# AI for Regulars — full content > Full text of every published guide, news item and tip, in plain markdown. > Canonical URLs live under https://aiforregulars.com. See https://aiforregulars.com/llms.txt for the index. --- # Best AI Apps for Kids by Age: 6, 10, and 14 Year Olds Source: https://aiforregulars.com/guides/ai-apps-kids-by-age Not every AI app is right for every child. A six-year-old needs something completely different from a fourteen-year-old, and the wrong tool at the wrong age can do more harm than good. Here is a practical, age-based guide to what actually works — and what to skip. ## Ages 6–8: AI That Asks, Not Answers Young children learn best by doing and talking, not by reading AI-generated answers. The best apps for this age group use AI to adapt games or ask guiding questions, not to hand over information. **What to look for:** Apps with no open chat box, limited vocabulary, and a parent dashboard you can check each week. Content should be reviewed and controlled by the app — not generated on the fly. **Apps worth trying:** - **Khan Academy Kids** — Uses character-based activities to adapt math and reading games. No chatbot, no risky content. Free. - **Duolingo ABC** — Adaptive reading and letter recognition for early learners. AI adjusts pace to the child. No chat features. - **Osmo** (with physical kits) — Combines screen and hands-on play using computer vision. Great for kindergarten through second grade. **Skip:** Any general-purpose chatbot (ChatGPT, Gemini, Copilot). They are not built for this age, and their content filters were designed with adults in mind, not six-year-olds. ## Ages 9–12: AI Tutors and Homework Helpers Tweens can start using AI tools that help them think through problems — but the keyword is "think through," not "give me the answer." The best tools for this age ask follow-up questions rather than just handing over solutions. **What to look for:** Tutoring-style AI that prompts reasoning, a parent or teacher dashboard, and age-appropriate content moderation. **Apps worth trying:** - **Khanmigo (by Khan Academy)** — Designed specifically for students. It asks Socratic questions rather than giving direct answers. Requires a Khan Academy account; family plans cost $4/month (free for teachers through school plans). - **Book Creator with AI** — Lets kids write and illustrate their own stories with AI suggestions. Good for creative writing without open-ended chat. **Safety note:** Even apps designed for this age benefit from a monthly check-in. Ask your child to show you what they used the app for last week — it keeps both of you in the loop. When my niece used Khanmigo for the first time, she kept trying to get it to just give her the answer. It kept asking "What do you think the first step would be?" — and eventually she solved it herself. That friction is the point. ## Ages 13–14: Real Chatbots With Real Guardrails Teenagers can start using general AI tools, but they need clear rules first. Without guidance, it is easy to slip from "help me understand this" into "write my essay for me." Set expectations before the app even opens. **What to look for:** Parental controls or family plan options, usage visibility, and an honest conversation about academic integrity. **Apps worth trying:** - **ChatGPT (with parental controls set)** — Teens 13+ can use it with an account and proper settings. Review [how to set up ChatGPT parental controls](/guides/chatgpt-parental-controls) before getting started. - **Microsoft Copilot** — Available free with any Microsoft account. Better integrated into school tools like Word and OneNote. - **Perplexity** — Search-style AI that shows its sources. Helps teens develop the habit of checking where information comes from, which is a skill worth building. **Academic integrity note:** At this age, it is worth having one direct conversation: using AI to write your work and submitting it as your own is dishonest, and most schools now have tools to detect it. Using AI to explain a concept you do not understand is just studying. ## What to Try Next If you want to go deeper on keeping kids safe across all apps — not just AI — [the guide to parental controls for AI apps](/guides/ai-parental-controls-all-apps) covers the settings worth turning on first. And if your child has already been using ChatGPT, [this parent's guide to ChatGPT and homework](/guides/kids-chatgpt-homework-parents-guide) walks through what to check. ## Sources - [Protecting Kids Online](https://consumer.ftc.gov/identity-theft-and-online-security/protecting-kids-online) — Federal Trade Commission - [Family Link — parental controls](https://families.google/familylink/) — Google - [Use Screen Time on your iPhone or iPad](https://support.apple.com/en-us/105121) — Apple --- # Your Bank's New AI Assistant: What It Can Do (and What It Can't) Source: https://aiforregulars.com/guides/ai-bank-assistant Your bank's app probably has a chat assistant in it now. You might have ignored it, or tapped it once and then closed it when you weren't sure what to ask. A lot of people do. But once you know what it's good at, it can save you real time. And once you know its limits, you'll know exactly when to talk to a person instead. This guide covers both sides. ## What Banks Are Rolling Out — and Why In 2026, Visa announced an AI Financial Assistant that lets cardholders ask plain-English questions about their own transactions, lock their card, and set spending alerts — all inside their existing bank app. (Source: Visa, 2026.) Major banks have been adding similar features of their own, and what all of these have in common is the same basic idea: instead of hunting through menus, you just ask. The technology is the same kind of AI that powers tools like ChatGPT — it reads your question, interprets what you mean, and pulls the relevant information from your account. The key difference from a general chatbot is that your bank's assistant is connected to your actual transaction data. It knows what you spent, where, and when. ## Five Things the Assistant Is Genuinely Good At **Finding a charge you don't recognise.** This is where it shines. Instead of scrolling through months of statements, you can describe what you're looking for: "Show me all charges from a company called 'Adobe' in the past six months" or "I see a charge for £14.99 on 12 July — what is it?" The assistant pulls up the exact transaction and shows you the full merchant name, which is often longer and more recognisable than the shortened version on your statement. **Explaining a fee.** Banks charge fees that feel mysterious — monthly maintenance fees, out-of-network ATM fees, foreign transaction fees. Ask: "Why was I charged $3.50 on 5 July?" It can usually tell you exactly what the fee was for and what you can do to avoid it in future. **Showing your spending by category.** Most bank assistants can produce a quick summary: "How much did I spend on restaurants last month?" or "What were my top five spending categories this year?" These aren't as detailed as a dedicated budgeting app, but for a quick picture, they're useful. **Setting alerts.** You can ask the assistant to notify you when your balance drops below a certain amount, when any charge over a certain size goes through, or when a specific recurring payment is taken. These alerts can catch problems early. **Locking your card.** If you can't find your debit card, you don't have to call anyone. Most bank assistants can lock it immediately. If you find the card later, you can unlock it the same way. Quick and done. ## How to Ask Questions That Actually Work The assistant understands plain English, but it responds better to specific questions than vague ones. Use dates or amounts when you have them. "Show me charges from May" is easier for it to handle than "show me old charges." Describe what you remember. "A charge I don't recognise" is less useful than "a charge for about £20 from a name starting with 'S'." Ask one thing at a time. Combining several requests in one message often confuses any chat system. Take it step by step. Here are examples of questions that work well: > "What did I spend at Tesco in June?" > "I see a charge for $9.99 labeled 'VZWRLSS' — what is that?" > "Lock my debit card. I'll unlock it when I find it." The assistant is better at looking things up than at giving you opinions. Keep that in mind and you'll find it much more useful. I use the bank assistant mainly to settle one question: whether a subscription I think I cancelled actually stopped. I'll ask "Show me all charges from Hulu in the last 12 months" and the answer is right there — no scrolling through statements, no phone call needed. It saves a surprising amount of time once you know to ask. ## The Hard Limits — Where the Assistant Falls Short This is the part that matters most for your financial safety. **It cannot give you reliable financial advice.** An MIT Sloan finance expert notes that AI chatbots work on probability patterns, not exact math — which makes them unreliable for precise financial calculations. (Source: MIT Sloan.) Asking the assistant "Should I pay off my credit card or put the money in savings?" or "Will I have enough to retire at 65?" will get you an answer. That answer may sound confident and thorough. Do not act on it without speaking to a licensed human adviser. **It has no fiduciary duty.** A licensed financial adviser is legally required to act in your best interest. The bank's AI assistant is not. It has no legal responsibility for bad advice. (Source: MIT Sloan.) This doesn't mean it's trying to mislead you — it isn't — but it does mean there is no recourse if it gets something important wrong. **Its math can be off.** For precise calculations — how much interest you'll pay over five years, what your net income is after a specific deduction — do not rely on the assistant alone. Verify the numbers yourself or ask a person. **It cannot resolve disputes or fraud claims.** The assistant can sometimes start a dispute process. The actual resolution requires a human. Ask it to connect you to a fraud specialist or dispute team, then follow through with a real person. ## What the Assistant Can (and Cannot) See Your bank's AI assistant has access to your account data — the same information you see when you log in: your balance, your transaction history, your account numbers. In most cases it can read that data, but it cannot move money without your explicit confirmation step. What it does not have access to is your broader financial life. It cannot see accounts at other banks, your investment portfolio, your credit score, or your tax history (unless those are explicitly linked to that specific bank). This matters because when the assistant gives you a picture of your finances, it is an incomplete picture. It only sees what is inside its own walls. ## What to Watch Out For **Impostor "bank AI" messages.** Scammers now send texts, emails, and messages that pretend to be your bank's AI assistant. They look official. They create urgency. They ask you to confirm your password, your card number, or a verification code. Your real bank AI will never ask for any of those things. If a message asks you to "verify your credentials" or "confirm your details to continue," close it immediately. Open your bank's official app directly — not through any link in the message — and report the contact. **Over-trusting the numbers.** The assistant can produce figures that look precise and authoritative — a projection of how your savings will grow, an estimate of fees you'll face. These figures can be wrong. Treat any calculation from the assistant as a rough guide, not a definitive answer. **Oversharing.** The assistant already knows your account data. Some people go further and share things it doesn't need: their full national insurance or Social Security number, their online banking password, or security question answers. The assistant has no legitimate reason to ask for any of these. If it seems to be asking, stop the conversation and call your bank directly. **Assuming the assistant speaks for the bank officially.** The assistant might explain a fee, a policy, or a rule. That explanation might be wrong or out of date. If something matters — whether you qualify for a fee waiver, what the early-closure penalty is — confirm it with a human representative. ## What to Try Next If you want to understand how your bank's AI protects your money behind the scenes — not the assistant you talk to, but the fraud detection running on every purchase — [How Your Bank Uses AI to Protect Your Money](/guides/how-your-bank-uses-ai-for-you) covers that in detail. And if you're interested in using AI to actively manage your budget rather than just look things up, [Best AI Budgeting Apps: What They Do and What to Expect](/guides/best-ai-budgeting-apps) is the natural next read. --- # Dangerous AI-Written Books Are Sold on Amazon — How to Spot One Before You Trust It Source: https://aiforregulars.com/guides/ai-books-amazon-dangerous Book by book, a new kind of shelf is filling up on Amazon. The titles look professional. The covers are polished. The author has a real-sounding name and a brief bio. But some of these books contain advice that could send you to a hospital — or worse — and the "author" may have published dozens of similar titles in the same month. ## Why AI-Generated Books Are a Safety Problem Most AI-generated books are harmless — low-quality but not dangerous. A mediocre AI-written biography or a repetitive self-help book wastes your money; it doesn't put you in danger. The problem is specific to categories where getting the information wrong can cause real, immediate harm. **Foraging guides are the clearest example.** In 2023, journalists at 404 Media found a wave of AI-generated mushroom foraging books on Amazon whose author bios didn't hold up to scrutiny — a claimed PhD from UC Davis that the university had no record of, decades of field experience with no trace anywhere outside the Amazon listing — and content that glossed over which species are dangerous while dwelling on unrelated details. The New York Mycological Society issued a public warning that books like these "could actually kill people if they eat the wrong mushroom because a guidebook written by an AI prompt said it was safe." In 2025, CBC reported a similar case in Saskatchewan: a self-published foraging guide with real inaccuracies about which local mushrooms are safe to eat, credited to an author with no verifiable background anywhere outside the book's own Amazon page — a sample of the text scored 100% on an AI-detection tool CBC used to check it. The danger isn't limited to foraging. The same pattern applies to books about **medication and drug interactions**, **medical diagnosis**, **infant care**, **wilderness survival**, and **travel safety**. These are categories where a single wrong fact can matter enormously. ## Red Flag 1: The Author Doesn't Exist Outside Amazon This is the fastest check, and it catches a lot. Search the author's name along with their stated expertise. If the book is about mushroom foraging in the Pacific Northwest, try "[Author Name] mycologist" or "[Author Name] foraging expert." A genuine expert in a safety-critical field leaves a trail outside their Amazon page: academic publications, professional society memberships, articles, media appearances, or at minimum a website with real content. **An author whose entire online presence is their Amazon page is a warning sign.** It doesn't prove the book is AI-generated — some first-time authors genuinely have no web presence — but combined with other signals, it matters. **Check the author bio on Amazon itself.** If it's vague ("nature enthusiast," "passionate about outdoor skills") rather than specific ("certified clinical nutritionist, 15 years at University of Oregon Health"), be skeptical. A real expert almost always has something concrete to establish credibility. ## Red Flag 2: Dozens of Books Published in Weeks or Months Click on the author's name to see their full publication list on Amazon. This is one of the clearest signals of AI-generated content. **A legitimate foraging expert or medical writer takes months or years to write a book.** If the same author has published 20 or 30 books in a three-month window — especially across wildly different topics — that volume is only achievable with AI. Amazon introduced a limit of 3 Kindle self-published books per day per author after this pattern became visible, but the problem persisted even after that change. **Watch for author names that vary slightly.** Content mills sometimes create clusters of similar author names — "James R. Miller," "J.R. Miller," "James Miller" — to stay under volume limits. If the writing style and topic focus are nearly identical across what look like different people, they may be the same AI operation. When I need a book on anything health-related or safety-critical, I skip Amazon's search entirely and start at a professional society's website instead. For foraging, that's the North American Mycological Association (namyco.org) or a regional mycological club. For medical topics, I look for books recommended by Mayo Clinic, the NHS, or a specialty medical society. These organizations vet what they recommend — Amazon doesn't. ## Red Flag 3: The Free Sample Sounds Generic Amazon lets you read a sample of almost every book before buying — use this for anything safety-critical. AI-generated text has a recognizable pattern even when it's grammatically correct: it tends to be fluent but hollow. You'll get technically accurate generalities without the specific, experience-grounded knowledge that makes a foraging book actually useful. A real foraging guide says something like "In the Pacific Northwest, Amanita phalloides starts appearing in October — it looks deceptively like a paddy straw mushroom, and that confusion has killed foragers." An AI-generated one says "Always check multiple characteristics before consuming any wild mushroom." **Look for personal anecdotes and specific expertise.** Books written by real experts almost always include moments where the author's specific experience shows — a mistake they made early in their career, a case where their advice differs from other guides and why, a vivid memory from fieldwork. AI can mimic this structure, but the anecdotes it generates tend to be vague and interchangeable. **Watch for hedging on exactly the details that matter.** Good foraging guides are specific about look-alikes and the signs that distinguish safe from toxic. Generic AI-generated text says things like "always consult an expert before eating any wild mushroom" — which is good advice but is also a way of avoiding specifics the author doesn't actually know. ## Red Flag 4: No Reviews Anywhere Except Amazon Amazon's own reviews can be faked, and for AI-generated books they sometimes are. Cross-referencing elsewhere takes a minute and catches a lot. **Search Goodreads.** Genuine readers who finish a book almost always leave a Goodreads rating or short review. A book with hundreds of Amazon ratings and near-zero presence on Goodreads is unusual — real books get talked about in both places. **Look for professional reviews.** Books in specialist fields — mycology, medicine, nutrition — are often reviewed by professional societies or specialty publications. A foraging guide with no mention anywhere outside Amazon, despite claiming expertise in a specialized field, is worth questioning. **Read the one-star and two-star Amazon reviews carefully.** Sometimes the most useful signal is in the negative reviews from actual readers who noticed something was wrong — not just the product quality, but the accuracy of the information itself. ## What to Watch Out For Even careful readers can be misled by AI-generated books that have been partially edited by a human — enough to pass a quick sample read but not enough to be reliable for safety-critical information. The honest answer is that **there is no foolproof test.** AI detection tools can flag texts as AI-generated when they aren't, and miss AI text that has been lightly edited. The checks in this guide — author credentials, publication volume, sample quality, cross-referencing reviews — reduce your risk but don't eliminate it entirely. For anything where being wrong has real consequences — mushroom foraging, medication, infant care, wilderness survival — the safest approach is to **prefer books from established publishers with identifiable, credentialed authors**. Major publishers have editorial processes that include fact-checking. A book published by a university press, a major health publisher, or a nonprofit field guide organization has cleared more gates than a self-published Kindle title from an author who also published 19 other books that month. This doesn't mean every self-published book is bad or every traditional publisher is reliable. It means the starting odds are different, and for safety-critical topics, those starting odds matter. ## What to Try Next The same instincts that catch AI-generated books also help with AI-generated reviews — [How to Spot Fake AI-Generated Product Reviews Before You Buy](/guides/spot-fake-ai-product-reviews) walks through the specific signals to look for on any product listing. And if you've ever wondered whether a piece of writing was produced by AI, [How to Tell If Text Was Written by AI](/guides/how-to-tell-if-text-is-ai) covers what the detection tools actually catch and where they fall short. ## Sources - ['Life or Death:' AI-Generated Mushroom Foraging Books Are All Over Amazon](https://www.404media.co/ai-generated-mushroom-foraging-books-amazon/) — 404 Media - [AI generated mushroom books – life and death](https://www.newyorkmyc.org/ai-generated-mushroom-books-life-and-death/) — New York Mycological Society - [Sask. mushroom pickers sound alarm on dangerously misleading AI-generated books](https://www.cbc.ca/news/canada/saskatchewan/mushroom-picking-ai-generated-1.7502667) — CBC News - [Amazon's Kindle Direct Publishing Will Limit Daily Number of New Titles](https://www.publishersweekly.com/pw/by-topic/digital/content-and-e-books/article/93207-kdp-will-limit-daily-number-of-new-titles.html) — Publishers Weekly - [Content Guidelines: Artificial intelligence (AI) content](https://kdp.amazon.com/en_US/help/topic/G200672390) — Amazon Kindle Direct Publishing --- # AI Browsers Explained: What ChatGPT Atlas, Comet, and Gemini in Chrome Actually Do Source: https://aiforregulars.com/guides/ai-browsers-explained You've probably heard one of these names in the past year: ChatGPT Atlas, Perplexity Comet, Gemini in Chrome. They all get lumped together as "AI browsers," but they're not the same product, they're not all available where you live, and one of them doesn't exist anymore. This guide sorts out what's actually real, what each one does, and the one safety habit worth building before you try any of them. An AI browser is a regular web browser with an assistant built directly into it — not a separate app you switch to, but a sidebar or chat panel that can see the page you're looking at, answer questions about it, and in some cases act on it: filling forms, comparing options across open tabs, or clicking through a multi-step task on your behalf. Right now there are two real ones worth knowing about: - **Perplexity Comet** — a free, standalone browser you install - **Gemini in Chrome** — an AI feature built into Google's existing Chrome browser, rolling out gradually by country There was a third: **ChatGPT Atlas**, which OpenAI launched in October 2025 with its own agent mode for multi-step tasks. OpenAI discontinued it less than a year later, folding its features into the main ChatGPT desktop app instead, and Atlas stopped working on August 9, 2026. If you're reading an older article that recommends downloading Atlas, that advice is out of date — don't go looking for it. That's worth pausing on for a reason beyond trivia: a major tech company launched a browser with full access to your open tabs and browsing habits, and killed it within ten months. That's a useful reminder not to over-invest trust, passwords, or sensitive accounts in any single AI browser, no matter how polished it looks on day one. Comet launched in July 2025 as a paid perk for Perplexity's most expensive subscribers, then went free worldwide on October 2, 2025. It's built on the same underlying engine as Chrome (Chromium), so it looks and feels familiar, with an AI sidebar added on top. In practice, Comet can: - Summarize a long article or webpage without you having to read the whole thing - Compare products across multiple open tabs — prices, reviews, specs - Manage and organize your open tabs - Navigate through an online store and assemble items in a cart That last one is where people usually get nervous, so it's worth being precise: Comet is designed to stop short of completing payment on its own. It can do the browsing and cart-building legwork, but it hands the final checkout step back to you for review. If you want to try it, you can download Comet from Perplexity's website and use it for browsing tasks the way you'd use a research assistant — just keep reading step 4 before you connect it to anything sensitive. Gemini in Chrome is Google's answer built into its existing browser rather than a new one to download. It launched to all US desktop users on September 18, 2025 (it had previously been limited to paying Google AI subscribers). It can explain a confusing part of a webpage in plain language, work across your open tabs to help you compare or research something, and Google has been gradually adding small "agentic" abilities — like helping book an appointment or add items to a cart — with the final step, such as payment, left for you to confirm. Here's the part that matters if you're reading this from Europe: as of this writing, Gemini in Chrome has not launched in the EU. It expanded from the US to Canada, India, and a handful of other countries during 2026, but EU availability is still pending, and it's widely assumed that's connected to compliance work required under EU digital regulation. Google hasn't given a firm EU launch date. If your Chrome doesn't show this feature and you're in an EU country, that's expected — it's not something broken in your settings. The single biggest thing to understand about any AI browser is a risk called **prompt injection**. It sounds technical, but the idea is simple: a webpage can contain instructions that are invisible or meaningless to you, but that the AI reads as if you had typed them yourself. It's the same basic trick as a phishing email — a message designed to fool someone into acting against their own interest — except here the target is the AI, not you directly. This isn't a hypothetical. Between August and October 2025, security researchers at Brave and other firms disclosed several real vulnerabilities in Comet, under names like "CometJacking." In one disclosed scenario, a webpage could hide instructions — for example inside an ordinary-looking spoiler tag on a Reddit-style post — that quietly told Comet's AI to hand over sensitive information, like a one-time login code, to an attacker, all without the person browsing noticing anything unusual on screen. Perplexity has patched the specific issues that were disclosed, but the underlying weakness — a webpage being able to influence what the AI does — is a structural risk in how these agentic browsers work, not a single bug that gets fixed once and never resurfaces. A few concrete precautions that actually help: - Don't leave an AI browser's agent or "do this for me" mode connected to email, banking, or payment accounts unless you trust the exact page it's about to act on - If the AI does something you didn't ask for — opens a page, fills a field, sends information — treat that as a red flag worth stopping and investigating, not a quirk to shrug off - Keep the browser updated; vendors patch these vulnerabilities as they're found, and an outdated copy stays exposed to issues that are already fixed for everyone else - Be extra cautious about clicking into unfamiliar or unexpected pages while an agent task is actively running in the background I installed Comet mainly to test the tab-comparison feature while researching this guide, and it's genuinely useful for pulling together specs from three or four open product pages. But I keep it logged out of anything financial. Given what the security researchers found, treating it like a smart research assistant rather than a trusted personal assistant feels like the right level of caution for now. If you're comfortable with the idea and not in a hurry: Comet is free, real, and fine to try for research-style tasks — summarizing, comparing, browsing — as long as you follow the precautions in step 4 and keep it away from sensitive accounts. Gemini in Chrome is worth turning on once it reaches your country, since it's a feature inside a browser you likely already use rather than a new one to trust from scratch. If you'd rather wait, that's a perfectly reasonable choice too. Nothing in a regular AI chat tool (like typing questions into ChatGPT, Gemini, or Perplexity's regular search) requires an AI browser, and this is a young, fast-moving category — the Atlas story shows that even a well-funded company can launch and cancel one within a year. There's no urgency to adopt an AI browser before it's actually useful to you. ## What to Try Next If a lot of this "AI acting on your behalf" territory is new to you, [What Is an AI Agent?](/guides/what-is-ai-agent) explains the underlying idea in plain terms before you decide whether to trust one with your browsing. And since the prompt-injection risk in this guide shares a lot in common with other AI-targeted scams, [How to Spot AI Phishing Emails](/guides/spot-ai-phishing-emails) covers the closely related pattern of AI being used to trick people directly. ## Sources - [OpenAI is shutting down Atlas, but its AI browser ambitions are still growing](https://techcrunch.com/2026/07/09/openai-is-shutting-down-atlas-but-its-ai-browser-ambitions-are-still-growing/) — TechCrunch - [Perplexity Launches Comet: a Browser Designed around AI-Assisted Interaction](https://www.infoq.com/news/2025/07/perplexity-comet/) — InfoQ - [Perplexity's Comet AI browser now free, Max users get new background assistant](https://techcrunch.com/2025/10/02/perplexitys-comet-ai-browser-now-free-max-users-get-new-background-assistant/) — TechCrunch - [Agentic Browser Security: Indirect Prompt Injection in Perplexity Comet](https://brave.com/blog/comet-prompt-injection/) — Brave - [CometJacking: One Click Can Turn Perplexity's Comet AI Browser Into a Data Thief](https://thehackernews.com/2025/10/cometjacking-one-click-can-turn.html) — The Hacker News - [Go behind the browser with Chrome's new AI features](https://blog.google/products/chrome/new-ai-features-for-chrome/) — Google - [Google brings Gemini in Chrome to US users, unveils agentic browsing capabilities, and more](https://techcrunch.com/2025/09/18/google-brings-gemini-in-chrome-to-us-users-unveils-agentic-browsing-capabilities-and-more/) — TechCrunch --- # Hidden AI Features Already in Your Phone's Camera Source: https://aiforregulars.com/guides/ai-camera-features-phone The camera on your phone is doing a lot more than you probably realize. Before the photo even saves, AI has already cleaned up the noise, adjusted the exposure, and applied portrait blurring if you were shooting a person. And after the photo saves, there are more tools waiting in your Photos app that most people never discover. This guide walks through the most useful ones — on both iPhone and Android. ## Step 1: Find Your AI Photo Editing Tools On iPhone, open a photo in the Photos app, then tap **Edit** in the top right. You will see a row of icons — look for the small sparkle or wand icon, which indicates AI-powered tools. On Pixel, open Google Photos, select a photo, tap **Edit**, and look for the **Tools** section at the bottom. On Samsung Gallery, open a photo and tap the pencil icon, then tap the Photo Assist icon (the blue circle with stars). All three platforms have a tool for removing people, objects, or distractions from photos. - **iPhone (Clean Up):** Tap the Clean Up tool (a circle with an X). The phone automatically highlights things it thinks you might want to remove — tap one to erase it. Or circle any object yourself with your finger. The AI fills in the background. - **Pixel (Magic Eraser):** In Google Photos, tap Edit → Tools → Magic Eraser. The phone suggests things to remove (strangers, power lines). Tap a suggestion or circle any area yourself. - **Samsung (Object Eraser):** Tap Edit → AI tools → Object eraser. Circle or tap any object. Samsung also offers a Generative Edit tool on newer Galaxy models that can fill in and extend backgrounds. All three preserve the original — you can revert any time before you save. Slightly blurry photos — from a quick movement or a missed focus — can often be recovered. - **Pixel (Photo Unblur):** In Google Photos, tap Edit → Tools → Photo Unblur. The phone detects blur and applies a correction. A slider lets you control how strong the effect is. - **iPhone:** The Photos app has a Sharpen adjustment under Edit that can help reduce the appearance of slight blur. Unlike Pixel's Photo Unblur, it is a manual slider, not an AI-powered tool. - **Samsung:** Galaxy AI includes Photo Remaster and Sharpen features under Edit → AI tools. Portrait mode is a camera AI tool that blurs the background to make your subject stand out. But the AI goes further — it also adjusts skin tones, smooths lighting, and on newer phones, lets you change the light source after the fact. - **iPhone:** Open a Portrait photo, tap **Edit**, and look for the **Lighting Effects** slider (the cube icon). You can switch between Natural Light, Studio Light, Contour Light, and more — these are not filters, they are AI-calculated adjustments to how the light appears to fall. - **Pixel:** Google's Portrait Mode includes Portrait Blur and Portrait Light. In Google Photos, open a portrait → Edit → Tools → Portrait Light. Drag to adjust where the light appears to come from. - **Samsung:** Remaster Portrait under Galaxy AI tools adjusts face and background separately. When you zoom in on distant subjects, the image usually gets blurry. Google's Zoom Enhance (available on Pixel 8 Pro and later, including all Pixel 10 models) applies AI to sharpen a zoomed photo after you take it. You take the shot at whatever zoom you used, then go to Edit → Tools → Zoom Enhance, and the AI sharpens the result. Samsung's Space Zoom on Galaxy S series phones does a similar thing in real time — the viewfinder stays sharp at high zoom because the phone is processing the image continuously before you even press the shutter. I used Magic Eraser to remove a stranger walking through the background of a family photo at a park. I expected it to leave a strange blurry patch — instead, the tree and grass filled in so naturally that I had to look at the original twice to see where the person had been. The tool is not perfect on complex backgrounds, but for simple cases like sky, grass, and pavement it is impressively clean. Taking group photos is frustrating because someone always blinks or looks away. Google Pixel 8 and later have a feature called **Best Take**: you take a burst of a few photos in quick succession, then go to Edit → Tools → Best Take. The phone analyzes every face in each frame and lets you swap in the best version of each person's face from any of the shots. The result is a composite where everyone looks their best. You do not need to turn on Night Mode on most phones — the camera detects low light and activates it automatically. It takes several fast exposures and blends them. The AI removes the parts that blurred from movement and stacks the sharp parts, giving you a brighter, cleaner photo without flash. **Action Pan** (available on some Pixel models) does the opposite — it intentionally keeps the subject sharp but blurs the background to suggest movement. Good for photos of kids running, bikes, or anything in motion. You will find it as a shooting mode in the camera app. ## What to try next Camera AI is just one slice of what your phone can do with AI built in. For a full rundown of non-camera features — live translation, call screening, writing tools — see [Hidden AI Features on Your Smartphone You Did Not Know About](/guides/hidden-ai-features-smartphone). And if you have old printed photos you want to rescue and restore, [How to Restore Old Photos with AI](/guides/restore-old-photos-ai) covers the tools that work best. --- # AI Captions and Live Transcription for Deaf and Hard-of-Hearing People Source: https://aiforregulars.com/guides/ai-captions-hard-of-hearing AI has made real-time speech-to-text genuinely useful in daily life — not perfect, but close enough to change how people who are deaf or hard of hearing navigate conversations, phone calls, meetings, and media. The best news is that many of the best tools are already on devices you own. Here is a practical breakdown of what's available, what each tool does best, and how to turn it on. ## Built-in OS Captions (No Download Required) Before reaching for a separate app, check what your device already offers. These built-in features have improved significantly in recent years. ### Android: Live Transcribe Google's Live Transcribe turns your Android phone into a real-time captioning device. Point the microphone toward whoever is speaking and read captions on screen. It shows pauses, identifies different speakers, and can even detect environmental sounds like laughter or applause. **How to enable:** Settings → Accessibility → Live Transcribe. On Samsung Galaxy phones, look under Hearing Enhancements. On Pixel phones it may also appear in the quick settings panel. **Cost:** Free. **Best for:** Face-to-face conversations, group settings, anywhere you can hold your phone. ### Windows: Live Captions Windows 11 includes Live Captions, which transcribes any audio playing through or into your computer — including video calls, streaming video, microphone input, and system audio. It runs as a floating bar you can dock to the top or bottom of your screen. **How to enable:** Settings → Accessibility → Captions → Live Captions, or press Windows key + Ctrl + L. **Cost:** Free with Windows 11. **Best for:** Video calls, streaming video, online meetings, any computer audio. ### Apple: Live Captions (iPhone/iPad and Mac) Apple's Live Captions transcribes speech in real time and integrates with FaceTime. On supported devices, it can caption FaceTime calls automatically. **How to enable:** Settings → Accessibility → Live Captions (toggle on). Requires iOS 16 or later and iPhone 11 or newer; English (US or Canada) must be your primary language. **Cost:** Free on supported devices. **Best for:** FaceTime calls, iPhone conversations, Mac audio. ## Dedicated Apps for More Capability | App | Platform | Best for | Cost | Notes | |---|---|---|---|---| | Live Transcribe | Android | In-person conversations | Free | Offline mode available | | InnoCaption | iOS / Android | Phone calls | Free (FCC-certified) | Real captioners assist AI | | Otter.ai | iOS / Android / Web | Meetings, recordings | Free tier; paid plans | Saves searchable transcripts | | Ava | iOS / Android | Group conversations | Free tier; paid plans | Multi-speaker tracking | | Google Meet captions | Web / App | Video meetings | Free with Google account | Auto-enabled in Meet settings | | Zoom auto-captions | Web / App | Video meetings | Free tier | Enabled in accessibility settings | ### InnoCaption (Recommended for Phone Calls) InnoCaption is designed specifically for captioning phone calls. It uses a combination of AI and human captioners to transcribe the other person's speech in real time while you talk normally. It is certified by the FCC under the Telecommunications Relay Service program, meaning it is provided at no cost to people with hearing loss. **How it works:** You make calls through the InnoCaption app. Captions appear on screen as the other person speaks. **Setup:** Download the app and create a free account. You self-certify hearing loss — no doctor's note is required. The service is US-only; FCC registration requires basic personal information including the last four digits of your Social Security number. ### Otter.ai (Recommended for Meetings and Recordings) Otter.ai transcribes meetings, interviews, lectures, and recordings. It identifies different speakers, saves searchable transcripts, and integrates with Zoom, Google Meet, and Microsoft Teams. It is not designed for real-time face-to-face conversation but excels for recorded or structured sessions. I tested Live Transcribe in a coffee shop and was surprised how well it handled background noise at a reasonable distance. The transcription was about 85–90% accurate in that setting — good enough to follow a conversation, not perfect enough to rely on for something critical. In a quiet room it was noticeably better. My recommendation: use it, but confirm important details in writing when the stakes are high. ## Captions for TV and Streaming Most streaming services now include captions as standard. A few things worth knowing: - **Netflix and Amazon Prime** both offer closed captions. Look for the CC or speech-bubble icon during playback. - **YouTube** auto-generates captions for most videos. They're imperfect but often usable. You can also search for videos that include human-made captions. - **Smart TVs** have accessibility menus for enabling broadcast captions — usually under Settings → Accessibility → Caption Settings (exact path varies by brand and model). Streaming apps like Netflix and Amazon Prime use their own in-app caption settings. For anything playing on a Windows computer, the built-in Live Captions feature can overlay captions on any video, regardless of whether the video's source offers them. ## Tips for Getting the Best Results A few practices make AI captions work significantly better: **Reduce background noise.** AI captioning accuracy drops with competing sounds. When possible, find a quieter spot or ask the other person to speak more slowly. **Position the microphone correctly.** For phone apps, the microphone faces the speaker — not toward you. For computer apps, a directional microphone rather than a built-in laptop mic makes a real difference. **Use wired headphones for calls.** Bluetooth audio sometimes introduces a slight delay that can throw off caption timing. **Enable speaker identification.** Apps like Otter.ai and Ava label each speaker separately, which makes multi-person conversations much easier to follow. ## What to try next If you are interested in broader accessibility tools, [AI Tools for Low Vision: What's Worth Using](/guides/ai-tools-low-vision) covers screen readers, magnification, and image description features. For a general introduction to AI assistants on your phone, [ChatGPT Voice Mode](/guides/chatgpt-voice-mode) explains how voice-based AI works and what to expect from it. --- # How to Use AI When Buying a Car (and Negotiate Like a Pro) Source: https://aiforregulars.com/guides/ai-car-buying The salesperson at the car dealership knows far more about pricing than almost any buyer who walks in. That information gap is not an accident — it's how dealerships make money. AI doesn't erase that gap overnight, but it gives you a fast, free way to close it before you ever step onto the lot. A 2025 study by CarEdge found that 1 in 4 car buyers now uses AI tools somewhere in the purchase process, and 88% of them said it was helpful. ## Figure Out What You Actually Need First Before you look at a single price, use AI to figure out which car makes sense for your actual life — not just your budget. **Be specific about your situation, not just your preferences.** The more detail you give, the more useful the answer. Try something like: > "I need a family car for two adults and three kids aged 5 to 12. We do one long road trip per year and daily school runs. Towing a small boat trailer would be a bonus. My budget is around $35,000. Compare the Toyota Highlander and the Honda Pilot for my situation, including reliability, cargo space, and total ownership costs." ChatGPT or Gemini will give you a side-by-side breakdown of real-world trade-offs — cargo space, reliability ratings, fuel economy, known issues — rather than just listing specs from the manufacturer's website. This is much faster than reading through five different review sites and trying to piece the picture together yourself. **Ask about ownership costs, not just the sticker price.** The purchase price is only part of what a car costs you. > "What are the typical 5-year ownership costs for a 2023 Toyota Camry vs. a 2023 Honda Accord, including insurance, maintenance, and fuel? Which tends to hold its value better?" AI has absorbed a lot of public data on these questions. Treat the output as a rough framework, not a precise forecast — verify figures on dedicated sites like Edmunds or Consumer Reports before making a decision. ## Find Out What a Fair Price Actually Is This is where most of the anxiety lives. Many buyers walk into a dealership with no idea whether the number on the sticker is $2,000 above market or $8,000 above market. AI can close that knowledge gap — not by giving you today's live price, but by teaching you the pricing framework. **Ask the AI to explain the key numbers before you go:** > "Explain the difference between MSRP, invoice price, and dealer holdback for a 2024 Honda CR-V. What's a realistic target price to negotiate to in the current market?" The AI can explain how dealer incentives work, what markup is typical, and which factors — a popular color, the end of the month, an outgoing model year — give you more or less room to negotiate. **Then verify on a live pricing site.** AI's training data has a cutoff and does not reflect this week's market. Always check KBB (Kelley Blue Book) or Edmunds for the current fair market range before you talk to a dealer. Use the AI to understand the framework; use the live sites to get the actual number. ## Prepare Your Negotiation Talking Points Walking into a dealership with a specific, written list of talking points changes the dynamic completely. AI is very good at helping you build one. > "I want to buy a 2024 Toyota RAV4 LE AWD. The Edmunds Fair Market Price is $31,200. The dealership has offered $33,400. Give me three specific, polite talking points I can use to negotiate the price down, including a mention of a competing offer I have from another dealer." The AI will produce language that sounds confident but not confrontational — the tone that actually works in these conversations. Keep the list on your phone or write the key points on a notepad. **One important rule: do not paste AI text verbatim into dealer emails.** Salespeople who work over email all day recognize AI-generated messages almost immediately, and it tends to make them less willing to deal. Use the talking points as a script to adapt in your own voice — not text to copy and paste directly. Before my last car purchase, I asked ChatGPT to walk me through every line item on a sample purchase agreement — documentation fees, dealer prep charges, registration fees — so I knew which ones were negotiable and which were fixed state costs. It took about ten minutes, and I walked into the dealership knowing exactly which numbers I could push back on without looking like I was guessing. ## Decode the Dealer's Email Offer When a dealership sends you a written quote, it's often designed to look thorough while burying the number that actually matters — how much you'll pay total. **Copy the email into the AI and ask it to analyze it:** > "Here is a quote from a car dealership. Tell me: what is the out-the-door price (the total I would actually pay), which fees are included, which of those fees are typically negotiable, and what is missing from this quote that I should ask about before agreeing to anything?" The AI will flag things like documentation fees (which vary widely and are often padded), dealer-added accessories you never asked for, and whether the quote includes taxes and registration. It can also spot if a monthly payment looks low because the loan term has been stretched to 72 or 84 months, rather than because the price is actually better. ## Handle the Finance Office With Confidence The finance office is where dealers often recover the margin they gave up on the sticker price. The finance manager's job is to sell you add-ons — extended warranties, paint protection packages, GAP insurance, tire-and-wheel protection — and most buyers agree to things they don't need because the conversation happens fast, the prices are buried in a monthly payment, and saying no feels uncomfortable. **Use AI to understand each add-on before you get there:** > "What is GAP insurance on a car loan, who actually needs it, and is it better to buy it from the dealer or somewhere else?" > "What is the difference between a dealer extended warranty and a manufacturer's extended warranty? Which tends to be better, and what should I watch out for in the fine print?" **Roleplay the finance office conversation.** This is one of the most useful things AI can do for you as a car buyer. Ask it to play the role of a finance manager: > "Act as a car dealership finance manager. Try to sell me an extended warranty, paint protection, and GAP insurance. I'll practice saying no to each one. Start the roleplay." After a few rounds, you'll have comfortable, non-confrontational answers ready. The finance office feels a lot less stressful when you've already had this conversation a few times in a low-stakes setting. ## Let Someone Else Negotiate Entirely If you'd rather skip the negotiation process completely, a handful of services will handle it on your behalf. Negoshify — founded by a former Apple engineer — is one of the newer options: you specify the car you want, they put down a refundable $500 reservation fee, and their team negotiates with dealers for you. Services like this cost money, but for buyers who find the whole process stressful, or who don't have time to go back and forth with three different dealerships, the time saved may be worth the fee. Think of it as paying for the outcome rather than the work. ## What to Watch Out For **AI price data can be out of date.** AI tools have training cutoffs, and car prices shift with supply and demand, interest rates, and model-year changeovers. Use AI to understand pricing dynamics; use KBB, Edmunds, or TrueCar for current numbers. **Don't share sensitive personal information with AI.** You do not need to tell ChatGPT your real name, address, Social Security number, or full financial picture to get useful car-buying help. Keep conversations general — "a buyer in my situation" or "a household with this income" — rather than including your actual details. **AI can't read the room.** Negotiation has a human element that no AI can replicate — sensing when the salesperson is genuinely at their limit, or when to stop pushing and close the deal. Use AI to prepare; use your own judgment once you're sitting across from someone. **Verify specific policies before you rely on them.** Dealer fee limits, which fees are negotiable in your state, and manufacturer incentive programs all change frequently. If AI tells you something specific about policy or law, verify it with the dealer or a state consumer protection website before acting on it. ## What to Try Next Once you have a purchase agreement in hand, AI can help you understand every line before you sign — [How to Use AI to Understand Any Contract or Legal Document](/guides/ai-explain-contracts) walks through exactly that process. And if you want ready-made scripts for other financial conversations, [AI Message Templates: Scripts for Awkward Conversations](/guides/ai-message-templates) has examples you can adapt to fit your own situation. --- # AI Companions for Seniors: Can They Really Help with Loneliness? Source: https://aiforregulars.com/guides/ai-companions-seniors Loneliness among older adults is a genuine health concern, and families often can't be there around the clock. AI companions have emerged as a possible answer — always on, never impatient, ready to chat at 3 a.m. But the honest question is: do they really help? The short answer is: they can help at the edges — filling quiet hours, prompting daily routines, giving someone to "talk through" a worry with. They are not a cure for loneliness, and they carry real risks if families use them as a hands-off substitute for human care. ## What AI Companions Actually Do Most AI companion products sit in one of two categories. **General-purpose chatbots** (like ChatGPT or Claude) can hold wide-ranging conversations about almost anything — family history, recipes, news, books. They respond in seconds, remember what you said earlier in the conversation, and never get frustrated. They are not designed specifically for seniors, but many older adults find them easy and surprisingly engaging. **Dedicated senior companion products** are built from the ground up for this audience. Some come as tablet apps; others, like ElliQ, are physical devices with a small screen and a "personality." They may include daily check-ins, medication reminders, trivia games, and the ability to alert a family member if the person seems distressed or skips a day. Both types can have real value. The difference is how much structure and caregiver integration you need. ## What the Research Says (and Doesn't) Studies on AI companions for older adults are still early and small. What researchers have reported is encouraging but cautious: some participants report feeling less isolated after regular use, and daily check-ins help people maintain a sense of routine. What is less clear is whether the effect lasts over months, and whether it holds for people with significant cognitive decline. The honest framing is that AI companions appear to reduce *the feeling* of loneliness in moments — the 2 a.m. worry, the quiet Sunday afternoon. They do not appear to replace the health benefits that come from real human relationships: shared physical presence, eye contact, touch, being truly known by another person. I spent a few evenings testing a general chatbot as a "companion," and what struck me was how easy it was to get absorbed in conversation. That's exactly the benefit — and the risk. If you set up an AI companion for a parent, plan regular check-ins yourself. The AI is the warmup act, not the main event. ## The Real Risk: Substitution The biggest concern among gerontologists and mental-health professionals is not that AI companions are harmful in themselves — it's that families and care facilities might use them to *reduce* human contact rather than enhance it. "Mom has her robot" can quietly become an excuse for fewer visits. Signs that an AI companion is being misused: - Family calls become less frequent because "she seems fine" - The person becomes distressed or anxious when the device is off or unavailable - Real social engagements (senior center, church, neighbors) drop off - The person reports the AI as their closest relationship If you notice any of these patterns, treat it as a signal to increase human contact, not adjust the AI settings. ## What AI Companions Are Not **They are not therapists.** If someone is experiencing depression, grief, or anxiety, a qualified counselor or the person's doctor is the right first call. An AI can listen, but it cannot diagnose, treat, or prescribe. Some products now include safety features that detect distress language and prompt emergency contacts — useful, but not a substitute for professional care. **They cannot replace physical presence.** Falls, medication confusion, changes in eating or hygiene — these are things a family member or professional caregiver notices in person. An AI conversation partner cannot see any of this. **They are not secure confidants.** Conversations with most AI services are logged on company servers. Seniors should know not to share financial information, Social Security numbers, passwords, or highly sensitive family matters during AI chats. ## How to Introduce One Well If you decide an AI companion could be genuinely useful for a parent or grandparent, a few practices make the difference between a helpful tool and a forgotten gadget: Start together. Sit with the person for the first few sessions. Show them how to start a conversation, help them find topics they enjoy, and make sure they feel in control of the device — not managed by it. Keep expectations realistic. Frame it as "a friendly thing to chat with when you feel like it," not "this will keep you company." The lower-pressure framing reduces disappointment. Keep human contact as the anchor. Schedule regular calls and visits that are not contingent on whether the AI is being used. The companion fills gaps; it does not replace the schedule. Watch for dependency. Check in monthly about how the person feels about the device. If the relationship starts to feel more important than human relationships, it's time to recalibrate. ## When to Involve a Doctor or Counselor If loneliness is severe — affecting sleep, appetite, motivation, or the person's sense of purpose — please talk with a doctor before reaching for a technology solution. Chronic loneliness at that level is a health issue that deserves clinical attention. A primary care doctor can refer to social workers, mental-health professionals, or community programs that address root causes. An AI companion is a complement to good care, not a shortcut around it. ## What to try next If you're new to AI tools altogether, [ChatGPT for Seniors: Your First Hour](/guides/chatgpt-for-seniors-first-hour) walks through getting started with no prior experience. And if you're thinking about other ways technology can support an aging parent at home, [AI Monitoring for Aging Parents: The Privacy Dilemma Families Face](/guides/ai-home-monitoring-elderly) covers the important consent questions before you install anything. --- # Using AI to Compare Medicare Plans During Open Enrollment Source: https://aiforregulars.com/guides/ai-compare-medicare-plans-open-enrollment Every autumn, tens of millions of people with Medicare get a stack of mail: plan comparison booklets, an Annual Notice of Change letter, maybe a call from an insurance agent. Medicare's Open Enrollment Period runs October 15 through December 7 every year, and it's the one window where you can freely switch between Original Medicare and a Medicare Advantage plan, change plans, or join or drop a Part D drug plan — with the change taking effect January 1. It's also, understandably, the moment a lot of people wonder if ChatGPT could just do the comparing for them. It partly can — but not the way you might expect. Here's how to use AI as a genuinely useful assistant in this process, without letting it (or a scammer pretending to be it) make the actual decision for you. ## What AI Is Good For Here — and What It Isn't **AI is useful for:** - Turning a dense Annual Notice of Change letter into a plain-English summary of what's actually different next year - Explaining unfamiliar terms — "formulary," "prior authorization," "Medigap," "star rating" — in context - Building a checklist of questions to ask an insurer, a doctor's office, or a SHIP counselor - Organizing a side-by-side comparison once you've already pulled the real numbers for two or three plans **AI is not reliable for:** - Calculating your actual yearly cost for a plan — that depends on live premium, deductible, and drug-pricing data that only the official Medicare Plan Finder has - Confirming whether your specific doctors or pharmacy are in a plan's network this year - Telling you which plan is "best" without seeing your real prescriptions, providers, and location - Anything involving your Medicare number — that never belongs in a chat with a tool or a website that contacted you first With that division in mind, here's a safe way to actually do this. Go to **medicare.gov/plan-compare**. You can browse plans without logging in, or sign in with your Medicare account to have your current drugs and pharmacies pre-filled. Enter every prescription you take and the pharmacy you use — this is the step that makes the cost estimates real, because the tool prices out your actual medications against each plan's formulary. Narrow the list down to two or three plans that look reasonable. This is your source of truth for the rest of the process. If you're already on a Medicare Advantage or Part D plan, your insurer was required to send you an Annual Notice of Change (ANOC) letter by the end of September, listing what's changing for next year: premium, deductible, drug tiers, and network. Copy the text of that letter (leaving out your Medicare number and other personal identifiers) into ChatGPT or Gemini and ask: > "This is my Medicare plan's Annual Notice of Change letter. Can you summarize in plain English what's changing next year, and flag anything that looks like it could cost me more?" This is where AI earns its place in the process — a three-page notice written in insurance language becomes a short, readable list. Once you have real numbers from Step 1 for two or three plans — premium, deductible, and estimated drug costs — you can ask AI to lay them out side by side: > "Here are three Medicare plans with their premium, deductible, and estimated yearly drug cost from the Medicare Plan Finder: [paste the numbers]. Can you put these in a simple comparison table and note the biggest differences?" Feeding it numbers you already trust, rather than asking it to produce numbers itself, is the difference between using AI as an organizer and using it as an unreliable source. Enrollment season is also scam season. Consumer-protection groups have flagged a rise in AI-generated phone calls, fake "Medicare advisor" chatbots, and lookalike websites that appear during Open Enrollment specifically to harvest Medicare numbers and bank details. A few rules that hold regardless of how convincing the caller or chatbot sounds: - **Medicare and Social Security never call, text, or email you first** to "confirm" your Medicare number or ask you to pay to keep your coverage active. - **A real agent can wait.** If someone is pressuring you to decide or hand over information immediately, that pressure is the warning sign, not the offer itself. - **If you didn't start the contact, don't continue it.** Hang up, close the page, and call 1-800-MEDICARE yourself to check whether anything is actually wrong with your account. Every state has a State Health Insurance Assistance Program (SHIP) — free, unbiased Medicare counseling with no ties to any insurance company. A SHIP counselor can review your plan options with you over the phone or in person, at no cost. Find your local SHIP at **shiphelp.org**, or call **1-800-MEDICARE (1-800-633-4227)**, available 24 hours a day. If anything about your comparison still feels uncertain after using AI, this is the step that resolves it. The single most useful thing I've seen AI do in this process isn't picking a plan — it's decoding the Annual Notice of Change letter. Those letters bury the one sentence that matters ("your monthly premium is increasing from $34 to $58") in a paragraph about federal regulations. Paste the letter in, ask "what's actually changing and does it cost me more," and you'll have the real answer in under a minute — then you can decide if it's worth digging further on the Plan Finder. ## The One-Sentence Version Let the official Medicare Plan Finder generate the real numbers, let AI turn those numbers and your Annual Notice of Change letter into something you can actually read and compare, and let a free SHIP counselor be your second opinion — and never let anyone who contacted *you* first be the one asking for your Medicare number. If you also received a confusing government letter about Social Security or another benefit this season, our guide on [using AI to understand government forms and benefits](/guides/ai-government-forms-help) covers the same paste-and-ask approach in more detail. ## Sources - [Open Enrollment](https://www.medicare.gov/health-drug-plans/open-enrollment) — Medicare.gov (official) - [Medicare Plan Finder](https://www.medicare.gov/plan-compare/) — Medicare.gov (official) - [What Is the Medicare Plan Finder?](https://www.kff.org/faqs/medicare-open-enrollment-faqs/general-enrollment-information/what-is-the-medicare-plan-finder/) — KFF - [Medicare vs. Medicare Advantage Open Enrollment](https://www.ncoa.org/article/whats-the-difference-between-medicares-open-enrollment-period-and-medicare-advantage-open-enrollment/) — National Council on Aging - [AI Makes It Next to Impossible to Detect Scams](https://www.aarp.org/money/scams-fraud/detecting-ai-fraud/) — AARP --- # How to Write a Dating Profile Bio With ChatGPT (Without Sounding Fake) Source: https://aiforregulars.com/guides/ai-dating-profile-bio If you've stared at an empty dating profile box for ten minutes and typed "I enjoy traveling and trying new restaurants" before deleting it in frustration, you're not alone — and you're not behind the times either. Using ChatGPT to help write a dating bio has gone from a niche trick to something a real chunk of singles now do, across every age group, including people well past their dating-app debut. The catch is that the same thing that makes AI good at this — smoothing out awkward phrasing, filling gaps, sounding confident — is exactly what can make a bio come out sounding like it was written by, well, an AI. This guide is about getting the help without the tell-tale flatness. ## Why This Isn't Cheating — But Also Isn't a Shortcut Think of ChatGPT here the way you'd think of a friend who's good with words looking over your shoulder: useful for organizing what you already know about yourself and suggesting phrasing you wouldn't have landed on alone, unhelpful the moment it starts making things up. The 2026 Singles in America study from Match and the Kinsey Institute — a survey of roughly 5,000 U.S. singles — found that 39% have already used AI somewhere in their dating life, with adoption climbing fastest among younger daters (67% of Gen Z, 46% of Millennials) but real and growing even further up the age range (24% of Gen X, 5% of Baby Boomers). Where people start to object is a specific failure mode: 30% said they'd consider it a dealbreaker if someone let AI handle their messages entirely, and 34% felt the same about AI-altered photos. The line isn't "did AI touch this," it's "does this still describe a real person accurately." ## Before You Open ChatGPT Spend five minutes jotting down real, specific things about your actual life — not the version of yourself you think sounds most impressive. A dating profile lives or dies on specifics: not "I like music" but "I've been slowly working through my dad's old vinyl collection and just discovered I like Fleetwood Mac more than he ever let on." AI can help you phrase that. It cannot invent it for you, because it doesn't know your dad had a vinyl collection. Skip "write me a dating bio" — it has nothing to work with and will default to generic filler. Instead, paste in a few genuinely specific details: your actual hobbies (even mildly embarrassing ones), a recent trip or plan, what you do on a normal Saturday, something you're weirdly proud of, and what you're actually looking for (casual, serious, not sure yet). The more concrete and true the input, the less generic the output. Try: "Using these details about me, write four short dating profile bios in different tones — one warm and low-key, one a bit funny, one direct and to-the-point, one that leads with what I'm looking for. Keep each one under 300 characters." Seeing four versions side by side makes it much easier to spot which lines actually sound like you and which sound like filler, than judging a single draft in isolation. Go through the draft you like best and delete every sentence that would still be true if someone else wrote it about themselves — "I love to laugh," "looking for my partner in crime," "work hard, play hard." These phrases show up constantly precisely because they're safe and vague, which is the opposite of what makes a bio memorable. Replace the gap with one more specific detail from your notes, even if it feels slightly odd or oversharing. Odd and specific beats smooth and generic. This is the single most effective edit. If a sentence makes you wince, stumble, or think "I would never actually say that," change the wording until it sounds like your own voice, not a press release about you. Short, plain sentences read as more human than a paragraph with three perfectly balanced clauses — AI tends to reach for the latter unless you push back on it. Read the final version and confirm every claim in it is literally true — the job title, the hobby, the number of countries you've visited. It sounds obvious, but it's an easy step to skip when a confident-sounding draft is right in front of you. A bio that oversells you creates an awkward gap to explain on a first date; a plain, accurate one doesn't. The tell I look for in my own drafts is the "resume voice" — a bio that lists three positive traits in a row like it's applying for a job. Real profiles that get good responses usually commit to one specific, slightly weird thing rather than three safe, impressive-sounding ones. If ChatGPT gives you "adventurous, loyal, and ambitious," ask it to throw that sentence out entirely and describe one actual thing you did last month instead. ## Keep the Prompt as Private as the Profile Don't paste your full name, phone number, home address, or workplace into the chat just because you're describing yourself — none of that belongs in a dating bio anyway, and there's no reason for it to sit in an AI chat log either. Our guide on [what not to tell an AI chatbot](/guides/what-not-to-tell-ai-chatbot) covers the broader version of this rule if you're unsure where to draw the line in any AI conversation, not just this one. ## What This Doesn't Fix Writing a sharper bio doesn't protect you from the other side of AI on dating apps: matches who use AI to run the entire conversation for them, or profiles that aren't real people at all. If a conversation starts feeling too smooth, too fast, or oddly evasive about specifics, our guide on [spotting an AI match on a dating app](/guides/dating-app-ai-bot-detect) walks through the signals worth watching for, and [our guide to AI romance scams](/guides/ai-romance-scam-dating-app) covers what to do if it goes further than an awkward chat. ## Sources - [15th Annual Singles in America Study Released](https://www.eurekalert.org/news-releases/1143932) — EurekAlert! (Kinsey Institute / Match press release) - [Singles in America survey reveals how relationships, marriage, money and family are changing](https://phys.org/news/2026-09-singles-america-survey-reveals-relationships.html) — Phys.org - [Tinder to Expand Facial Verification Feature Across the U.S.](https://www.tinderpressroom.com/2025-10-22-Tinder-to-Expand-Facial-Verification-Feature-Across-the-U-S-,-Setting-a-New-Standard-for-Dating-Safety) — Tinder Newsroom --- # When AI Denies Your Loan, Rental, or Benefits: Your Rights and How to Push Back Source: https://aiforregulars.com/guides/ai-decision-appeal-rights Automated systems now make — or heavily influence — decisions about who gets a loan, who can rent an apartment, and who qualifies for a government benefit. When a rejection arrives, it can feel final and unexplained. Often the letter offers nothing beyond "we are unable to approve your application at this time." You have more options than that letter suggests. ## How to Tell If a Decision Was Automated Not every rejection is automated, but the clues are often there: - The letter uses phrases like "our automated review process," "based on your profile," or "our system" - A decision arrived within seconds or minutes of you applying online - The rejection gives no specific reason, or only a generic one - It references a credit score or "scoring model" If you're unsure, you can simply ask: "Was this decision made entirely by an automated system, or did a human reviewer assess my application?" Companies are often required to tell you, and many will answer straightforwardly. ## What Rights Do You Have? Rights vary by country and type of decision — but the direction of travel is clear. In many EU member states, people have long had the right to a meaningful explanation of automated decisions that significantly affect them, and the right to request human review. These protections are being strengthened under the EU AI Act — see our guide to [the EU AI Act and your rights](/guides/eu-ai-act-your-rights) for what that means in practice. In the United States, credit decisions specifically are regulated under the Fair Credit Reporting Act, which requires lenders to give "adverse action notices" explaining why credit was denied. Rights for other types of automated decisions vary by state and sector. A Consumer Reports survey of 2,022 U.S. adults found that 83% wanted to know what information an AI system used to make a decision about them, and 91% wanted the right to correct inaccurate data used in such decisions. The demand for these rights is clear — and in many places, the legal framework is beginning to match it. ## The Three-Part Request When you decide to push back, make three specific asks — in writing, ideally in one clear letter or email: **1. The specific reason for the rejection.** Ask them to explain, in plain terms, what factors led to the decision. "We were unable to approve your application" is not an explanation. You are asking for the actual reasons. **2. The data they used.** Ask what information — especially from third-party sources like credit bureaus or background-check services — was used to reach the decision. **3. A human review of your case.** Ask that a real person, not only the algorithm, look at your application again. This is especially important when you have context the system couldn't know: a recent improvement in your financial situation, a medical emergency that caused a temporary gap in payments, or a factual error in the data they hold about you. ## How AI Can Help You Draft Your Request This is where AI becomes a tool working *for* you. A chatbot like ChatGPT or Claude can help you: - Turn a difficult situation into a clear, polite, professional letter - Understand what the rejection letter is actually saying in plain English - Find the right words if you find formal writing difficult Try this prompt: > "I was rejected for [a personal loan / a rental application / a housing benefit] and I believe the decision was made automatically. Help me write a polite, professional letter asking: (1) the specific reasons for the decision, (2) what data was used about me, and (3) a human review of my case. I want to sound calm and professional, not angry." Review the draft, add any specific details that are relevant, and send it. Keep the original rejection letter and a copy of everything you send. I've used AI to decode rejection letters full of legal boilerplate. Pasting the letter text (with personal details removed) and asking "what is this actually saying, in plain English?" cuts through the fog immediately. What sounds like a final answer is often, in practice, an invitation to respond. ## How to Spot and Correct Wrong Data Automated systems rely on data — and data has errors. Common problems include: - **Credit report mistakes**: A debt that belongs to someone else, a payment marked late when it was on time, an account that should have been closed years ago - **Identity mix-ups**: Your record confused with someone who has a similar name, date of birth, or who previously lived at your address - **Outdated information**: A record that no longer reflects your current situation In most countries, you are entitled to a free copy of your credit report from the main credit bureaus, and you have the right to dispute errors you find. Contact the bureau in writing, explain the error clearly, and provide evidence: bank statements showing on-time payments, a letter confirming a debt was settled, or a government ID showing your correct address. Keep copies of every document you send and receive. If the error is in data held by the company that rejected you (rather than a credit bureau), ask them directly to correct it under your data subject rights. ## When to Escalate If the company won't explain, won't offer a human review, or you believe the decision was based on incorrect or discriminatory data: - **In the EU**: Your national data protection authority can investigate complaints about misuse of automated decision-making. Financial complaints also go to the relevant national financial ombudsman. - **In the US**: The Consumer Financial Protection Bureau (consumerfinance.gov/complaint) has handled complaints about financial products, though budget and staff cuts since 2025 have sharply reduced its capacity — check its current status, and consider your state attorney general's office as an alternative. - **In the UK**: The Financial Ombudsman Service handles banking and insurance disputes. The Information Commissioner's Office handles data rights. - **Everywhere**: Many legal aid organizations offer free initial consultations. They can tell you whether your situation is worth pursuing further. Don't assume a first rejection is final. Many companies reverse or modify decisions once a human reviews a well-documented, polite appeal. ## What to Watch Out For **Don't share sensitive details with a chatbot.** When using AI to draft your appeal letter, describe your situation in general terms. Do not paste your full credit card number, Social Security number, date of birth, or account statements. The letter does not need those details — and once you paste sensitive data into a chatbot, you have shared it with a third-party service. **AI can mis-state the law.** A chatbot may confidently tell you that you have a specific legal right — or don't have one. Laws differ by country, change over time, and vary by the type of decision involved. Use AI to help you write and understand documents, not to give you legal advice. If a legal right matters to your case, verify it with an official source or a professional. **Watch for appeal deadlines.** Some rejections have a short window in which you can formally appeal — missing it can waive your rights. Read your rejection letter carefully for any dates or response deadlines before you do anything else. **Stay calm and keep records.** Escalation works better with documented, professional communication than with frustrated demands. Keep a note of every contact: who you spoke to, when, what was said or written, and what happened next. ## What to Try Next If your rejection involved health insurance specifically, [how to appeal an AI health insurance denial](/guides/ai-health-insurance-appeal) has a step-by-step process built around that situation. If you are dealing with confusing paperwork from a government agency, [using AI to fill out government forms](/guides/ai-government-forms-help) can help you understand what they are asking for and how to respond clearly. --- # AI Detectors Tested: Accuracy, False Positives and What Teachers Should Know Source: https://aiforregulars.com/guides/ai-detectors-tested When ChatGPT became widely available in late 2022, a new category of software appeared almost immediately: AI detectors. The promise was appealing — paste in text, get a percentage, know if a human or machine wrote it. Schools started subscribing. Parents started checking homework. Employers started screening job applications. The reality turned out to be much messier. Several years of research and real-world use have made one thing clear: these tools are genuinely useful for understanding statistical patterns in text, and genuinely unreliable for judging any single piece of writing. Understanding why — and what the tools actually measure — is the most important thing before using one. ## How AI Detectors Work Every AI detector analyzes text for statistical properties that tend to differ between AI output and human writing. The two main signals are **perplexity** and **burstiness**. **Perplexity** measures how surprising each word choice is, given the words before it. AI tends to pick highly probable, predictable words. Human writers make more unexpected choices — a metaphor, a slang term, a long word where a short one would do. Low perplexity suggests machine-like predictability. **Burstiness** measures how much sentence length varies. Humans tend to mix very short sentences with longer ones in an uneven rhythm. AI tends toward more uniform sentence lengths, especially in formal writing. These are reasonable ideas. The problem is that many humans write in ways that score as low-perplexity and low-burstiness — especially people who write carefully, formally, or in English as a second language. ## Comparison: What to Look for in a Detector The table below compares the main categories of AI detection tools across criteria that matter for practical use. It does not include invented accuracy percentages — those vary too much by use case and prompt style to be meaningful. The qualitative ratings reflect patterns widely reported in independent testing and published research. | Criterion | Free browser tools | School / LMS integrations | API-based tools | Open-source tools | |---|---|---|---|---| | **False positive risk on ESL text** | High | High | Moderate to High | Varies widely | | **False positive risk on formal human writing** | High | Moderate to High | Moderate | Varies | | **Detection of lightly edited AI text** | Low | Low to Moderate | Moderate | Low | | **Detection of paraphrased AI text** | Very Low | Low | Low | Very Low | | **Explains why it flagged text** | Rarely | Sometimes | Sometimes | Depends on tool | | **Audit log / evidence trail** | No | Sometimes | Yes | No | | **Cost** | Free | Subscription (per school) | Pay per use | Free | | **Suitable as sole evidence of cheating** | No | No | No | No | The last row is the same across every category, because no tool currently available meets the standard of evidence needed to accuse someone of academic dishonesty on its own. I ran several essays through three different free detectors to see what came back. The same essay flagged as "likely AI" by one tool came back as "mostly human" by another. One of the essays I tested was written entirely by a colleague who is a very careful, structured writer. It flagged as AI on two of three tools. That should tell you something about how much weight to put on any single result. ## The False Positive Problem False positives — cases where the detector flags human writing as AI — are the central failure mode. They are well-documented, widely reported, and serious. Some groups are more likely to be flagged than others: **Non-native English speakers** write in patterns that match AI statistical signatures more closely. Formal vocabulary, careful grammar, and structured paragraphs are all traits that score as low-perplexity. **Students who write formally** for academic assignments — the way they are often taught to write — produce text that many detectors find suspicious. **Writers who draft carefully and edit** tend to produce smoother, more predictable text than writers who dash things off. There is no way to know from the outside whether a false positive is happening in any given case. That is the core problem. A result of "98% AI" tells you that the text scores similarly to AI-generated text. It does not tell you that AI generated it. ## What Detectors Cannot Catch Modern AI, when prompted to write conversationally, with imperfections, or in a specific person's style, can produce text that scores as fully human on nearly every detector. Anyone motivated to evade detection can do so easily: - Ask the AI to "write like a high school student" or "make it sound casual" - Edit a few sentences manually after generating - Run the text through a free paraphrase tool - Ask the AI to vary sentence lengths and include contractions This means a student who takes cheating seriously is unlikely to be caught by a detector. A student who did not try to evade detection might be caught. The tools end up being slightly better at catching careless use of AI than deliberate misuse. ## What Actually Helps For teachers, the more durable approaches involve the writing process rather than the final product: - Ask students to submit drafts at multiple stages, not just a final document - Include in-class writing components that mirror out-of-class assignments - Ask students to discuss their work: what sources they used, what was difficult, what they would change - Look for inconsistencies between a student's verbal explanation and what the essay argues A student who used AI to write an essay will typically struggle to explain it. A student who wrote it — even with AI assistance for research or editing — will have something to say about their own thinking process. For parents, the same principle applies. If you are curious whether your child used AI for an assignment, ask them to walk you through what they did. The conversation is more informative than any detector. ## Tools You'll Encounter Three tools come up most often in schools and households. Knowing what they are helps set expectations before trying one. **GPTZero** (gptzero.me) was built specifically for educators. A free account scans up to 10,000 words per month; paid plans start at around $15 per month. It highlights the individual sentences that triggered detection and integrates with Google Classroom. **Turnitin AI Detection** is not available as a standalone purchase — it is bundled into institutional Turnitin subscriptions and controlled by the school or district, not by individual teachers. **Scribbr AI Detector** (scribbr.com) is free with no account required, limited to 1,200 words per submission. It is the most transparent of the three tools about its own limitations. All three produce false positives, all three can be evaded in a few minutes of editing, and none meets the standard of evidence needed to treat a result as proof of anything. ## What to Try Next To understand what patterns actually show up in AI writing — beyond what a detector measures — read [How to Tell If a Text Was Written by AI](/guides/how-to-tell-if-text-is-ai). If you want a practical guide for talking to your kid about AI and homework, [My Kid Uses ChatGPT for Homework — A Parent's Guide](/guides/kids-chatgpt-homework-parents-guide) has a calm, step-by-step approach. --- # What Happens to Your Accounts, Photos, and AI Chats After You Die — and How to Plan Now Source: https://aiforregulars.com/guides/ai-digital-estate-plan When someone passes away, the people they leave behind often can't access the simplest things — the email inbox, the folder of family photos in the cloud, the streaming subscriptions still charging every month. Passwords are locked away, companies have no standard process for grieving families, and years of memories can sit unreachable behind a forgotten login screen. The good news is that this is one of the easier parts of estate planning to address — if you take an afternoon to set things up now. ## What the Law Says About Your Digital Accounts Your email, cloud storage, and social media accounts are governed by the same kind of legal rules as physical property — but the specifics are newer and still evolving. As of February 2025, 47 US states have adopted some version of RUFADAA (the Revised Uniform Fiduciary Access to Digital Assets Act), which gives your legal executor or trustee the right to manage your digital accounts. What RUFADAA does not automatically grant: the right to read your private communications. Most versions specifically protect message content — your executor can typically close your Gmail account, but not necessarily read your emails or messages without your prior consent. In plain terms: the law helps with account management; the actual keys to your memories and conversations still depend on you leaving them behind. **The practical gap:** Laws change slowly; tech companies move faster, and their terms of service still treat accounts as non-transferable. The built-in tools that actually matter — Facebook Legacy Contact, Google Inactive Account Manager — are worth setting up today regardless of what your state's law says. ## Your Photos and Cloud Storage This is where most families feel the loss most sharply. Decades of photos, home videos, scanned documents — stored in cloud accounts that a grieving spouse or child cannot open. **Google Photos and Google Drive:** Google's Inactive Account Manager lets you designate up to 10 trusted people who receive access, or a full download of your data, if your account goes dormant for a period you choose (3, 6, 12, or 18 months). You can also tell Google to delete your account after that window. Setting it up takes about five minutes at myaccount.google.com/inactive. **Apple iCloud:** Apple has a Digital Legacy program that lets you name legacy contacts who can request access to your iCloud data, including photos, notes, and files. **The gap:** If you don't set these up in advance, your family will need a court order to gain access — a process that takes time, costs money, and isn't guaranteed. Setting it up yourself takes a few minutes and removes that burden entirely. ## Social Media: What You Actually Want to Happen **Facebook** gives you two choices: designate a Legacy Contact who can write a pinned tribute post and manage your profile after you're gone (but cannot read your messages or log in as you), or request that your account be deleted entirely when you die. Go to Settings → Memorialization Settings. **Instagram** allows memorialization — the account stays up labeled "Remembering [Name]" — or removal upon a family member's request with proof of death. **X (formerly Twitter):** Family members can request account deactivation; there's no legacy contact feature as of 2026. **The most useful thing you can do:** Write down what you want to happen to each account, and tell someone. A clear document that says "delete my Facebook, keep Instagram as a memorial, give my sister the email login" will do more practical good than any legal instrument most families will actually use in time of grief. ## Your AI Chat History — and Why It's Different This one doesn't get enough attention yet. If you've used ChatGPT, Gemini, Claude, or any AI assistant regularly, your chat history contains something your email doesn't: the questions you asked when you were worried, confused, or thinking something through in private. Most AI platforms — including OpenAI's ChatGPT — have no provisions for heir or executor access. There's no "legacy contact" feature for your AI chat history. What there is: an account that anyone who knows your password can open and read everything, or delete everything, with no verification that you've passed. This matters because your AI chat history might include conversations about health conditions, financial worries, relationship questions, or other private thoughts. Right now, if someone logs into your account after you're gone, they can read all of it — or erase it permanently. > To export your ChatGPT history before doing anything else: go to Settings → Data Controls → Export data. You'll receive an email with a download link within a few hours. **What you can do today:** Decide what you want to happen to your AI chat history, and document that decision. Most platforms let you turn off history saving, delete it, or export it. If you want it deleted after your death, leave clear instructions and account access information for someone you trust — in a password manager with emergency access, or a sealed envelope kept with your other important documents. I keep a short document with my main account logins — email, cloud storage, the password manager master key — in a sealed envelope with our other important papers. My spouse knows where it is and what to do. It's not a sophisticated tech solution, but it's the one that actually gets used when it matters. ## Subscriptions: The Bill Nobody Cancels This is the practical, unglamorous side of digital estate planning that catches families off guard: subscriptions keep charging a deceased person's credit card for months because nobody knew they existed. Make a list of every monthly or annual charge: streaming services, cloud storage (iCloud+, Google One, Dropbox), news subscriptions, software licenses, phone plans. Your bank statement is the fastest place to find them. This list is a genuine gift to whoever handles your finances afterward. ## Cryptocurrency: No Keys Means No Money Cryptocurrency stored in a personal wallet — not on an exchange — is only accessible with the private key or seed phrase. If that information dies with you, the cryptocurrency is gone. Not recoverable by lawyers, courts, or the exchange. Exchanges like Coinbase have processes for estate claims, but self-custody wallets do not. If you own any meaningful amount of cryptocurrency, getting the private key or seed phrase into trusted hands — securely but accessibly — is urgent. ## Your AI Likeness After You're Gone AI tools can now clone a person's voice or generate a realistic video likeness from a small amount of source material. What rights do your family have if someone creates an AI "version" of you — for a scam, for a podcast, or for entertainment — after you're gone? The law is beginning to catch up. Washington state passed legislation effective June 2026 setting civil penalties of up to $3,000 per violation for unauthorized digital replicas of a person's voice or likeness. At the federal level, the bipartisan NO FAKES Act — still pending as of mid-2026 — would treat a person's voice and likeness as an inheritable property right, giving your family legal standing to challenge unauthorized use. Estate lawyers are already writing about this in 2026 primers on AI and the digital afterlife, including resources from elderlawanswers.com. If you feel strongly about how your likeness may or may not be used, ask an estate attorney about including a digital likeness provision in your will or trust. A signed, dated written statement of your wishes has more force than nothing, even without a formal document. ## A Simple Digital Estate Checklist You don't need to do all of this today. But each item is something a family member will need to figure out in a difficult moment if you don't: - **Inventory:** List your major accounts — email, cloud storage, social media, subscriptions, financial accounts, AI chat tools. - **Passwords:** Use a password manager and set up emergency access for a trusted person, or keep a physical record in a secure location. - **Google Inactive Account Manager:** Set up at myaccount.google.com/inactive. - **Facebook Legacy Contact:** Configure in Settings → Memorialization Settings. - **Apple Digital Legacy:** Add a legacy contact in Settings → [Your Name] → Legacy Contact. - **AI chat history:** Decide whether to delete it, export it, or leave access instructions for someone. - **Cryptocurrency:** Ensure private keys or seed phrases are in trusted hands. - **Subscriptions:** Leave a list so they can be cancelled promptly. - **Digital likeness:** If it matters to you, note your wishes in writing. ## What to Try Next Understanding what data AI platforms already hold about you is a good starting point — [What Does AI Know About You?](/guides/what-does-ai-know-about-me) covers the data trail you leave when using ChatGPT and similar tools. If you're wondering whether AI can help draft a simple will or other legal documents, [Can You Use AI to Write Legal Documents?](/guides/ai-legal-documents-safe) walks through where AI genuinely helps and where you still need a human. --- # Preparing for Your Doctor's Appointment with AI: Step by Step Source: https://aiforregulars.com/guides/ai-doctor-visit-prep The average primary care visit lasts less than 20 minutes. That's not much time to describe your symptoms, ask everything you wanted to ask, and understand what the doctor tells you. AI can't add time to your appointment, but it can help you use every minute of it better. This tutorial walks through using ChatGPT (or any general AI chatbot) before, during, and after a doctor's visit. No medical training required. **Important reminder:** AI helps you prepare and understand — your doctor does the actual evaluating and diagnosing. Before opening any AI tool, spend five minutes writing down what you've been experiencing in your own words. Don't try to sound medical — just describe it the way you'd tell a friend. Note the symptom, when it started, how often it happens, what makes it better or worse, and how much it's affecting your day. This raw material is what you'll feed into the AI in the next step. Open ChatGPT (or whichever AI you prefer) and paste your notes in. Use a prompt like this: ``` Here is how I've been feeling lately. Help me organize this into a clear, short symptom summary I can share with my doctor: [paste your notes] ``` The AI will restructure your description into a more readable format — chronological order, grouped by body system, or whatever makes the most sense. Read it over and correct anything that doesn't sound right. You know your body; the AI is just helping with the organization. Now ask AI to generate questions you might want to raise with your doctor: ``` Based on these symptoms, what are 5 to 7 questions I should ask my doctor at my appointment? ``` You can also prompt it with what you already know: ``` My doctor previously mentioned I might have [condition]. What questions should I ask to understand my options? ``` Review the list and cross off anything that doesn't apply. Add any questions you already had. The goal is to arrive with a written list so you don't forget things once you're in the exam room. If a previous appointment, lab result, or online patient portal used terms you didn't understand, this is a good time to ask: ``` What does "elevated TSH" mean in simple terms? I saw it on my blood work. ``` ``` My last visit notes mentioned "mild mitral regurgitation." Can you explain what that is in plain English? ``` AI is good at this kind of translation. Just remember it's giving you a general explanation, not a specific assessment of your results. Only your doctor can interpret what a number means for you specifically. If you already have a sense of what the appointment is about, ask the AI to help you understand the possibilities: ``` What are the most common reasons a doctor would order a colonoscopy? What should I expect? ``` ``` If my doctor recommends starting blood pressure medication, what are some things I should know going in? ``` This reduces the chance that you'll feel caught off guard and forget to ask something important in the moment. I started doing this before every specialist appointment and noticed a real difference. My cardiologist told me I was "one of the most prepared patients" she'd seen. That preparation came from spending about 15 minutes with ChatGPT the night before. The questions I brought up led to a test she hadn't initially planned — and it turned out to be useful. Doctors write discharge summaries quickly, often in shorthand. If you leave with a paper or portal message you don't fully understand, paste the relevant parts into ChatGPT: ``` Here is what my discharge notes say. Can you explain what each instruction means in plain English? [paste notes — but remove your name, date of birth, and any identifying information] ``` Pay attention to: - Any follow-up actions you're supposed to take - Medications added or changed - Warning signs listed under "return to ER if…" If the AI's explanation contradicts what you remember the doctor saying, call the office to clarify. Don't guess. Your symptom log and question list are useful beyond this visit. Keep a simple document — even a notes app on your phone — where you track: - Symptoms and when they started or changed - Medications (name, dose, when you started) - Questions that were answered (and what the answer was) - Questions still open Next appointment, you'll have a running history instead of starting from scratch. AI can help you update it: just paste your current notes and add "update this with my latest visit" after you describe what happened. ## What to try next For help understanding what medications you've been prescribed, see [How to Use AI to Understand Your Prescription and Side Effects](/guides/ai-prescription-explainer). And if you'd like to get better at writing prompts so AI gives you more useful answers, [How to Write Prompts](/guides/how-to-write-prompts) covers the basics in plain English. --- # Using AI to Understand Contracts and Bills (Plain-English Translator) Source: https://aiforregulars.com/guides/ai-explain-contracts Contracts are written by lawyers to protect companies, not to be understood by the people signing them. Bills arrive with line items that seem designed to be ignored. AI has flipped this — you can now get a plain-English explanation of almost any clause in seconds. Here's how to do it safely and effectively. ## Before You Start: The Privacy Warning This is the most important step. Before pasting any document into an AI tool, **remove your personal information:** - Your full name, address, and phone number - Social Security or national ID numbers - Account numbers and policy numbers - The other party's identifying information You're asking AI to explain the *language* of a document, not to analyze your specific situation. The clause "Tenant is responsible for all repairs under $150" means the same thing whether you paste it alone or with your name and address attached — remove the personal details first. Don't paste an entire 20-page lease and ask "what does this say?" That produces an overwhelming response and often misses what you care about. Instead, find the specific section that worries you — the early termination clause, the late fee terms, an unexpected line on your bill — and copy just that part. A paragraph or two is the right amount to work with. This structure works for most documents. Fill in your document type and paste the redacted section: ``` Explain the following section of a [lease / phone contract / insurance policy] in plain English. I'm not a lawyer. Focus on what it means for me as the [tenant / customer / policyholder], and flag anything that seems unusual or that I should ask the other party about. [paste the redacted clause here] ``` This framing gets you a more practical answer than just asking "what does this mean?" Phone bills are full of fees with vague names. Use this prompt: ``` Here is a section of my phone bill. Explain each line item in one sentence. For any fee that seems unusual or potentially avoidable, flag it. [paste the bill section — remove account numbers] ``` Common discoveries: administrative fees that can sometimes be removed with a phone call, data overage charges that suggest a plan change would save money, or insurance premiums you forgot you were paying. Insurance letters use particularly dense language. This prompt cuts through it: ``` I received the following letter from my insurance company. Explain what it is telling me to do, what deadline (if any) I have, and what happens if I don't respond. [paste the letter text — remove policy numbers and personal details] ``` This gets you the three things you actually need: what's happening, when, and what the consequence of doing nothing is. I've used this for insurance renewal notices I would have otherwise skimmed past. One of them had a significant rate increase buried in paragraph four. Asking AI to summarize "what's actually changing" surfaced it immediately. After the initial explanation, ask specific follow-up questions to build a fuller picture: ``` Is this early termination fee typical for apartment leases? ``` ``` What questions should I ask the landlord about this clause before signing? ``` ``` What does "pro-rated" mean in the context of this rental agreement? ``` You're not trying to get legal advice — you're trying to know what you're agreeing to before you sign it. AI explanation is the right tool for: - Routine bills and standard contract language - Understanding what a clause says before asking questions - Deciding whether something is worth flagging to a professional You need a lawyer, accountant, or other professional for: - Contracts involving significant money (property purchases, business agreements) - Anything you're about to sign that has obligations you can't easily exit - Documents related to a dispute, lawsuit, or formal legal notice A free consultation with a local legal aid society can tell you whether your situation needs paid help. ## Not Legal Advice AI explanations help you understand language — they don't constitute legal advice, and they can be wrong, especially on jurisdiction-specific rules or unusual clause interpretations. Use AI to understand what you're reading; use a professional to decide what to do about it. ## What to Try Next If you need to dispute a charge after understanding your bill, [how to reach a real human in customer service](/guides/bypass-ai-customer-service) has the phrases and tactics that actually work. For understanding tax documents specifically, [can AI help me file my taxes](/guides/ai-for-taxes) covers what's safe to ask and what isn't. --- # How to Spot AI-Generated Fake Profiles and Posts on Facebook Source: https://aiforregulars.com/guides/ai-fake-facebook-profiles Not everyone on Facebook is who they claim to be. Fake profiles have existed for years, but AI has made them faster to create and harder to detect at a glance. A convincing profile photo of a real-looking person can now be generated in seconds. A backstory, a set of posts, even a comment history can be produced just as fast. Here's how to spot them before you accept a friend request, join a group, or trust a recommendation. ## Red Flags in the Profile Photo The profile photo is usually the first thing to check. AI-generated face images are increasingly realistic, but they still have common tells. **The photo looks almost too polished.** Real people's photos have lighting variations, context in the background, and occasional imperfections. AI-generated faces often look studio-lit and perfectly symmetrical in a way real photographs rarely are. **Look at the edges — especially ears and hair.** AI image generators often struggle with fine detail at the boundaries of a face. Hair strands may blend unnaturally into the background. Earrings may be asymmetrical or partially missing. Glasses may have distorted or mismatched frames. **The background is generic or blurred out completely.** Legitimate profile photos usually have a real place in the background — a room, a street, a vacation spot. AI-generated images often feature a blurred or plain background because there's no real context to show. **Run a reverse image search.** On a computer, right-click the profile photo and choose "Search image" in Chrome or Firefox. On mobile, hold the photo and look for a copy or search option, then paste it into [Google Images](https://images.google.com) or [TinEye](https://tineye.com). If the same face appears on multiple unrelated accounts or websites, it is almost certainly fake. ## Red Flags in the Account History The profile photo is the first check. The account history is often more revealing. **Recent account creation.** Facebook shows when an account was created under "About." Fake accounts are often new — created within the last few months. Scam campaigns frequently create large batches of profiles at once. **No personal photos beyond the profile picture.** Real people's timelines have birthday photos, family snapshots, and event pictures. A profile with one or two photos total — usually the profile picture and a cover image — is suspicious. **Few real-life connections.** Fake profiles often have a small number of friends, many of whom are also new or look suspicious. Mutual friends you don't recognize from real life are worth noticing. **Posts that look like content, not life.** Real people post personal things — opinions, events, family moments. Fake profiles post a stream of articles, memes, or politically charged content with little personal commentary. This is because the "person" is a managed account with no real life to share. When someone I don't recognize sends me a friend request, I look at three things in under a minute: when the account was created, how many personal photos they have, and whether any mutual friends are people I actually know in real life. If two out of three look off, I decline and move on. ## Viral-Bait Groups and Coordinated Pages Fake profiles don't usually operate alone. They're often part of coordinated networks that use Facebook groups to spread content. **Watch out for groups with names designed to feel local or personal** — names like "Concerned Parents of [Your City]" or "[Your Neighborhood] Community Watch" that suddenly appear and post alarming or politically charged content. These groups use a local feel to earn trust quickly, but they were created elsewhere and have no real community connection. **Engagement patterns in these groups look automated.** The same few profiles comment on every post within minutes. Comments are often generic ("So true!" or "This is terrifying!") rather than personal and varied. Real communities have a range of voices, including people who disagree with each other. **Links in these groups often go to unfamiliar sites.** The goal is frequently to drive traffic to content-farm articles (the kind covered in the [fake news guide](/guides/ai-generated-fake-news)) or to collect email addresses through a sign-up form. ## What to Do When You Find a Fake Profile **Don't engage.** Don't comment, argue, or try to expose them publicly. Engagement — even hostile engagement — helps the account look more active and legitimate to the platform algorithm. **Report the profile directly.** Click the three dots on the profile or any post, choose "Find support or report," and select the most fitting option — usually "Fake account" or "Spam." Facebook and Instagram investigate reports and often remove fake accounts, especially when multiple people report them. **Leave groups that seem coordinated.** If you joined a group and it's filled with what look like fake accounts pushing a specific narrative, you can leave quietly without engaging or announcing it. **Be selective about friend requests you didn't initiate.** Fake profiles need to connect with real people to gain reach. Staying selective about who you accept limits their ability to use your network. ## What to Try Next Fake profiles often go hand in hand with fake photos more broadly — [How to Spot AI-Generated Photos](/guides/spot-ai-generated-photos) covers the visual tells in more depth. And if a suspicious account has also sent you a voice message or called you, [AI Voice Cloning Scams](/guides/ai-voice-cloning-scams) explains how that piece of the puzzle works. ## Sources - [How to report a Facebook profile or Page pretending to be you or someone else](https://www.facebook.com/help/306643639690823) — Meta - [Search with an image on Google](https://support.google.com/websearch/answer/1325808) — Google - [Scams — consumer advice](https://consumer.ftc.gov/scams) — Federal Trade Commission --- # How to Fact-Check a Headline with AI in 60 Seconds Source: https://aiforregulars.com/guides/ai-fake-news-fact-check A suspicious headline lands in your feed. Your first instinct might be to share it — or argue about it — before you've verified anything. This workflow gives you a way to check in about 60 seconds, using AI tools you likely already have access to. You don't need an account, an app, or any technical skills. You just need a browser. Before you fact-check anything, be clear about what the claim actually is. Headlines are often vague or misleading even when the article underneath is more nuanced. Ask yourself: *What specific fact would have to be true for this headline to be accurate?* Write it down or copy it to your clipboard. A headline like "Scientists Say Coffee Causes Cancer" would need a specific study from a named institution to be real. That's what you'll search for — not the headline itself. Standard chatbots like a basic ChatGPT window aren't ideal for fact-checking because they don't always pull from current sources. Instead, open one of these: - **Perplexity AI** — free at [perplexity.ai](https://perplexity.ai), shows inline citations by default - **Microsoft Copilot** — free at [copilot.microsoft.com](https://copilot.microsoft.com), connected to the live web - **Google AI Overview** — appears automatically at the top of many Google searches All three pull from live web sources and show you where the information came from. That's what separates them from a regular chatbot for this purpose. Don't just paste the headline. Ask a question that forces the AI to find sources. A reliable prompt looks like this: ``` Is it true that [specific claim]? Show me which news outlets or official sources confirm or deny this. ``` For example: ``` Is it true that the FDA banned food dye Red 3 from all foods in 2025? Show me which official sources or major news outlets reported this. ``` The more specific your question, the more useful the answer. Avoid yes/no questions — you want sources, not just a verdict. This step is the most important one. The AI summary can still be wrong or incomplete. What matters are the sources it links to. Look for citations from: - Major newspapers (New York Times, Washington Post, Guardian, AP, Reuters) - Government websites (.gov domains) - Academic institutions (.edu domains) - Established broadcast networks (BBC, NPR, PBS) If the sources are all from sites you've never heard of, or if there are no citations at all, treat the claim as unverified — regardless of how confident the AI summary sounds. I've learned to be especially careful when an AI confidently summarizes something but then links to obscure sites I can't recognize. Confidence in the summary does not mean the sources are reliable. I always click at least one link and read the actual article before I decide anything. Open a new browser tab and search the claim on a site you already trust. A reliable shortlist: | Outlet | Best for | |--------|----------| | AP News (apnews.com) | Breaking news, US and international | | Reuters (reuters.com) | Business, global news | | BBC News (bbc.com/news) | International stories | | NPR (npr.org) | US politics, health, science | | Snopes (snopes.com) | Viral claims and urban legends | | PolitiFact (politifact.com) | Political statements | | FactCheck.org | US election and policy claims | If none of these have covered a story that supposedly happened recently, that's a strong signal to pause before sharing. After your check, you have three options: **It checks out** — share if you want; you could even mention you verified it. **It doesn't check out** — don't share. If a friend already shared it, a private message explaining what you found is usually well-received. **You're not sure** — wait. A real story will still be real tomorrow. There's no cost to waiting; there's a real cost to spreading false information. If you found the story on social media, most platforms let you report it under a "false information" or "misleading content" option. ## What to Try Next Now that you have a fast verification workflow, [AI-Generated Fake News: Spotting Content Farms](/guides/ai-generated-fake-news) explains the broader patterns to watch for — so you'll start recognizing suspicious sources before you even click. If you're also wondering whether text you're reading was AI-written rather than human-written, [How to Tell If Text Is AI-Generated](/guides/how-to-tell-if-text-is-ai) covers the most reliable signals. ## Sources - [Find information in faster & easier ways with AI Overviews in Google Search](https://support.google.com/websearch/answer/14901683) — Google - [Perplexity AI](https://perplexity.ai) — Perplexity AI - [Microsoft Copilot](https://copilot.microsoft.com) — Microsoft - [FactCheck.org](https://www.factcheck.org/) — Annenberg Public Policy Center, University of Pennsylvania - [Snopes](https://snopes.com) — Snopes Media Group - [PolitiFact](https://politifact.com) — Poynter Institute --- # Fake Hotel Sites Made with AI Are Everywhere This Summer — How to Spot Them Source: https://aiforregulars.com/guides/ai-fake-travel-booking Summer is peak season for travel — and AI has made travel scams harder to spot than ever before. Booking.com reported in 2024 that AI had fuelled a 500–900% increase in travel scams over the previous 18 months. Separately, the FTC's Consumer Sentinel Network logged more than 58,000 travel, vacation, and timeshare fraud reports in 2024 totaling $274 million in losses — and that only counts the people who filed a report. This summer the fake sites look professional: polished photos, convincing reviews, realistic pricing — all generated by AI in a matter of hours. The good news is that scammers are still making the same fundamental mistakes, and once you know what to look for, most fake sites give themselves away quickly. ## Red Flag 1: The Domain Doesn't Quite Match the Real Thing Legitimate booking platforms use short, simple URLs: booking.com, hotels.com, expedia.com, airbnb.com. Fake sites mimic these but add words, hyphens, or swap a letter or two: booking-hotels-secure.com, expedia-deals-summer.net, airbnb-direct-rentals.com. **Read the address bar before you enter any personal information or payment details.** Scammers register domains that look almost right at a glance, counting on the fact that most people don't read URLs carefully enough to catch a subtle difference. **Look up when the site was registered.** Go to whois.domaintools.com, enter the domain, and check the creation date. Legitimate booking platforms have existed for years or decades. If a site claiming to be a major hotel-booking service was registered three weeks ago, close the tab. **A padlock icon means the connection is encrypted — not that the site is honest.** Any website, including a fraudulent one, can display the padlock. HTTPS tells you nobody is intercepting the connection; it says nothing about whether the destination is real. ## Red Flag 2: The Photos Look Perfect — Maybe Too Perfect AI image generators can create hotel lobbies, pools, and guest rooms that don't exist anywhere in the real world. These images tend to look more polished than real hotel photography — no awkward furniture angles, no mismatched towels, no evidence that actual people have ever been in the room. **Do a reverse image search on the property photos.** Right-click any photo and choose "Search image with Google" (or press and hold on mobile). If the image shows up on a stock-photo site or appears as a different hotel in a different country, the listing is using stolen photos — and almost certainly a scam. **Know what AI-generated photos get wrong.** AI struggles with small, specific details: text on signs tends to be garbled nonsense, reflections in mirrors and windows look slightly wrong, and repeating patterns on towels or carpets sometimes drift in unnatural ways. Furniture proportions can look almost right but slightly off. If everything looks perfect and nothing looks lived-in, take a closer look. **Check the address on Google Maps.** A real hotel will appear on Maps with photos taken by actual guests — imperfect angles, variable lighting, the background noise of a place people genuinely visit. A hotel that exists only as a slick website, with no Maps presence at all, is a serious warning sign. ## Red Flag 3: The Price Is Too Good and the Pressure Is High A legitimate discount is usually 10 to 20 percent off the going rate, not 60 percent below what every other platform is showing. Scammers use dramatically low prices to attract clicks, then pair them with urgency tactics — countdown timers, "only 1 room left," "deal expires in 5 minutes" — designed to stop you from taking the time to verify the site. **Compare prices on two or three platforms you trust.** If the rate on an unfamiliar site is dramatically lower than what Booking.com, Expedia, and the hotel's own website are all charging, that gap is the scam. **Slow down when someone rushes you.** Urgency is a manipulation tactic, not a feature of legitimate travel deals. No real hotel will rescind an available room because you took ten minutes to check a second source. When I'm researching a place to stay, I open the hotel's own website in a separate tab before booking through any third-party site. Most hotels now have a "Book Direct" button, and they often match or beat third-party prices — no middleman, no chance of a fake platform in between. ## Red Flag 4: They Ask to Move the Conversation Off the Platform This is one of the most reliable warnings. You find a listing, express interest, and the "host" or "agent" asks you to continue on WhatsApp, Telegram, or by email — usually with a story about their inbox being full, the platform having technical issues, or the ability to offer you a better deal privately. The moment you leave the platform, you lose its payment protection, dispute resolution, and fraud guarantees. The scammer then collects payment directly and disappears. The FTC's June 2026 consumer alert on summer travel scams covers this same broader pattern — fake booking sites and lookalike ads — and its core advice applies just as much here: never pay by wire transfer, gift card, or cryptocurrency, no matter how convincing the seller sounds. **Never complete a booking or payment outside the official platform.** If you found it on Airbnb, book it on Airbnb. If a "hotel agent" contacts you through an ad or search result and immediately asks to move to a private channel, decline and look elsewhere. ## Red Flag 5: They Won't Accept a Credit Card This is the clearest signal that something is wrong. Every legitimate hotel and booking platform accepts credit cards. Scammers avoid them because credit card charges can be disputed — and chargebacks cost them. If a site or "agent" insists on payment by: - **Wire transfer** — money sent and gone, nearly impossible to recall - **Gift cards** — untraceable, a classic fraud payment method - **Cryptocurrency** — irreversible and anonymous ...that is not a payment preference. That is a warning. The FTC's advice is explicit: never pay for travel with gift cards, wire transfers, or cryptocurrency. If the seller insists on these methods, the booking is not real. ## Safe Booking Habits That Significantly Cut Your Risk **Stick to platforms and hotels you've used before.** Booking.com, Expedia, Hotels.com, Airbnb, and VRBO have fraud detection teams, verified listings, and dispute processes. They're imperfect — fake listings do occasionally appear — but they're dramatically safer than a site you've never heard of. **Call the hotel directly to confirm the booking.** Find the hotel's phone number on Google Maps (not on the suspicious website), call them, and ask if they have a record of your reservation and if they work with the platform you found them on. This takes less than two minutes and catches most scams. **Always pay by credit card.** Debit cards offer weaker protections. Other payment methods offer almost none. **Screenshot everything before you pay.** Save the URL, booking confirmation page, advertised price, and any chat messages. Scam sites often disappear within hours of collecting payment; your screenshots are your only evidence. ## What to Do If You Already Paid Act quickly — time matters more than it might seem. **If you paid by credit card:** Call your card issuer immediately and dispute the charge. Explain that the booking site was fraudulent. In the US, the Fair Credit Billing Act generally gives you 60 days from your billing statement to dispute it in writing; in the UK, Section 75 or a chargeback claim through your card network are the equivalent routes. Earlier is always better regardless of which applies to you. **If you paid by wire transfer:** Call your bank the same day. Banks sometimes can recall a wire if it hasn't cleared yet — the window is narrow, often just a few hours, but it's worth the call. **If you paid by gift card or cryptocurrency:** Recovery is extremely difficult, but document everything — card numbers, transaction IDs, screenshots — and report it anyway. **Report the fraud:** File with the FTC at reportfraud.ftc.gov and with the FBI's Internet Crime Complaint Center at ic3.gov. Reporting doesn't guarantee recovery, but it contributes to investigations that catch these operations. ## What to Watch Out For Even legitimate booking platforms occasionally list properties with misleading photos or inaccurate descriptions — that's a frustration, not necessarily fraud. This guide focuses on outright scams: sites where the property doesn't exist, the booking is never confirmed, or the payment goes to someone with no intention of providing the accommodation. If you book a real hotel through a real platform and the room doesn't match the photos, that's a complaint for the platform's customer service team, not evidence of a criminal scam. Keep your booking confirmation email, contact the platform first, and escalate from there. Social media ads are one of the fastest-growing entry points for travel fraud — scammers buy targeted placements on Facebook, Instagram, and TikTok that look indistinguishable from legitimate promotions. Be especially skeptical of travel deals you see advertised in your social feed, even if they look professionally produced. ## What to Try Next If you're using AI tools to plan your trip, [Best AI Travel Planners: What They're Actually Good At](/guides/best-ai-travel-planners) covers which tools genuinely help and where they fall short. Since travel scam tactics overlap closely with phishing, [How to Spot an AI Phishing Email Before It Gets You](/guides/spot-ai-phishing-emails) is worth reading alongside this one. ## Sources - [Booking.com warns of 900 per cent rise in AI-related travel scams](https://www.bitdefender.com/en-us/blog/hotforsecurity/up-to-900-surge-in-travel-scams-warns-booking-com) — Bitdefender HotForSecurity - [How to avoid a travel scam this summer](https://consumer.ftc.gov/consumer-alerts/2026/06/how-avoid-travel-scam-summer) — Federal Trade Commission - [Consumer Sentinel Network Data Book 2024](https://www.ftc.gov/system/files/ftc_gov/pdf/csn-annual-data-book-2024.pdf) — Federal Trade Commission - [Fair Credit Billing Act](https://www.law.cornell.edu/wex/fair_credit_billing_act_%28fcba%29) — Cornell Law School Legal Information Institute - [Section 75 and chargeback protection](https://www.moneyhelper.org.uk/en/everyday-money/credit/how-youre-protected-when-you-pay-by-card) — MoneyHelper (UK government-backed guidance) - [Internet Crime Complaint Center (IC3)](https://www.ic3.gov/) — FBI --- # How to Research Your Family Tree with AI Source: https://aiforregulars.com/guides/ai-family-history-genealogy Tracing your family history used to mean spending months writing letters to county courthouses and squinting at microfilm. AI has changed what's possible — tools can now scan billions of records in seconds and flag likely matches for your ancestors. The research still takes your judgment and time, but the AI does the heavy lifting of finding the needle in the haystack. Here's how to use these tools well, starting with where most people begin. Before opening any app or website, write down everything you know about your family: names, approximate birth years, towns, countries of origin, and how people are related. Even rough information helps. Start with living relatives and go backward. Talk to the oldest family members you can — grandparents, great-aunts and uncles. Ask about: - Full names (including maiden names for women) - Where people were born and grew up - When they came to this country (if applicable) - Any family stories about ancestors This starting information is what you'll feed into genealogy tools. The more specific you are, the more accurate the AI matches will be. Go to **familysearch.org** and create a free account. FamilySearch is the best starting point because it's free and has an enormous collection — billions of records from over 170 countries. Start adding what you know: your name, your parents, your grandparents. FamilySearch will automatically start suggesting record matches — birth certificates, census records, immigration records, military records — based on the names and dates you enter. Each suggestion comes from a source document you can view. Always click through to the original document to verify. **Ancestry** (ancestry.com) has the largest commercial genealogy database and the most sophisticated AI hint system. When you add a person to your tree, Ancestry's AI automatically scans its records and other users' trees for potential matches and flags them as "hints" — a little leaf icon next to the person. Ancestry requires a subscription, but offers a free trial. When you get a hint: 1. Click the leaf to see the suggested match 2. Compare every detail against what you already know 3. If it matches on name, birth year, AND birth location, it's likely the right person 4. Click "Review" to accept or ignore the hint 5. Always look at the original source document, not just the summary Don't accept hints automatically — AI hint systems are good at surfacing possibilities, but a common surname can produce many false matches. Older records — 1800s census entries, ship manifests, church registers — are often handwritten in styles that are hard to read. ChatGPT is surprisingly good at transcribing these. Upload a photo or scan of the document directly into ChatGPT and use a prompt like: ``` This is a handwritten census record from approximately 1880. Please transcribe everything you can read, noting any words you're uncertain about with [?]. The document is in English. ``` For non-English documents: ``` This appears to be a German church baptism record from the 1850s. Please transcribe the text and translate it to English. Note any words that are unclear. ``` ChatGPT won't get everything right, especially on very faded or unusual handwriting, but it often gets 80–90% of a document that would otherwise take hours to puzzle out. Always check its transcription against the original. I uploaded a photo of my great-great-grandmother's immigration record from 1903 — a handwritten ship manifest with cramped cursive. ChatGPT transcribed it in about 20 seconds, including the village she came from in Poland, which I had never been able to read. That village name unlocked a whole new branch of the family. It's not magic, but it's close. This is the most important step, and the one most people skip. Genealogy errors compound quickly — if you accept a wrong match for your great-grandfather, every person you add above him in the tree is wrong too. Good verification habits: - **Look at the original document**, not just the transcription (transcriptions contain errors) - **Require at least two matching details** before accepting a record as your ancestor — name plus birth year is not enough; add location, parent names, or spouse names - **Be skeptical of other users' trees** — they spread errors freely. Use them as hints to find source documents, not as sources themselves - **Note your sources** as you go. Most genealogy software has a "source" field — use it When the evidence is uncertain, mark it as uncertain. Guesses feel like facts six months later when you've forgotten the context. If you've done a DNA test (through AncestryDNA, 23andMe, or MyHeritage DNA), **MyHeritage** (myheritage.com) is worth using alongside Ancestry. It has particularly strong records from Central and Eastern Europe, Scandinavia, and Latin America — regions where Ancestry's coverage is thinner. MyHeritage also offers Record Matching (similar to Ancestry's hints) and Smart Matches (matching your tree against other users' trees). The free tier lets you build a tree and see limited hints; more features require a subscription. Note that as of 2025, MyHeritage no longer accepts DNA uploads from other testing services — to get DNA matches on MyHeritage, you would need to order a MyHeritage DNA kit directly. ## Verification: The Golden Rule AI tools in genealogy are powerful at surfacing possibilities. They are not reliable at confirming them. Every match needs a human — you — to look at the original evidence and decide if it's right. The professional genealogical community has a standard called the Genealogical Proof Standard, which boils down to: search thoroughly, cite your sources, resolve conflicting evidence, and write a reasoned conclusion. AI makes the searching faster. The reasoning is still yours. ## What to try next Once you've found photos of your ancestors, you can bring them to life using [MyHeritage's animation and colorization tools](/guides/myheritage-ai-animate-photos). And if you have scanned documents or PDFs of family records, the guide on [uploading PDFs to ChatGPT](/guides/upload-pdf-ask-chatgpt) shows you how to extract information from them efficiently. ## Sources - [FamilySearch — free family history records](https://www.familysearch.org/en/) — FamilySearch - [Genealogy research](https://www.archives.gov/research/genealogy) — U.S. National Archives - [MyHeritage](https://www.myheritage.com/) — MyHeritage --- # AI Features in Zoom, Teams and Meet: What They Actually Do Source: https://aiforregulars.com/guides/ai-features-zoom-teams Meeting platforms have added AI features quickly over the past couple of years, and the marketing makes them sound nearly magical: summaries, transcripts, smart recaps, action items extracted automatically. The reality is more modest but still genuinely useful once you understand what these tools actually do well and where they fall short. This guide walks through the AI features in Zoom, Microsoft Teams, and Google Meet — how to turn them on, what they produce, and how much you can trust the output. When any of these platforms says it will "summarize your meeting," the underlying process is roughly the same: the platform transcribes the audio (speech to text), then passes that transcript to a large language model that produces a condensed version. This means the summary is only as good as the transcript. Background noise, multiple people talking at once, heavy accents, or unclear audio all degrade transcript quality, and a bad transcript produces a bad summary. In a well-run meeting with clear audio and one person speaking at a time, the results are impressive. In a noisy group discussion, they're less reliable. What AI summaries do well: capturing the main topics discussed, the general direction of decisions, and any explicit action items someone said out loud clearly. What they miss: the subtext, the decision that emerged from a long back-and-forth, or the commitment made half-jokingly that everyone understood was serious. Zoom AI Companion is available on most paid Zoom plans at no extra charge. To enable it for a meeting you're hosting: - Start or schedule your meeting in the Zoom desktop app. - Look for the AI Companion button in the meeting toolbar (it may be under "More"). - Select what you want: Meeting Summary, Smart Recording, or both. - Zoom will display a notice to all participants that AI features are active. After the meeting, Zoom emails the host a meeting summary with a topic overview, key discussion points, and a next steps section. The summary is also available in the Zoom web portal under your AI Companion section. Microsoft's AI meeting features are delivered through Copilot, which requires a Microsoft 365 Copilot license — this is separate from a standard Microsoft 365 subscription and has an additional per-user cost. If you're at a company, check with IT or your manager whether you have access. When Copilot is enabled in a Teams meeting: - You can ask Copilot questions about the meeting while it's still happening — for example, "What were the key points from the last ten minutes?" - After the meeting, a Recap tab appears with an AI-generated summary, transcript, and action items. - The summary integrates with other Microsoft tools — you can reference meeting notes from within Outlook or a Teams channel. The standout feature of Teams Copilot compared to the other platforms is the real-time Q&A: you can ask questions during the meeting without interrupting. This is useful for joining a meeting late and quickly catching up on what you missed. Google Meet's AI features are included in Google Workspace plans, with different tiers getting different features. Both live transcription and AI-generated meeting notes with Gemini are available on Business Standard and higher plans; neither is included in the entry-level Business Starter plan. To turn on transcription in Meet: - During a meeting, click the "Activities" icon (the three shapes icon) in the bottom-right corner. - Select "Transcripts" and click "Start Transcript." - A notification appears to all participants. - After the meeting, the transcript is saved automatically to your Google Drive in a Google Docs file. For meeting summaries powered by Gemini, look for the "Take notes for me" option in the same Activities panel. This generates a structured summary with discussion topics and action items that also saves to Drive. This step sounds obvious, but it's where most people run into problems with AI meeting features. The instinct is to forward the AI summary directly — to share the action items list with the team or send the recap to a client. Resist that instinct until you've read it. Common issues to check: - **Wrong attribution.** "Action item: John will send the report by Friday" sometimes attaches to the wrong person, or gets assigned to someone who just mentioned the report rather than committed to sending it. - **Missing decisions.** If a decision emerged from discussion rather than being stated explicitly, AI often misses it. - **Oversimplification.** A nuanced discussion can get summarized as one simple statement that doesn't capture the conditions or caveats. - **Invented specifics.** Rarely, the AI fills a gap with a plausible-sounding detail that wasn't said. Look for any item that you don't remember being discussed. A quick review — five to ten minutes right after the meeting — catches most of these. The time saved by AI transcription is far more than the review takes. I started using Zoom AI Companion for internal team meetings and found that the summaries are most useful not as the official record, but as a personal memory aid. When I need to remember what we decided three weeks ago, the AI summary usually gets me there in thirty seconds instead of scanning through notes. For anything that involves commitments to people outside the team, I write my own brief recap from the AI draft. Some meetings should not have AI transcription running at all. Before every meeting where you're considering enabling AI features, ask: would everyone in this meeting be comfortable knowing this conversation is being transcribed and summarized? Cases where you should think carefully or skip AI entirely: - HR conversations (performance reviews, terminations, complaints) - Legal discussions with attorney-client privilege implications - Medical or health-related conversations - Contract negotiations where you don't want a record of your opening position - Any meeting where you'd need explicit informed consent you haven't obtained Participants will see the transcription notice — but seeing a notice is not the same as understanding that a summary will be saved and could be shared. The professional standard is to ask, not just to notify. ## What to try next If you want a dedicated AI note-taker that works across all three platforms and adds more features, the [comparison of Otter, Fireflies, and Zoom AI Companion](/guides/ai-note-taker-comparison) covers the standalone tools in detail. For using AI to write up meeting follow-ups and action item emails, the [guide to AI prompts for office work](/guides/ai-prompts-office-work) has templates you can use immediately. --- # Best AI Fitness Apps: Which One Actually Builds a Plan That Sticks? Source: https://aiforregulars.com/guides/ai-fitness-apps There's no shortage of fitness apps. The AI ones stand out because they actually adjust: if you skip a workout, they shift the plan. If you're sore, they dial down the intensity. If you only have 20 minutes, they rebuild the session on the spot. But "AI-powered" has become marketing noise. Here's what the top options actually offer — and which one matches your situation. ## What to Look For Five factors that predict whether a fitness app will actually stick for you: | Factor | Why it matters | |---|---| | Beginner-friendliness | Apps that start too hard get abandoned in week two | | Equipment awareness | Does it know you only have resistance bands? | | Recovery tracking | Does it factor in rest days and muscle soreness? | | Medical sensibility | Does it flag exercises to avoid with common injuries? | | Explanation quality | Do you understand WHY you're doing each exercise? | ## The Top AI Fitness Apps ### 1. Fitbod — Best for Strength Training Fitbod builds personalized weight training plans based on your available equipment, fitness level, and which muscles have been worked recently. Its recovery tracking is genuinely good — it tracks muscle fatigue and avoids programming chest exercises the day after you just did bench press. **Beginner-friendliness:** High — you can set starting weights very low and the app adjusts. **Medical caution:** You can mark exercises as "disliked" or "excluded," which is useful if a movement bothers a joint. It's not a substitute for physical therapy advice. **Cost:** Free trial; subscription $15.99/month (or $95.99/year). ### 2. Future — Best If You Want Human Accountability Future pairs you with a real certified personal trainer who uses AI tools to design and adjust your program. You check in with your coach via text. This hybrid model — human accountability plus AI efficiency — has significantly higher retention than fully automated apps. **Beginner-friendliness:** Very high — the human coach can address questions no AI reliably can. **Medical caution:** Human coaches can account for injuries in ways AI cannot. **Cost:** ~$199/month — significantly more expensive than other apps. **Best for:** People who've tried apps and quit, who respond better to accountability. ### 3. Freeletics — Best for Bodyweight Training Freeletics uses AI to generate bodyweight workout plans that require zero equipment. It's particularly good for travel workouts or anyone who can't or won't go to a gym. Community features add a social motivation layer. **Beginner-friendliness:** Medium — workouts can be intense. Use the intensity slider and start low. **Medical caution:** Bodyweight movements done with poor form can still cause injury. Start slow. **Cost:** Free tier available; paid subscription for full AI coaching. ### 4. Whoop — Best for Recovery and Readiness Whoop is a wearable device paired with an AI app that tracks your sleep, recovery, and exertion level. It tells you how ready your body is for hard exercise on any given day. It doesn't generate workout plans but pairs well with another app. **Best for:** People who over-train or want data-driven recovery guidance. **Cost:** Subscription model that includes the hardware. ## Beginner-Friendliness Scores Ratings below are based on published user reviews, app-store feedback, and independent assessments — not in-house testing. | App | Starts easy? | Adjusts well? | Explains exercises? | Good for 50+? | |---|---|---|---|---| | Fitbod | ⭐⭐⭐⭐ | ⭐⭐⭐⭐⭐ | ⭐⭐⭐ | ⭐⭐⭐⭐ | | Future | ⭐⭐⭐⭐⭐ | ⭐⭐⭐⭐⭐ | ⭐⭐⭐⭐⭐ | ⭐⭐⭐⭐⭐ | | Freeletics | ⭐⭐⭐ | ⭐⭐⭐⭐ | ⭐⭐⭐⭐ | ⭐⭐⭐ | | Whoop | ⭐⭐⭐⭐ | ⭐⭐⭐⭐⭐ | ⭐⭐⭐ | ⭐⭐⭐⭐ | I've used Fitbod on and off for two years. The one thing I had to learn: the app's first suggested weight is often too ambitious for beginners. Always start 20–30% lighter than it suggests for the first week. The app adjusts up quickly once it has real data from you. ## A Medical Caution AI fitness apps are useful tools, but they are not medical professionals. If you have heart disease or high blood pressure, joint injuries, a chronic condition like diabetes or osteoporosis, or haven't exercised for several years, talk to your doctor before starting any new fitness program — AI-designed or otherwise. A good AI app will ask about injuries during setup. Take those questions seriously and answer honestly. "I want a hard program" is less useful information than "I have a bad left knee." ## What to Try Next If you're also interested in how AI can help with health questions beyond fitness, [AI and Health Symptoms: When to Use It and When to Call Your Doctor](/guides/ai-health-symptoms) covers that territory carefully. And if you're helping a parent or older family member get started, [AI for Seniors: A Beginner's Guide](/guides/ai-for-seniors) covers tools designed with that age group in mind. ## Sources - [American Heart Association recommendations for physical activity in adults](https://www.heart.org/en/healthy-living/fitness/fitness-basics/aha-recs-for-physical-activity-in-adults) — American Heart Association - [WHOOP](https://www.whoop.com/) — WHOOP - [Freeletics](https://www.freeletics.com/) — Freeletics - [Future](https://www.future.co/) — Future --- # AI Price Alerts: Never Overpay for Flights Again Source: https://aiforregulars.com/guides/ai-flight-price-alerts Flight prices swing by hundreds of dollars for the same seat on the same route. The difference between paying $300 and $600 is often just timing — and AI price tools have gotten very good at predicting those swings. Here's how to set up alerts so the right deal comes to you, instead of you hunting for it. You don't need a paid app for this. Three free tools cover most travelers: - **Google Flights** — the most accurate for major routes; connects to your Gmail so alerts land in your inbox automatically - **Hopper** — shows a simple "buy now" or "wait" recommendation based on its AI prediction model - **Kayak** — good for comparing prices across many airlines at once Start with Google Flights if you're not sure. It requires no app download and works in any browser. Go to [flights.google.com](https://flights.google.com) and enter your departure city, destination, and rough travel dates. You don't need exact dates — you can leave the return date flexible. On the results page, look at the **Price graph** tab near the top. This shows how prices change across different departure dates. Cheaper days show up in green; expensive ones in red. Once you've entered your route, look for the **Track prices** toggle near the top of the results. Click it. Google will ask you to sign in with your Google account. After sign-in, you'll automatically receive an email whenever the price changes significantly for that route — up or down. You can manage or delete alerts at any time from Google Flights' "Tracked flights" section. Download the Hopper app (free, iOS and Android). Search your route and tap **Watch this trip**. Hopper will track the price and send a push notification when it thinks the best time to buy has arrived. What makes Hopper useful: instead of just showing prices, it explicitly says "prices are likely to drop — wait" or "this is a good price — buy now." Keep expectations realistic. Hopper's predictions are good but not perfect, especially for holiday travel or unusual routes. Most people search for fixed dates and miss better deals one or two days away. In Google Flights, switch to the **Calendar view** (the grid icon near the date fields). This shows you price differences across the whole month at a glance. If your trip is flexible by even 2 days, this view often reveals a 20–40% price difference for the same route. When you get an alert saying prices have dropped, move quickly. Good flight deals on popular routes can disappear within hours. Open the alert email or app notification, verify the price is still showing, and book directly with the airline or through the tool. Don't sit on a deal "to think about it" — the algorithm that found it is running for everyone. I set alerts on Google Flights for every trip I'm even considering — not just confirmed ones. It takes 30 seconds and has surprised me with deals on routes I thought were always expensive. The worst that happens is I ignore the alert. The best is I save $200 without lifting a finger. ## What to Expect (Realistic) AI price alerts work best on: - Popular routes with lots of flights (major US cities, Europe hubs) - Trips planned 6–12 weeks out - Flexible travelers who can shift dates by a day or two They work less well on: - Very niche or small airports - Last-minute bookings (under 2 weeks out) - Holiday peaks where prices rarely drop ## What to Try Next Once you've got your flights sorted, use an AI travel planner to build the full itinerary. [Best AI Travel Planners Tested](/guides/best-ai-travel-planners) shows which tool builds the best day-by-day plan. And to keep your whole trip on budget, [ChatGPT as Your Monthly Budget Planner](/guides/chatgpt-monthly-budget) shows you how to track spending as you go. --- # AI for Seniors with Memory Challenges: Helpful Reminders or a Risk? Source: https://aiforregulars.com/guides/ai-for-seniors-memory Forgetting where the keys are is one thing. Forgetting whether you took your blood pressure medication is something else entirely. For seniors managing mild memory changes, AI reminder tools offer real practical help — a voice that prompts the 8 a.m. pill, a calendar that talks, an assistant that can be asked "what do I have today?" at any hour. But technology that compensates for a problem can also hide it. That's the tension this guide addresses. ## What AI Reminder Tools Can Actually Do Today's AI tools can handle a surprising range of memory-support tasks: **Medication reminders.** A smart speaker like Amazon Echo or Google Nest can announce reminders at set times. More specialized apps can confirm whether the person responded to the reminder and notify a caregiver if they didn't. **Appointment management.** A phone assistant (Siri, Google Assistant) can read out the day's calendar when asked. For someone who finds reading a calendar screen difficult, hearing "You have a doctor's appointment at 2 p.m." is much easier. **Daily task prompts.** Reminders to eat lunch, take a walk, water plants, or call a family member can reduce the anxiety that comes with feeling like you might be forgetting something. **Questions and answers.** A chatbot can answer "What day is it?" or "When did I last eat?" — questions that can feel embarrassing to ask a person but are comfortable to ask a machine. These tools work best for mild, everyday forgetfulness — the kind that comes with normal aging and does not indicate disease. For more significant memory challenges, they remain useful but need more careful setup and caregiver oversight. ## The Risk: When Reminders Mask Decline Here is the concern that doctors and family caregivers raise most often: if an AI is handling everything a person used to handle for themselves, how do you notice when things are getting worse? A missed medication used to be visible. Now the app handles it. A forgotten appointment caused a call from the doctor's office. Now the assistant manages the calendar. The person functions adequately — on paper — while the underlying memory problem progresses unobserved. This is not an argument against using AI tools. It is an argument for using them thoughtfully, with regular human check-ins that are not filtered through the technology. When I tested several reminder apps, I found they all made it very easy to set up prompts — and very easy to forget there is a real person on the other end of those prompts. The most useful thing any family member can do is stay curious about what the person actually remembers, not just what the app reminds them of. ## Signs That Memory Needs Medical Attention, Not an App AI reminder tools are appropriate for everyday forgetfulness in people who are cognitively well or experiencing mild age-related changes. They are not appropriate as a substitute for medical assessment when the following are present: - Forgetting recent events (not just where the keys are, but entire conversations from an hour ago) - Getting confused about time, place, or familiar people - Asking the same question repeatedly in a short span - Difficulty completing familiar tasks (cooking a meal they've made hundreds of times) - Personality or mood changes alongside the forgetfulness - Judgment changes — unusual financial decisions, poor safety awareness If any of these are present, the right next step is a conversation with the person's primary care doctor. Many cognitive changes are treatable or manageable when caught early. An AI app cannot assess this; only a clinician can. ## How to Set Up AI Reminders Wisely If you decide AI reminders are appropriate, these practices make the setup work better for everyone: **Keep it simple.** One smart speaker in a frequently used room is better than a complex system with multiple devices and apps. Complexity leads to confusion. **Involve the person.** Let them help set the reminders — time, wording, which things to track. Ownership increases the chance they engage with the prompts rather than dismiss them. **Don't automate judgment.** Reminders for medication and appointments are great. But don't use AI to replace calls from family or decisions the person can still make themselves. Preserve autonomy wherever possible. **Build in regular human check-ins.** Decide in advance that at least once a week, a family member will check in specifically about how the person feels their memory is — not just "how are you?" but "what have you been up to? Tell me about your week." This conversation reveals what the app cannot. **Review what the tools are compensating for.** Every few months, ask: are there tasks or functions the app is covering that the person used to do independently? Is the list growing? If yes, that's worth discussing with a doctor. ## Involving the Person's Doctor If you set up AI reminders for a parent or loved one, it is worth mentioning to their doctor. A physician can: - Do a baseline cognitive assessment so future changes are measurable - Advise whether the current tools are appropriate for the person's situation - Flag signs of progression that you might not recognize as significant - Refer to a geriatric specialist or neurologist if needed The doctor is the right person to judge whether technology is an appropriate support tool or a way of avoiding a harder conversation about care needs. ## What to try next For a broader look at AI tools that help seniors stay connected and engaged, [AI Companions for Seniors](/guides/ai-companions-seniors) is a good companion read. If you're also thinking about at-home safety tools, [AI Monitoring for Aging Parents](/guides/ai-home-monitoring-elderly) covers the privacy and consent questions you'll want to think through. --- # AI for Seniors: Everything a Beginner Over 65 Needs to Know Source: https://aiforregulars.com/guides/ai-for-seniors You've probably heard the phrase "artificial intelligence" more times than you can count lately. People talk about it on the news, grandchildren mention it at dinner, and ads for AI apps seem to pop up everywhere. This guide cuts through the noise and shows you five real things AI can help you with today — on any device you already own, at your own pace. ## 1. Voice Assistants: Ask Out Loud, No Typing Needed If you have an Amazon Echo, Google Home, or an iPhone, you already use AI. When you say "Alexa, what's the weather today?" or "Hey Siri, call my daughter," that's AI doing the work. You don't type anything. You just talk. Voice assistants are the gentlest starting point because there's nothing new to learn — you're having a conversation. They set kitchen timers, read you the news, play your favorite music, and answer quick questions like "what time does the pharmacy close?" or "how many cups are in a gallon?" Not sure which voice assistant to choose? Our comparison of [Alexa, Google Assistant, and Siri](/guides/alexa-google-siri-comparison) explains the differences in plain English so you can pick the one that fits your household best. The one thing to know: voice assistants work best for simple, single questions. For longer conversations or more complex help, a chat AI (the next section) does the job better. ## 2. AI Chat: A Patient Helper for Any Question An AI chatbot — ChatGPT is the most well-known one — is like having a very patient librarian available twenty-four hours a day. You type a question (or speak it out loud), and it writes back a plain, clear answer. What can you ask? Almost anything. "Explain my blood pressure reading to me in simple words." "Write a birthday card message for my nephew who loves fishing." "What are some gentle exercises for bad knees?" The AI doesn't get frustrated, doesn't rush you, and you can ask it to explain anything again in simpler terms. If typing is difficult, [ChatGPT has a voice mode](/guides/chatgpt-voice-mode) where you simply speak your question — no keyboard needed. Our tutorial [ChatGPT for Seniors: Your First Hour](/guides/chatgpt-for-seniors-first-hour) walks through setting it up step by step, including how to make the text as large as you need. One important rule: AI chat is a good starting point, not a final word on medical or legal matters. Treat it like a well-read friend, not a doctor. Always confirm serious health or financial questions with a real professional. ## 3. Photo Tools: Restore and Organize Old Pictures This one surprises a lot of people. AI can take an old, faded, scratched photograph and restore it — clearing up the image, sharpening blurry faces, and even adding color to black-and-white photos. Google Photos and Apple Photos do some of this automatically, and specialized apps go further. Our guide to [restoring old photos with AI](/guides/restore-old-photos-ai) covers both free and paid options and shows you exactly what results to expect. You can also use AI to help write down the stories behind those photos. Many older adults are using ChatGPT as a writing partner to create a [memoir or family history](/guides/ai-write-life-story-memoir) — telling their story one conversation at a time, no writing experience needed. ## 4. Scam Alerts: A Second Opinion on Suspicious Messages Phone scams targeting older adults are common, and they're getting more convincing. One of the most practical uses of AI right now is using it as a quick second opinion before you respond to anything that feels off. If you get a strange text message or email, copy the text and paste it into an AI chat. Ask: "Does this look like a scam?" A good AI will spot the red flags — pressure to act quickly, requests for gift cards, links to unfamiliar websites — and explain why something seems suspicious. Two scams that specifically target older adults have become much more sophisticated because of AI: - **Voice cloning scams:** Criminals can now clone a family member's voice using a short audio clip from social media. They call pretending to be your grandchild in trouble. Learn how this works in our guide on [AI voice cloning scams](/guides/ai-voice-cloning-scams) — and read [The Grandparent Scam 2.0](/guides/grandparent-scam-ai) for a family safety plan. - **Romance scams:** Fake "people" on dating sites are increasingly AI chatbots. Our guide on [AI romance scams](/guides/ai-romance-scam-dating-app) shows you the warning signs. ## 5. Reminders and Planning: Never Forget an Appointment Voice assistants make reminders simple. You can say "remind me every morning at eight to take my blood pressure pill" and it will do exactly that — no screen, no typing, no settings to dig through. AI chat can also help you prepare for doctor appointments. Before a visit, try typing: "Help me write down my questions for my cardiologist. I want to ask about chest tightness and about the new medication she mentioned." The AI helps you organize your thoughts into a clear list you can bring with you. Our full tutorial on [preparing for doctor visits with AI](/guides/ai-doctor-visit-prep) walks through this process step by step. If you or someone you care for is dealing with memory challenges, we have a separate guide on [AI tools for seniors with memory difficulties](/guides/ai-for-seniors-memory) — covering what genuinely helps, what the risks are, and when to involve a doctor. The first time I used an AI chatbot, I didn't know what to ask, so I typed a question I'd been meaning to look up for weeks: whether it was safe to take ibuprofen with my arthritis medication. The answer was clear, thorough, and it told me to confirm with my pharmacist — exactly the right advice. That felt like a safe and genuinely useful first step. ## Where to Start: Pick Just One Don't try all five at once. Pick whichever sounds most useful to you right now: - If you have a smart speaker → ask it a question today - If you want to chat → visit [chatgpt.com](https://chatgpt.com) in any web browser - If you have old photos you'd love to restore → open Google Photos and tap "Memories" - If you got a suspicious text → copy it into an AI chat and ask "is this a scam?" - If you forget appointments → set one voice reminder today and see how easy it is --- ## All Senior Guides on This Site Once you've taken your first step, the resources below go deeper on every topic. They're all written in plain English with no assumed technical knowledge. ### Starting Out with AI Chat - [ChatGPT for Seniors: Your First Hour, No Jargon](/guides/chatgpt-for-seniors-first-hour) — slow, step-by-step first session with everything from sign-up to voice mode - [ChatGPT Voice Mode: Talk Instead of Type](/guides/chatgpt-voice-mode) — speak your questions out loud; no keyboard needed - [ChatGPT Privacy Settings Explained](/guides/chatgpt-privacy-settings) — five minutes to set this up correctly before you use it regularly - [Is ChatGPT Free? Plans, Limits, and What You Get](/guides/is-chatgpt-free) — clear answer on what costs nothing and what the paid version adds ### Health and Medical - [Preparing for Your Doctor's Appointment with AI](/guides/ai-doctor-visit-prep) — build a symptom log and a list of questions before every visit - [AI Prescription Explainer: Understand Your Medication Label](/guides/ai-prescription-explainer) — paste in the label text and get a plain-English explanation - [AI Health Gadgets for Seniors: What's Worth Buying](/guides/ai-health-gadgets-seniors) — honest look at health-monitoring wearables and fall-detection devices - [AI Tools for Memory Challenges: Helpful or a Risk?](/guides/ai-for-seniors-memory) — what AI reminders and scheduling tools can and can't do for people with early memory difficulties - [Brain Training Apps: Do They Actually Work?](/guides/brain-training-ai-seniors) — honest review of Lumosity, BrainHQ, and similar apps ### Staying Independent at Home - [AI Companions for Seniors: Can They Help with Loneliness?](/guides/ai-companions-seniors) — voice-based AI companions explained, with realistic expectations - [AI Home Monitoring for Elderly: What Families Need to Know](/guides/ai-home-monitoring-elderly) — safety cameras, fall detection, and check-in systems - [AI Tools for Low Vision and Hearing Loss](/guides/ai-tools-low-vision) — accessibility features built into phones and standalone apps ### Safety and Scam Protection - [AI Voice Cloning Scams: How to Protect Your Family](/guides/ai-voice-cloning-scams) — understand the trick where criminals clone a family member's voice - [The Grandparent Scam 2.0: AI Edition — A Family Safety Plan](/guides/grandparent-scam-ai) — updated version of the most common scam targeting older adults - [AI Romance Scams: Fake Chatbots on Dating Sites](/guides/ai-romance-scam-dating-app) — how to tell when an online "person" is actually AI - [AI Scam Call Blockers: Do They Work?](/guides/ai-scam-call-blocker) — tools that filter suspicious calls before they reach you ### Life Stories and Family History - [Write Your Life Story or Memoir with AI Help](/guides/ai-write-life-story-memoir) — using ChatGPT as a patient writing partner, one conversation at a time - [AI for Family History and Genealogy](/guides/ai-family-history-genealogy) — finding records, writing family stories, and organizing old documents - [Restore Old Photos with AI: Free and Paid Options](/guides/restore-old-photos-ai) — bring faded or damaged family photos back to life ### Retirement and Finances - [AI and Your Retirement Pension: What You Can and Can't Ask AI](/guides/ai-retirement-pension-guide) — how AI can help you understand your benefits — and what it can't decide for you ### Voice Assistants and Devices - [Alexa, Google Assistant, or Siri: Which Should You Choose?](/guides/alexa-google-siri-comparison) — plain comparison of the three main voice assistants - [Tablet Setup for AI Beginners: Step-by-Step](/guides/tablet-setup-ai-beginners) — setting up a tablet for the first time with AI features in mind --- # How Students Can Use AI Without Cheating: 12 Legit Uses Source: https://aiforregulars.com/guides/ai-for-studying Most schools now have some kind of AI policy — and most students have already used AI at least once for school. The gap between using it well and using it wrong is smaller than it looks, but it matters. The 12 uses below are the ones that genuinely help you learn, that most schools allow, and that build skills rather than replacing them. At the end, there is a plain-English version of where the line actually is. ## The 12 Legit Uses **1. Explain a concept you did not understand in class** Paste the definition from your textbook and ask: "Explain this to me like I'm 14 and never heard of it before." AI is endlessly patient and will try different explanations until one makes sense. No judgment for not knowing. **2. Quiz yourself before a test** ``` I'm studying [topic] for a test on [subject]. Ask me 10 multiple-choice questions, one at a time. Wait for my answer before giving the next question. At the end, tell me which ones I got wrong and why. ``` This is one of the most effective study methods — retrieval practice — and AI can generate it for any subject instantly. **3. Get feedback on your own draft** Write your essay or paragraph first. Then: ``` Here is a paragraph I wrote for an assignment. Tell me: Is the argument clear? Are there sentences that are hard to follow? What would make it stronger? Do not rewrite it for me — give me specific feedback only. [paste your paragraph] ``` This is using AI the same way you would use a writing tutor. **4. Break down a hard math problem step by step** Show your own work first. Then ask the AI to show its work so you can see where yours went wrong. If you skip your own work and just copy the AI's answer, you will not understand the method on the test. **5. Summarize a long reading after you have read it** Use AI summaries to check your understanding *after* you have read, not as a substitute for reading. Ask: "What are the main arguments in this passage? Did I miss anything?" Then compare to your own notes. **6. Create a study schedule** ``` I have a [subject] exam in [X] days. I need to cover these topics: [list]. I have [X hours] per day available. Build me a study plan that covers everything without cramming it all into the last night. ``` **7. Translate academic language into plain words** "What does 'photosynthesis is an endergonic process' actually mean in simple terms?" AI is excellent at turning textbook language into sentences a person would actually say. **8. Practice writing in a language you are learning** Write a paragraph in Spanish, French, Mandarin, or whichever language you are studying. Ask the AI to correct your grammar and explain each mistake. This is active practice, not shortcutting — the thinking is yours. **9. Prepare for a class debate or discussion** "Give me 5 arguments for and 5 arguments against [topic] so I can prepare for a class discussion." You are still doing the thinking in the debate — you are arriving prepared rather than empty-handed. **10. Understand what feedback on a paper actually means** Teachers write comments like "underdeveloped argument" or "cite your evidence." If you do not know what to do with that, ask: "My teacher said my argument is underdeveloped. What does that usually mean, and what would fixing it look like?" That turns vague feedback into a concrete next step. I used to think getting help from a tutor was somehow less legitimate than figuring it out alone. That is not how learning works. AI is the most patient tutor ever built — the only question is whether you use it to think more or to think less. **11. Build subject vocabulary before a unit** "Give me the 15 most important vocabulary words for a unit on the Civil War and explain each one in a sentence." Quiz yourself on these before class — it makes everything the teacher says make more sense. **12. Check your own logic, not your wording** ``` Here is my argument: [paste your paragraph or outline]. Is the logic sound? Are there any steps I'm missing or assumptions I'm making without realizing it? Don't rewrite anything — just point out gaps. ``` This is different from asking AI to write your argument. You are asking it to push back on your own reasoning so you can strengthen it yourself. ## Where the Line Is Most school policies use a version of this rule: **the thinking and the writing must be yours.** - Using AI to understand, to practice, or to get feedback on your work = fine. - Using AI to produce the work you submit as your own = not fine. The grey area is editing. If you ask AI to "improve" your essay and it rewrites whole sections, and you submit that — you have crossed the line. If you ask for feedback on your argument and you rewrite it yourself based on that feedback — that is legitimate. When in doubt, ask your teacher. That question itself shows integrity, and it protects you if there is ever a dispute later. ## What to Try Next If your parents want to understand the homework side of this, [the parent's guide to ChatGPT and homework](/guides/kids-chatgpt-homework-parents-guide) covers the same territory from their perspective. And if you want to get sharper at writing prompts that actually help you study rather than just answer questions, [the guide to writing good prompts](/guides/how-to-write-prompts) covers the core habits quickly. --- # Can AI Help Me File My Taxes? Source: https://aiforregulars.com/guides/ai-for-taxes Tax season brings a pile of confusing forms, unfamiliar acronyms, and decisions that feel risky to get wrong. AI can take some of that confusion away — if you know what to ask it for, and what to never ask it for. This guide is written primarily for US taxpayers, but the general advice about how to use AI applies anywhere. Tax rules themselves are country-specific, so always verify against your country's official tax authority. ## What AI Is Actually Good At for Taxes ### Explaining Forms and Terms Tax documents are full of jargon. AI is genuinely useful for decoding it. You can paste a line from a form (without personal details) and ask what it means. For example: ``` What does "Box 12, Code DD" mean on a W-2 form? ``` You'll get a plain-English explanation. The same approach works for terms like "adjusted gross income," "standard deduction," "1099-NEC," or "Schedule C." ### Walking You Through What Documents to Gather Ask ChatGPT something like: ``` I'm a W-2 employee who also did some freelance work last year. What documents should I gather before filing my taxes? ``` It will give you a useful checklist: W-2 from your employer, 1099-NEC from freelance clients, receipts for deductible expenses, and so on. ### Suggesting Deductions to Research AI can generate a general list of deductions that might apply to your situation. This is not official tax advice — it's a list of things worth researching or asking your accountant about. ``` I work from home as a freelancer. What deductions might I qualify for? ``` Common ones it might mention: home office deduction, business equipment, internet service, and health insurance premiums for self-employed people. Treat this as a starting checklist, not a confirmed list for your actual return. I used ChatGPT to understand the difference between a 1099-NEC and a 1099-MISC before meeting with my accountant. Going in knowing what those forms meant let me ask better questions and not waste the appointment on basics. ## What AI Cannot Do for Your Taxes This part is just as important as the list above. **AI cannot file your return.** It has no connection to the IRS, HMRC, or any tax authority. Filing still has to happen through official tax software, a tax professional, or the tax authority's own website. **AI doesn't know your full situation.** A correct answer to a tax question depends on your total income, filing status, state, prior-year carryovers, and many other details. AI gives general information, not personalized advice. **AI's knowledge may be out of date.** Tax laws change every year. A model trained even a year ago may not know about new rules, limits, or credits. Always check current guidance at IRS.gov for US taxes. **Never share real personal data.** Your Social Security number, income figures, and bank details should never be typed into a public AI tool. If you want AI to look at something specific, rewrite the example with fake numbers. ## When to Use Tax Software Instead For straightforward returns — one W-2, standard deduction, no major complications — dedicated tax software (TurboTax, H&R Block, FreeTaxUSA, IRS Free File) is the right tool. These programs walk you through each section, apply current rules, catch common mistakes, and submit electronically. Use AI alongside tax software to understand what it's asking, not instead of it. ## When to Hire a Professional Some situations genuinely need a human expert: - You're self-employed with significant business income and expenses - You own rental property - You had a major life change: marriage, divorce, inheritance, selling a home - Your return spans multiple states - You're dealing with an IRS notice or audit A CPA or enrolled agent costs money, but in complex situations, the deductions they find or the mistakes they prevent often make the fee worthwhile. A free initial consultation with a local legal aid society or tax clinic can help you decide whether paid help is necessary. ## What to Try Next For understanding other confusing documents — lease agreements, phone bills, insurance letters — [AI as a plain-English contract translator](/guides/ai-explain-contracts) has safe prompts to use. And if you want to understand what personal data AI already has on you, [what does AI know about me](/guides/what-does-ai-know-about-me) is worth a read before sharing anything with an AI tool. --- # What Happens to Your Data When You Use AI — GDPR Explained Simply Source: https://aiforregulars.com/guides/ai-gdpr-your-data Every time you type a question or a problem into an AI chatbot, those words leave your device and travel to a company's server. That part is unavoidable — the AI lives there, not on your phone or laptop. But what happens to those words after they arrive? That is where it gets important, and where your rights as a European resident come in. Data privacy is the top concern many Europeans have about AI tools. That concern is well founded — and this guide will help you understand what is actually going on, in plain English. ## What "Your Prompts May Train the Model" Actually Means When you type something into a consumer AI chatbot and press send, your message goes to the company's servers. The AI generates a reply, which comes back to you. So far, so normal — this is how almost every web service works. The extra step is this: many consumer AI services, by default, keep a record of that conversation and may use it to improve their model. "Improving the model" means the company's engineers review patterns in conversations (sometimes with human reviewers) and use those patterns to make the AI smarter over time. In practice, this means words you typed — your questions, your worries, your personal situation — could become part of what the AI learns from. Not in a way that lets someone search for your exact message, but your words do contribute to the model's future behaviour. This is not a secret. It is in the terms of service. Most people just never read them. ## The Difference One Setting Makes Almost every major AI service now has a privacy or data settings page. In that page, there is usually an option along the lines of: "Use my conversations to improve the model." It is often **on** by default for free accounts. Turning it off takes about thirty seconds, and it means your new conversations will not be used for training. What it does NOT do: - It does not delete conversations already stored on the server. - It does not prevent the company from keeping the conversation for safety monitoring or legal reasons. - It may not apply to conversations you had before you changed the setting. To actually remove stored data, you need to use your GDPR rights — explained below. ## Your Three Core GDPR Rights, in Plain English GDPR stands for General Data Protection Regulation. It is EU law that gives you real, enforceable control over your personal data. Here are the three rights most relevant to AI tools. ### 1. The Right to Access You can ask any company: "What personal data do you hold about me?" This is Article 15 of GDPR. They must tell you within one month, and they must give you a copy in a usable format; for complex or numerous requests they can extend that by a further two months, but only if they tell you within the original month and explain why. This is called a Subject Access Request (SAR). For an AI tool, this might include: your account details, your conversation history, and any inferences the company has drawn from your use. ### 2. The Right to Erasure (the "Right to Be Forgotten") Under Article 17, you can ask a company to delete the personal data it holds about you, with the same one-month-plus-two-months timeline as above. This right is not absolute: Article 17(3) lets them refuse in limited circumstances — for example, if they are legally required to keep the data, or need it to establish or defend a legal claim — but they must explain why. For AI tools, this typically means: deleting your account, your stored conversations, and any training data derived from your conversations (though this last part can be technically complicated, and companies may not be able to remove your influence from an already-trained model). ### 3. The Right to Object Under Article 21, you can tell a company: "Stop using my data for this purpose." For AI training, this means objecting to your conversations being used to improve the model. The company must stop, unless they can show a compelling reason to continue. In practice, many companies make it easier to use the privacy settings toggle than to go through a formal objection process — but both routes are available to you. ### What If a Company Doesn't Respond? If a company misses the deadline on a request to access, delete, or object, or you think their response falls short, you can complain to a supervisory authority. In the UK, that's the ICO (Information Commissioner's Office); in the EU, it's the national data protection authority for the country the company is established in, or where you live. The authority can investigate the complaint and order the company to fix the problem — but it does not settle disputes over compensation between you and the company, or pay damages on your behalf. If you believe you suffered actual harm and want compensation, that has to go through the courts. ## Concrete Steps You Can Take Today Here is a simple checklist. You do not need to do all of these — even one or two make a real difference. **Step 1 — Find the privacy settings.** In ChatGPT, go to Settings → Data Controls. In other tools, look for Settings → Privacy or Help → Privacy Policy → Your Rights. **Step 2 — Turn off training.** Toggle off any option that says "improve the model," "use conversations for training," or similar. Save. **Step 3 — Delete your conversation history.** Most AI tools let you delete individual chats or all history. Deleting them from your view also removes them from the company's training pipeline going forward. **Step 4 — Send a deletion request if you want a clean slate.** Use the company's privacy request form (usually in the Help Centre or Privacy page). State clearly: "I am exercising my right to erasure under GDPR. Please delete all personal data held about me, including conversation history." **Step 5 — Use a more private mode for sensitive topics.** Some tools offer an "incognito" or "temporary chat" mode that does not save the conversation at all. Use this when discussing anything you would not want stored. > "I would like to ask a question about my personal situation but I want to keep it private. Is there a way to use you without saving this conversation?" You can literally ask the AI this — it will tell you what modes or settings are available. When I am discussing anything sensitive — a health worry, a financial question, a family situation — I use temporary chat mode and keep details vague. I describe my situation in general terms rather than typing names, exact amounts, or ID numbers. It is a small habit that costs nothing and keeps the risk low. ## What to Watch Out For **Never paste medical, financial, or ID details.** Even with training turned off, your text is stored on the company's servers during the session. A full name plus a medical condition plus a date of birth is exactly the kind of combination that makes personal data sensitive. Use general descriptions instead. **Screenshots and voice recordings are data too.** If you share a photo or use a voice assistant, that image or audio goes to the server. Be just as careful with visual and audio input as you are with typed text. **"Free" tools and your data.** When a service costs nothing, the company has to pay its bills somehow. For many AI tools, one answer is using what users share to make the product better — which they can then sell to businesses. This is not automatically bad, but it is worth being aware of. **Phishing and fake "data deletion" forms.** You will occasionally see emails or messages claiming to help you delete your AI data. Always go directly to the official website — never click a link in an unexpected email. Legitimate GDPR requests go through the company's own privacy page. **Training off ≠ data gone.** Turning off training is an important step, but it is not the same as asking for your data to be deleted. If you want a full clean slate, you need to submit a formal erasure request. ## What to Try Next Once you have sorted your privacy settings, the next step is understanding exactly what data these tools already hold about you — and how to see it. Our guide [What Does AI Know About Me?](/guides/what-does-ai-know-about-me) walks through how to request a full data export from major AI services. If you use ChatGPT specifically, the step-by-step walkthrough in [ChatGPT Privacy Settings](/guides/chatgpt-privacy-settings) shows you every toggle in plain language. And if you want to go further — opting out of training across multiple services at once — [How to Opt Out of AI Training and Request Deletion of Your Data](/guides/opt-out-ai-training-delete-data) covers the full process. ## Sources - [Regulation (EU) 2016/679 (GDPR) — full text](https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX%3A32016R0679) — EUR-Lex, Art. 12 (deadlines), Art. 15 (access), Art. 17 (erasure), and Art. 21 (objection) - [Art. 17 GDPR – Right to erasure ('right to be forgotten')](https://gdpr-info.eu/art-17-gdpr/) — GDPR-info.eu (readable version with commentary on the Art. 17(3) exceptions) - [Making a complaint](https://ico.org.uk/make-a-complaint/) — UK Information Commissioner's Office - [Right of access](https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/individual-rights/individual-rights/right-of-access/) — UK Information Commissioner's Office --- # AI-Generated Fake News: How to Tell Real Stories from AI Content Farms Source: https://aiforregulars.com/guides/ai-generated-fake-news AI content farms have become one of the fastest-growing sources of misinformation online. These are websites — sometimes hundreds of them run by the same anonymous operator — that use AI to churn out article after article with little or no human involvement. The goal isn't to inform you. It's to get clicks, run ads, push a political view, or all three. The good news: once you know the patterns, spotting these sites takes only a minute or two. ## Five Warning Signs of an AI Content Farm ### 1. The Site Has No Real People Behind It Legitimate news outlets — even small local ones — have named journalists, an "About" page with real information, and a way to reach an editor. Content farms usually have none of this. Look for: articles with no byline, or bylines that link to author pages showing no photo, no bio, and dozens of articles published in a single day. If someone supposedly wrote 40 stories this week, they almost certainly didn't. ### 2. The Site Name Sounds Like News But Isn't Content farms often use names that feel familiar — something like *American Daily Report* or *National News Tribune*. They pick generic names on purpose to create a sense of authority without earning it. A quick check: search the exact site name plus the word "fake" or "misinformation." If others have reported it before, you'll find that quickly. ### 3. The Story Only Appears on Unknown Sites Real news breaks in one or two places and gets picked up by larger outlets quickly. Fake news often appears on dozens of small sites at once — sometimes with nearly identical wording — but never gets picked up by a recognizable outlet. If you search a headline and only find it on sites you've never heard of, treat that as a red flag. ### 4. The Writing Feels Slightly Off — But Trust This Less Than You Used To Older AI-generated text had reliable tells: - Sentences that are technically correct but oddly vague - Quotes that sound like summaries rather than real speech - No specific dates, place names, or named eyewitnesses - Paragraphs that could be shuffled without changing the meaning These still show up, and three or four together are worth a closer look. But by 2026, the more capable models write with specific-sounding dates, named sources, and realistic quotes just as easily as vague ones — including invented specifics that never happened. That means writing style alone is a weaker signal than it was even a year or two ago. Treat it as one input among several, not as your main test — the source-check method below (find who published it first, check whether a recognized outlet corroborates it) works regardless of how good the writing is, because it doesn't depend on spotting a stylistic tell. ### 5. The Emotional Volume Is Turned Up High Words like *shocking*, *explosive*, *they don't want you to know*, or *the truth finally revealed* are designed to bypass careful thinking. Solid journalism uses measured language even for serious topics. I've started a personal habit: if a headline makes me feel instant outrage or fear, I wait 30 seconds before clicking or sharing. That pause is usually enough time to ask myself whether I've actually verified anything. The emotional jolt is often the whole point of the headline. ## A Four-Step Source Check Anyone Can Do When you see a story that feels off, run through this before sharing it. **Step 1: Find who published it first.** The story on your feed may be a copy of a copy of a copy. Search the key claim in quotes and look for the earliest version of it. **Step 2: Evaluate the outlet.** Go to the site's About page. Look for named editors and a physical address. Search the domain on [Media Bias/Fact Check](https://mediabiasfactcheck.com) — they maintain a database of unreliable sites. **Step 3: Look for corroboration.** A real story will appear on at least one recognized outlet — a major newspaper, the Associated Press, Reuters, or a public broadcaster like BBC or NPR. If only obscure sites carry it, be skeptical. **Step 4: Run a fact-check search.** [Snopes](https://snopes.com), [PolitiFact](https://politifact.com), and [FactCheck.org](https://factcheck.org) cover many viral stories. A quick search takes 30 seconds. ## Why AI Makes This Harder Now Before AI writing tools existed, producing a convincing fake article took real effort. Now anyone can generate a hundred convincing articles overnight. This means: - The volume of fake content is much higher than it used to be - The writing quality is higher too, making gut-feel detection less reliable - One person can run dozens of "news" sites simultaneously - The same false story appears in many slightly different versions, making it look corroborated when it isn't This isn't a reason to distrust every article you read. It's a reason to spend 60 seconds checking before you share. ## If You Find a Fake Story Spreading - Don't share it — even to argue against it. Sharing boosts its reach. - Report it on the platform you found it. Most have a "false information" option. - If a friend shared it, a private message tends to land better than a public correction. ## What to Try Next If you want a step-by-step system for checking any headline on the spot, [How to Fact-Check a Headline with AI in 60 Seconds](/guides/ai-fake-news-fact-check) walks you through a quick workflow using AI search tools. And if you're also seeing suspicious videos or photos, [How to Spot a Deepfake Video](/guides/how-to-spot-deepfake-video) covers the same practical, calm approach. ## Sources - [Media Bias/Fact Check](https://mediabiasfactcheck.com) — Media Bias/Fact Check - [FactCheck.org](https://www.factcheck.org/) — Annenberg Public Policy Center, University of Pennsylvania - [Snopes](https://snopes.com) — Snopes Media Group - [PolitiFact](https://politifact.com) — Poynter Institute --- # How to Use AI to Understand Government Forms and Benefits (Social Security, Medicare, DMV) Source: https://aiforregulars.com/guides/ai-government-forms-help Government paperwork is written by lawyers for other lawyers. A Social Security determination letter can run three pages without ever quite saying what happens next. A Medicare Annual Notice of Change lists premiums and coverage changes in dense paragraphs of fine print. If you have ever put one of these in a drawer because reading it felt impossible — this guide is for you. AI tools like ChatGPT and Gemini can now decode that language in seconds. You paste the text, ask a plain question, and get a plain answer. This guide walks you through exactly how to do that safely — including where you can trust AI and where you need to verify with an official source before acting. ## What AI Can (and Can't) Do With Government Documents **AI is genuinely useful for:** - Explaining what a letter or form is saying in plain English - Summarizing the key points and any deadlines mentioned - Creating a checklist of what you need to gather or do - Preparing questions to bring to a phone call or in-person appointment - Explaining the difference between Medicare Part A, B, and D in the context of your specific plan letter - Walking you through what an appeal process looks like **AI is not reliable for:** - Your exact benefit amount (AI can cite typical amounts but may be out of date) - Whether you personally qualify for a specific program - Filing on your behalf or confirming that a deadline has passed safely - Replacing legal or official advice for complex situations like appeals or disability determinations With that in mind, here is exactly how to use it. You do not need to scan or photograph your letter. If you have a paper copy, just type out the paragraph or section you don't understand. If you have a digital PDF, copy the relevant text from it. **Focus on the specific part you're confused about.** You don't need to paste a 12-page Medicare plan summary in full. Find the section you don't understand — the paragraph about your new premium amount, the part that says you need to respond by a certain date, or the benefit determination — and paste just that portion. If you have a PDF on your computer, you can also upload it directly to ChatGPT (with a paid account) or Google Gemini, which can read PDF files without any copying. See [How to Upload a PDF and Ask ChatGPT Questions About It](/guides/upload-pdf-ask-chatgpt) for the step-by-step on that. Go to chat.openai.com (ChatGPT) or gemini.google.com (Gemini). Both are free to use for this kind of task. You do not need a paid subscription to get a plain-English explanation of a government letter. **Before you paste:** remove your Social Security number, date of birth, and full name from the text. AI does not need that information to explain what the document means — the legal requirements and procedures are the same regardless of who you are. Removing those details reduces your privacy exposure without affecting the quality of the explanation. Paste the letter text into the chat box and continue to the next step. Type a question like this: > "This is text from a government letter I received. Can you explain what it's saying in plain English? What is it telling me, and what do I need to do?" Or, for a specific form: > "I have form SSA-44. Can you explain what this form is for and what it's asking me to fill in for each section?" The AI will give you a summary. If any part is still confusing, ask a follow-up: "What does 'IRMAA' mean?" or "Can you explain what a Medicare Part D late enrollment penalty is?" You can ask as many follow-up questions as you need — there is no limit. Follow up with: > "Based on this letter, can you give me a checklist of everything I need to do? Include any deadlines mentioned and what documents I might need to gather." This is where AI earns its keep. It will pull the action items out of the bureaucratic language and list them clearly. You end up with a short, readable to-do list instead of a three-page letter you have to re-read four times to find the one important sentence buried in paragraph seven. After you understand the letter, ask: > "What are the most important questions I should ask when I call Social Security about this?" or > "What should I bring to a DMV appointment to handle this situation?" AI can help you go into any phone call or office visit prepared. This saves time and reduces the chance of being sent home because you forgot one document — which anyone who has dealt with the DMV knows is a real possibility. This step is not optional. AI can misstate benefit amounts, eligibility thresholds, and program rules — especially for something as frequently updated as Medicare premiums or Social Security cost-of-living adjustments. **Before you do anything that matters** — file a form, miss a deadline, change your Medicare plan, or appeal a decision — check the official source: - Social Security: **ssa.gov** or **1-800-772-1213** (Mon–Fri, 8 a.m.–7 p.m. Eastern) - Medicare: **medicare.gov** or **1-800-MEDICARE** (1-800-633-4227, 24/7) - Veterans benefits: **va.gov** or **1-800-698-2411** - Utility assistance (LIHEAP): **liheap.acf.hhs.gov** - State-specific DMV and benefits: search for "[your state] DMV" or "[your state] benefits portal" AI is the translator. Those official sites are the source of truth. If you receive a letter about an IRMAA surcharge — an extra amount added to your Medicare Part B or Part D premium because Social Security says your income was too high two years ago — and your income has since dropped (due to retirement, job loss, divorce, or death of a spouse), you can appeal it using form SSA-44. I've found that pasting the IRMAA notice into ChatGPT and asking "what are the qualifying life events for an IRMAA appeal?" gets you a clear list in under a minute. It does not replace talking to SSA, but you'll walk into that call knowing exactly what to ask for. ## Common Situations Where This Works Well **Social Security letters.** Award letters, overpayment notices, and Supplemental Security Income (SSI) determination letters all tend to use dense legal language that doesn't say plainly what happened or what you need to do. Paste the relevant section and ask what it means and whether you need to respond. **Medicare Annual Notice of Change (ANOC).** Every fall, Medicare plans mail this document listing what's changing for the next year. Most people don't read it. Paste the premium and coverage change section into AI and ask: "What is changing in my plan next year, and is there anything here I should be concerned about?" **DMV forms.** Registration renewals, title transfers, name changes after marriage or divorce — DMV forms vary by state but are consistently confusing. Paste the form instructions and ask what you'll need to bring and in what order to complete the fields. **Veteran benefits.** VA benefit letters, pension award letters, and disability rating notices are notoriously dense. The approach is the same: paste the text, ask for a plain summary, ask what to do next. **Utility assistance programs.** Programs like LIHEAP (Low Income Home Energy Assistance Program) or state-run water assistance often have confusing eligibility paperwork. AI can explain the criteria and help you understand whether you likely qualify — though you'll need to apply through the official program. ## What to Watch Out For **AI hallucinates benefit rules.** This is the most important caution. SSA and Medicare rules change every year. An AI might cite a 2023 income threshold that has since been adjusted, or describe an appeal process that has changed. The Social Security Administration has moved toward automated decision-making — called "auto-adjudication" — where computers approve or deny claims without human review. That makes accurate self-filing more important than it used to be, which means AI's errors matter more too. Always verify numbers. **SSA's own AI phone system has documented problems.** The SSA launched an AI-powered phone assistant in 2025. Reporters documented cases where the system failed to understand callers after multiple attempts. If you call SSA and the automated system is not working for you, say "representative" or press 0 to request a human agent. **Do not paste your full Social Security number.** A chat interface is not a secure channel for your SSN, bank account numbers, or Medicare ID. Redact those before pasting anything. **Deadlines are serious.** AI may mention that a letter has a response deadline, but it may not catch every deadline, and it cannot file anything on your behalf. Missing a Medicare open enrollment window or an appeal deadline can have real, hard-to-reverse consequences. **This approach works outside the US too.** If you're in another country, the same method applies to your equivalent paperwork — pension notices, health insurance correspondence, tax agency letters. Paste the text, ask for a plain-language explanation, then verify with the official source. AI tends to know US government systems better than others, so verification matters even more outside the US. ## What to Try Next If your letter involves pension income or retirement timing decisions, [How AI Can Help You Understand Your Pension and Retirement Options](/guides/ai-retirement-pension-guide) goes deeper on that specific situation. For contracts, leases, or insurance policies where the language is just as dense, [How to Use AI to Understand Any Contract or Legal Document](/guides/ai-explain-contracts) covers the same paste-and-ask approach for those. --- # Can AI Help You Through Grief? What These Apps Can and Cannot Do Source: https://aiforregulars.com/guides/ai-grief-support Grief doesn't follow a schedule. It shows up at 3 in the morning when you can't sleep, on a Tuesday afternoon when a familiar song comes on the radio, at a dinner table with one fewer person at it. The people you want most are sometimes the ones who are gone — and the people who are still here don't always know what to say. Some people have started turning to AI in those moments, either to write out their feelings or to try something more unusual: a conversation with a digital version of the person they lost. This guide explains what both of those experiences actually are, what they might help with, and where the real risks lie. ## What a Regular Chatbot Can Help With (and What It Can't) Writing about grief — putting feelings into words — is one of the things that genuinely helps, and decades of research on expressive writing support that. A general-purpose chatbot like ChatGPT or Claude gives you a place to do that anytime, without worrying about burdening a friend or waking someone up at midnight. **What a chatbot does well during grief:** - Gives you a space to write out whatever you're feeling at 3am, without judgment - Asks gentle follow-up questions that help you say more than you might on your own - Responds to "I'm having a really hard day" with warmth, at any hour - Helps you draft a difficult message — to your children about what happened, or to a family member you've been avoiding **What it cannot do:** - Provide grief therapy or replace a trained counselor - Know you, your history, or the person you lost - Hold what you share in actual confidence (conversations may be visible to company staff or used for model training — check the app's privacy policy) - Be there for you the way another human being can A prompt that works well: > I'm grieving the loss of my [mother/father/partner] and I'm having trouble processing it. Can you just help me write out what I'm feeling right now? The chatbot will follow your lead rather than trying to fix things too fast. That said: if your grief is severe — if you haven't eaten or slept for days, if you're having thoughts of harming yourself, or if you feel unable to function — an AI chatbot is not the right starting point. In the US, please call or text **988** (Suicide and Crisis Lifeline). You don't need to be in immediate danger to reach out. One thing I've heard from people who find typing difficult when they're upset: use the voice feature in a chatbot app — ChatGPT has one — to speak your thoughts out loud, then read back what the AI transcribed and organized. Some people find it easier to speak grief than to write it, and hearing your own words played back can sometimes help you understand what you're actually feeling. ## What Are "Griefbots"? How Digital Replicas Work A griefbot is something different from a regular chatbot. Instead of a general-purpose AI, it's a system trained on data from a specific deceased person — their voice recordings, texts, emails, or social media posts — to simulate what it might be like to have a conversation with them after they're gone. Two of the better-known services work like this: **HereAfter AI** (hereafter.ai) is designed to be set up while someone is still alive. The person records voice conversations — stories, memories, answers to questions from family members — and the AI learns to respond in their voice. After the person dies, family members can ask questions and hear responses drawn from what was recorded. **You, Only Virtual** builds a text-based replica using a deceased person's messages and social media posts. Family members can submit the data after someone has died. Both create something that can sound or read like the person — responding to questions, sharing memories, using familiar phrases. But the underlying technology is pattern-matching: the system cannot know what the person didn't record, and it will sometimes say things they never would have said. Researchers who've studied griefbots have found roughly 70% accuracy in responses, meaning a meaningful portion of what the bot says contains errors, fabrications, or uncharacteristic phrases. ## Why Some People Find Comfort — and Why the Research Is Complicated A small qualitative study of 10 people who used digital replicas found that interacting with the replica offered a form of connection that human support sometimes couldn't fully provide. Specifically, people said it let them "talk" to the person they lost without worrying about burdening the living. That need — to say something to someone who is gone — is real and deep, and it doesn't disappear just because a person has died. Scientific American covered the emerging field under the headline "Can AI Griefbots Help Us Heal?" and found genuine ambivalence among both users and researchers: some found real comfort, others found the experience unsettling or ultimately unhelpful. But a 2026 study covered by TechXplore found that ongoing use of what researchers call "deadbots" can fuel pathological grief — a more prolonged and disabling form of mourning that doesn't naturally ease over time. Grief therapists warn that griefbots can extend the denial phase, making it harder to eventually reach acceptance. And because the bots achieve only partial accuracy, hearing a deceased parent say something out of character — because the bot got it wrong — can feel like a small but painful betrayal. The research base is still very thin. No large controlled studies exist yet. What we have is a handful of small qualitative reports, a growing number of clinical warnings, and a wide range of individual experiences. That doesn't mean these tools are necessarily harmful, but it does mean going in with your eyes open. ## Legal Rights and Consent: Who Controls a Digital Likeness? The law is catching up with this technology, though slowly. **Washington state** passed a law in June 2026 setting civil penalties of up to $3,000 per violation for creating unauthorized digital replicas of real people. The law applies to commercial uses of a person's likeness without their consent. **The NO FAKES Act**, a federal US bill, is pending in Congress and would create nationwide standards for digital replicas of voice and likeness. Outside the US, protections vary significantly. In most places there is currently no clear legal framework governing whether a family member can or cannot authorize a digital replica of a deceased person without that person's own prior consent. **Practically, this means:** - A reputable service will require documented consent — either from the person while they were alive, or from a next-of-kin with legal authority over the estate - Services operating in legal gray areas may not ask for consent at all - If a company approaches you unsolicited, offering to build a digital replica of someone you've lost, treat that as a red flag ## Staying Connected to Real Human Support AI works best as a supplement to human support, not a replacement for it. A few places to find genuine human connection during grief: **Hospice bereavement programs.** If your loved one received hospice care, the hospice is typically required to offer bereavement support to the family for up to 13 months after the death — at no cost. Many people don't know this. Call the hospice that was involved and ask what's available. Hospice News reported in April 2026 that AI grief tools are presenting new complexities in bereavement care — and that clear guidelines around consent and data privacy are still lacking in this space (Hospice News, April 2026). **Grief support groups.** In-person and online grief groups run by hospitals, hospices, and community centers are free or low-cost. Connecting with someone who has genuinely been through loss is a different experience from any AI interaction. **Grief therapists.** Psychology Today's therapist finder (psychologytoday.com/us/therapists) lets you filter by specialty, insurance accepted, and location. Many grief therapists offer sliding-scale fees. **Crisis support.** If you're in the US and struggling with thoughts of suicide or feeling overwhelmed, call or text **988** (Suicide and Crisis Lifeline). You don't have to be in immediate danger to call — if grief feels unmanageable, reaching out is the right move. ## If a Company Offers to Build an AI Clone of Your Loved One Whether the offer comes from a well-meaning family member or a commercial service, here are the questions to ask before you say yes: 1. **What data will be used?** Voice recordings, texts, emails, social media? Who collects and stores it, and how is it protected? 2. **Who owns the output?** Can you access the underlying data, export it, or request permanent deletion? 3. **What happens if the company closes?** Is there a plan for what happens to the digital replica? 4. **Did the person ever express a wish about this?** Anything written, recorded, or shared with someone you know? 5. **How will this affect others who loved them?** Children, surviving parents, and siblings may feel very differently about a digital recreation. There's no obligation to say yes — and no obligation to explain yourself if you say no. ## How to Document Your Own Wishes If you want to be proactive about what happens to your own digital likeness after you die, here's what you can do now: **Write it down.** A clearly written statement — "I do/do not consent to a digital replica being created from my voice, texts, or social media" — added to your advance directives or will carries real weight. Make sure the executor of your estate knows it's there. **Create your own version while you're alive.** Services like HereAfter AI are specifically designed for this: you record your voice, your stories, your answers to questions from family members, and the AI creates something your family can use after you're gone — with your full knowledge and consent. **Have the conversation.** Whatever you decide, tell someone who will outlive you. A written document helps, but a real conversation is even better. ## What to Try Next If you're thinking more broadly about AI companionship for yourself or an older family member, [AI Companions for Seniors: What They Can and Can't Replace](/guides/ai-companions-seniors) covers that ground in depth. And before you share anything personal with any AI tool — especially during a vulnerable time — [What You Should Never Tell an AI Chatbot](/guides/what-not-to-tell-ai-chatbot) is worth reading first. --- # AI Health Gadgets for Seniors: What Works, What's Overhyped Source: https://aiforregulars.com/guides/ai-health-gadgets-seniors Tech companies love to put "AI" on health products for seniors. That label sometimes means a genuinely smart system — and sometimes just means the device connects to an app. This guide cuts through the marketing and looks at which categories of AI health gadgets actually deliver for older adults, and which ones need more realistic expectations. All health emergencies require calling 911 or a medical professional — no gadget replaces that. ## How to read this guide The devices below are organized by category. Within each category, a table summarizes the key options, followed by an honest assessment of what works and what to watch out for. Setup difficulty is rated on a simple scale: **Easy** (works out of the box or nearly so), **Moderate** (needs a one-time setup with app), or **Requires help** (most seniors will need assistance to configure). ## Medication dispensers Missing doses — or doubling up — is one of the most common and serious health risks for older adults managing multiple prescriptions. Automated dispensers address this directly. | Device | What it does | Setup difficulty | Subscription needed? | |---|---|---|---| | Hero | Stores up to 10 medications, dispenses at set times, alerts caregivers | Moderate | Yes | | Livi | Locked canister, voice reminders, app alerts | Moderate | Yes | | MedMinder | Flashing lights, phone call reminders, caregiver portal | Requires help | Yes | | Simple pill organizer + phone alarm | Low-tech but effective for most people | Easy | No | **What works:** Smart dispensers are genuinely useful for people managing five or more medications with complex timing. The locking mechanism prevents double-dosing, and caregiver alerts mean a family member gets a notification if a dose is skipped. **What's overhyped:** The AI in most dispensers is fairly limited — it's mostly a timer with alerts. The real value is the dispenser mechanism and the caregiver connection, not anything meaningfully "intelligent." Also, many require a monthly fee that adds up fast. **Practical tip:** Before spending money on a smart dispenser, test whether a simple weekly pill organizer combined with phone alarms solves the problem. For many people, it does. ## Fall detection Falls are the leading cause of injury for people over 65. Devices that detect falls — and summon help automatically — address a real and serious need. | Device | How detection works | Setup difficulty | Requires phone? | |---|---|---|---| | Apple Watch (Series 4+) | Motion sensors detect hard falls, calls emergency contacts | Moderate | Yes (iPhone) | | Garmin Instinct / Vivosmart | Activity monitoring, automatic incident detection | Moderate | Paired phone for alerts | | Medical Guardian / Life Alert pendant | Wearable button, 24/7 monitoring center | Requires help | No | | Google Pixel Watch | Similar to Apple Watch, Google ecosystem | Moderate | Yes (Android) | **What works:** Wearable fall detection that doesn't require pressing a button is the key innovation. Apple Watch's fall detection has been credited with summoning help in real situations where the wearer was unconscious. That's significant. **What's overhyped:** No wearable catches every fall, and false positives (detected falls that aren't real) are common during exercise or vigorous movement. These devices reduce risk — they don't eliminate it. **Practical tip:** A traditional medical alert pendant with a 24/7 call center remains the most reliable option for immediate human response. Smartwatch fall detection is a good additional layer, not a replacement, especially for someone with high fall risk. My grandfather refused to wear a "medical alert" pendant because it felt stigmatizing. An Apple Watch was a completely different conversation — it's a watch. The fall detection is almost incidental. Sometimes the social acceptability of a device matters as much as its features. ## Health monitoring wearables Fitness trackers and smartwatches now measure heart rate, blood oxygen levels, irregular heart rhythm, and sometimes blood pressure. Here's an honest take. | Device | What it monitors | Setup difficulty | Validated medically? | |---|---|---|---| | Apple Watch (ECG feature) | Heart rhythm, fall detection, blood oxygen | Moderate | ECG cleared by FDA | | Withings ScanWatch | Heart rate, blood oxygen, ECG | Moderate | ECG cleared by FDA | | Omron blood pressure watch | Blood pressure, heart rate | Moderate | Yes (blood pressure feature) | | Fitbit Charge / Sense | Heart rate, blood oxygen, sleep | Easy to Moderate | Varies by feature | **What works:** Heart rhythm monitoring (ECG) on devices like Apple Watch is genuinely useful for detecting atrial fibrillation, a condition that often has no obvious symptoms but significantly raises stroke risk. This is a case where an AI-assisted feature can catch something a person wouldn't otherwise know about. **What's overhyped:** Blood oxygen readings on consumer wearables vary in accuracy, especially on darker skin tones. They can be directionally useful but shouldn't be used to make clinical decisions. **Practical tip:** If you're interested in heart rhythm monitoring for medical reasons, talk to your cardiologist first. They can tell you whether a consumer device is appropriate for your situation or whether a medical-grade holter monitor is more appropriate. ## Voice assistants and AI companions Smart speakers (Amazon Echo, Google Nest) can provide medication reminders, answer health questions, and connect to emergency services. **What works:** Simple reminders, hands-free timers, and quick answers to general health questions. For someone with limited mobility or vision, being able to ask questions out loud is a real quality-of-life improvement. **What's overhyped:** AI companions marketed specifically for seniors (there are several subscription services) make big claims about reducing loneliness and improving cognitive function. The evidence for those specific claims is limited. **Practical tip:** An Amazon Echo or Google Nest is inexpensive and easy to try. Start there before paying for a specialized companion service. ## What to try next For AI tools focused specifically on vision and accessibility, see [AI Tools for People with Low Vision](/guides/ai-tools-low-vision). If you're interested in keeping mentally sharp with AI, [Keep Your Brain Sharp with AI](/guides/brain-training-ai-seniors) covers learning tools that are designed with seniors in mind. --- # Denied by Health Insurance? How to Write a Winning Appeal with AI Source: https://aiforregulars.com/guides/ai-health-insurance-appeal Getting a denial letter from your health insurance company is a genuinely awful feeling — especially when the treatment your doctor ordered seems obviously necessary. Here is what most people don't know: insurance companies deny roughly one in three claims on the first submission, and more than 70% of those denials get overturned when someone takes the time to appeal. The problem is almost nobody does — fewer than 1% of denied ACA marketplace claims were appealed in 2024. AI tools can help you write a strong appeal letter in under an hour, even if you have never done it before. The denial letter is your roadmap. Insurance companies are required to tell you exactly why they denied the claim, and that reason is the foundation of your appeal. Look for: - The **specific reason for denial** — common ones include "not medically necessary," "experimental or investigational," "out of network," or "prior authorization not obtained." - The **claim or procedure code** (often a CPT or ICD code) that was denied. - The **appeals process instructions** — your insurer must include these by law. - The **deadline** to file your appeal. If the letter is hard to understand, take a photo of it and use one of the purpose-built tools in Step 4, or ask a general chatbot: > "I received a health insurance denial letter. Here is what it says: [paste the denial reason]. Can you explain in plain English what this means and what I should focus on in my appeal?" This step takes two minutes and is easy to skip under stress — don't. Appeal deadlines for most plans run from **30 to 180 days** from the date on the denial letter; 60 days is common for ACA marketplace plans, and Medicare Advantage plans often allow up to 60 days as well. Missing the deadline can forfeit your right to appeal entirely. Write the date somewhere prominent and set a phone reminder for a week before. If you are close to the deadline and don't have time to write a full letter yet, you can often file a brief notice of appeal to preserve your right, then follow up with full documentation. Call your insurer's member services line and ask about this option. A strong appeal letter doesn't just argue that the denial was wrong — it proves it, with documents. Collect as many of these as apply: - A **letter of medical necessity** from your doctor explaining why this specific treatment is needed for your condition. This is the single most important document. - Relevant **medical records** — diagnosis notes, test results, or a treatment history showing why alternatives didn't work. - The **specific policy language** describing coverage for the denied treatment. Ask your insurer for a copy of your Summary of Benefits and Coverage (SBC) if you don't have it. - A **clinical guideline or peer-reviewed study** showing that your treatment is standard of care for your diagnosis. Your doctor can usually point you to one. You don't need all of these for every appeal. The letter of medical necessity plus the denial reason is the starting point — add more evidence as you can gather it. You have two main options, and either can produce a solid first draft. **Purpose-built appeal tools (safer and easier):** - **Counterforce Health** (counterforcehealth.org) — free, designed specifically for fighting claim denials, with a claimed roughly 80% success rate. You describe your situation and it generates a structured appeal letter. - **Fight Health Insurance** (fighthealthinsurance.com) — upload a photo of your denial letter and it drafts an appeal. Also free. These are the better choice for most people because they're designed to handle sensitive health information appropriately and already know what an effective appeal letter needs to include. **General AI chatbots (ChatGPT, Claude, Gemini):** These work well if you're careful about what you share. Do not paste your full medical records into a general chatbot. Instead, share only what the appeal letter needs: the denial reason, your diagnosis, the procedure code, and the relevant policy language. The next step shows you exactly how. Whether you're using a purpose-built tool or a general chatbot, giving it a specific, structured prompt produces a much better first draft than asking it to "write an appeal letter." For a general chatbot, try this prompt: > "I need to write a health insurance appeal letter. My claim for [describe the treatment, e.g., 'an MRI of my right knee'] was denied with the reason: '[paste the denial reason exactly].' My diagnosis is [describe it briefly, e.g., 'chronic knee pain following an injury that has not responded to six weeks of physical therapy']. My doctor has provided a letter of medical necessity. Please draft a formal appeal letter that: (1) states what was denied and when, (2) explains why the treatment is medically necessary for my specific condition, (3) references clinical evidence or guidelines supporting this treatment, and (4) requests a written response within the plan's standard appeal timeframe. Keep the tone formal but clear — the reader is an insurance case reviewer." A good draft will have a clear statement of what was denied, your specific medical circumstances, a reference to clinical evidence, and a firm but polite request to overturn the denial. When I help someone draft an appeal, I always ask the AI for a second version at the end: "Now write a shorter version of the key argument in two paragraphs, in case I need to summarize it in a phone call." It's surprisingly useful to have a spoken version ready when the insurer calls to discuss the appeal — you don't want to be reading a long letter aloud. The AI draft is a starting point, not the finished product. Before you submit: **Check every fact.** Verify that the diagnosis name, procedure code, and any policy references match your actual documents. AI can produce confident-sounding text with plausible but incorrect details — a wrong CPT code can slow or derail your appeal. **Add your personal story.** One or two sentences about how the denied treatment affects your daily life, or what happens if you don't receive it, can make a real difference. Case reviewers are human; a specific personal detail carries weight alongside the clinical argument. **Have your doctor review it.** Especially the clinical evidence section — they may want to add a specific guideline reference or correct a medical term. Many doctors' offices have billing staff who handle insurance appeals routinely and can help. **Check for state-specific requirements.** Some states have additional rules about appeal language or required attachments. Your state insurance commissioner's website usually lists these. Submit through whatever channel your denial letter specifies — online portal, fax, or mail. Certified mail gives you documented proof of delivery, which matters if there is ever a dispute about the deadline. Keep copies of: - Your complete appeal letter with all attachments - Confirmation of submission (reference number, fax confirmation page, or mail tracking number) - The original denial letter Note the date you submitted and the expected response time. Insurers are typically required to respond to internal appeals within 30 days for standard requests and 72 hours for urgent ones. If your insurer upholds the denial after the internal appeal, you still have options — good ones. **External independent review:** Under federal law, most health plans in the US must offer external review by an independent organization. The external reviewer's decision is binding on the insurer. Request this in writing immediately after the internal denial. **State insurance commissioner:** File a complaint at your state insurance department. They can investigate whether the denial was appropriate and sometimes intervene directly with the insurer on your behalf. **Medicare Advantage:** The Medicare appeals process has multiple levels — internal plan appeal, reconsideration by a Qualified Independent Contractor, a hearing before an Administrative Law Judge, and beyond. The Medicare Rights Center (medicarerights.org) offers free counseling for navigating these steps. **Patient Advocate Foundation** (patientadvocate.org): Free case management for people dealing with serious denials. They know the system and can guide you through escalation at no cost. ## What to Watch Out For **AI may miss state-specific rules.** Appeal requirements vary by state — deadlines, required forms, and specific language differ. A general chatbot won't know your state's rules. Check your state insurance department's website for local requirements. **Don't share more than the appeal needs.** A general chatbot conversation is not a HIPAA-covered channel. Share the denial reason, diagnosis name, and procedure code — not your full medical history, Social Security number, or insurance member ID. **Review the AI draft before submitting.** AI can produce confident-sounding text with factual errors — wrong procedure codes, misquoted policy language, or invented clinical study names. Read every sentence before you sign your name to it. **Insurers are increasingly using AI to auto-deny.** Stateline reported in November 2025 that patients are now deploying AI tools to fight AI-driven denials — and it's working. Purpose-built tools are specifically designed to respond to algorithmic denials with the kind of structured, evidence-based language those systems send to human reviewers. **Urgent appeals move faster.** If treatment is time-sensitive, ask your doctor to flag the case as urgent when writing the letter of medical necessity. Urgent appeals must be resolved within 72 hours for most plans, compared to 30 days for standard ones. ## What to Try Next If the denial letter or policy document uses confusing legal language, [How to Use AI to Understand a Contract or Legal Document](/guides/ai-explain-contracts) shows you how to get plain-English explanations without a lawyer. Before your next appointment, [How to Prepare for a Doctor's Visit with AI](/guides/ai-doctor-visit-prep) can help you frame the medical necessity conversation with your doctor so the letter they write is as strong as possible. --- # Should You Ask AI About Your Health Symptoms? Source: https://aiforregulars.com/guides/ai-health-symptoms If you've ever Googled a symptom and ended up convinced you had something terrible, you already know that health searches can go sideways fast. AI chatbots like ChatGPT can feel like a step up — they give you a real answer instead of a list of links. But they come with their own risks, and understanding those risks is what separates a helpful tool from a dangerous shortcut. ## What AI can safely help with Think of a good AI chatbot as a well-read friend who happens to have read a lot of medical textbooks. That friend can explain what "edema" means, tell you the difference between a sprain and a strain, or help you put your symptoms into words before a doctor's visit. That's genuinely useful. Specifically, AI is solid for: - **Understanding terminology.** If your lab results mention "elevated creatinine" or your discharge notes say "idiopathic," AI can explain those terms in plain English. - **Organizing a symptom log.** You can describe what you've been feeling and ask AI to help you write it down clearly for your doctor. - **Generating questions to ask.** "What questions should I ask my doctor about this medication?" is a great prompt. - **Learning about a diagnosis you already have.** If your doctor just told you that you have Type 2 diabetes, AI can help you understand what that means in general terms. I tested this with a vague symptom — "occasional dizziness when I stand up" — and found that ChatGPT gave a helpful explanation of orthostatic hypotension while explicitly telling me to get it checked by a doctor. That's the right behavior. If an AI ever tells you it can diagnose you, that's a red flag. ## What AI cannot do — and where it gets dangerous Here is the hard line: **AI cannot diagnose you.** It cannot examine you. It doesn't have access to your medical history, your test results, or the ability to observe how you actually look and feel. It works by predicting likely text responses based on patterns in training data — which means it can miss things, get things wrong, or reassure you when it shouldn't. The risks are real: - AI might describe the most common cause of a symptom and miss the rare but serious one you actually have. - It might tell you something sounds benign when it needs urgent attention. - It might make you feel confident enough to skip a doctor visit you actually need. Never use AI to decide whether to take or stop taking medication. Never use it to evaluate a symptom in a child without getting a real clinical opinion. Never let a reassuring AI answer delay you from getting emergency care. ## When to call your doctor or 911 — no AI first Call 911 or go to an emergency room immediately if you experience: - Chest pain, pressure, or tightness - Sudden shortness of breath - Severe headache that came on suddenly ("worst headache of my life") - Weakness, numbness, or drooping on one side of your face or body - Confusion, trouble speaking, or difficulty understanding others - Coughing up blood or vomiting blood - A feeling that something is seriously wrong In any of these situations, do not stop to type into a chatbot. Call for help. ## Your health data and privacy This is the part most people skip, and it matters. When you describe your symptoms to a free AI chatbot, that conversation may be stored, reviewed for safety, or used to improve the model — depending on the service's privacy policy. Most consumer AI tools are **not HIPAA-compliant**, which means they do not have the same legal protections as your doctor's patient portal. Practical rules for protecting yourself: - **Don't include identifying details.** Describe your symptom without your name, date of birth, location, or insurance information. - **Use your doctor's patient portal instead** for anything sensitive. Many now include AI-assisted features that are covered by health privacy rules. - **Check the privacy settings.** Services like ChatGPT let you turn off training on your conversations. See our guide on [ChatGPT privacy settings](/guides/chatgpt-privacy-settings) for exact steps. - **On shared devices, clear the chat** after discussing anything personal. ## How to get the most out of an AI health conversation When you do use AI for health research, frame your questions carefully. Here are some prompts that work well: ``` Explain what [medical term] means in simple English. ``` ``` I've been having [symptom] for [how long]. Help me write a clear description to give my doctor. ``` ``` What are some questions I should ask my doctor if I've been told I might have [condition]? ``` ``` What does this phrase from my discharge notes mean: "[exact phrase]"? ``` After the AI responds, read it critically. Look for phrases like "this could be" or "in some cases" — these signal uncertainty, and real uncertainty warrants a real conversation with your doctor, not more AI queries. ## What to try next If you want to make the most of AI before your next appointment, read [Preparing for Your Doctor's Appointment with AI](/guides/ai-doctor-visit-prep) — it walks through building a symptom log and question list step by step. And if you're unsure how much to trust AI answers in general, [Can You Trust ChatGPT?](/guides/can-you-trust-chatgpt) explains how these systems work and where they fall short. --- # AI Hearing Aids Explained: What the 'AI' Label Actually Means Source: https://aiforregulars.com/guides/ai-hearing-aids-explained Walk through a Costco Hearing Center or browse hearing aids online and almost every device now claims to be "AI-powered." That label covers a real range — from a genuinely useful noise-filtering chip to little more than a companion app with a trendy name attached. This guide breaks down what the AI actually does, what it costs, and how to tell a real feature from a marketing sticker. ## What "AI" Actually Does Inside a Hearing Aid Older hearing aids amplified sound using fixed rules — turn up all sound above a certain pitch, for instance. That helps with volume but doesn't do much to separate a voice from restaurant clatter or wind noise. The AI in newer hearing aids is typically a small neural network — a chip trained on millions of recorded sound scenes — running directly on the device. Its job is to constantly classify what you're hearing: is this quiet, one person talking, a group conversation, or a noisy room? Based on that read, it adjusts amplification and noise reduction dozens of times per second, aiming to keep speech clear while turning down everything around it. The easiest way to tell whether a specific model's "AI" claim is substantive is to search the brand name plus "deep neural network" or "AI chip." Brands with a real engineering story behind the feature — like Jabra's Enhance Pro 30 or Phonak's Sphere — publish details about what the chip was trained on and how it performs in noise testing. If a product's site only shows a vague "AI-powered" badge with no explanation, that's worth a second look before you spend the money. ## OTC vs. Store-Fitted: Two Different Ways to Buy This is the part that confuses the most people, because both categories now use AI and both get sold at Costco. **Over-the-counter (OTC) hearing aids** became a legal category in the US after an FDA rule took effect in October 2022. It lets adults who believe they have mild to moderate hearing loss buy self-fitting hearing aids directly — no prescription, no medical exam, no audiologist visit required. You typically set them up yourself using a phone app that runs a basic hearing test and tunes the device to your results. **Store-fitted hearing aids** sold at places like Costco's Hearing Centers are a different, prescription-grade category. A licensed hearing instrument specialist gives you a free hearing test in-store and programs the device specifically for you, with follow-up visits included. You don't need a doctor's referral to book an appointment, but the fitting is done by a professional, not an app. Both categories can include AI-based noise reduction. The difference is who does the fitting — you, or a specialist — not necessarily the underlying chip. ## What It Actually Costs Self-fitting OTC models like the Lexie B2 Plus (built with Bose audio technology) sit at the lower end, generally around $999 per pair. You fit them yourself through an app, and they include rechargeable batteries and Bluetooth streaming from a phone. Brands like Eargo push further into discreet, nearly invisible designs — the Eargo 8, for example, sits deep in the ear canal and starts around $2,700 a pair. You're generally paying for size and comfort as much as additional AI processing at this tier. At Costco Hearing Centers, brands like Sennheiser's Sonite Rise and Rexton run about $1,600 per pair, while Jabra's Enhance Pro 30 — which Costco markets as its first model with a dedicated AI chip trained on millions of real-world sound scenarios — runs about $1,700. All three prices include the in-store hearing test, fitting, and follow-up visits, which is why this tier can come in below what the same technology level costs at an independent audiology clinic. ## Does Medicare Cover Any of This? Original Medicare does not cover hearing aids, AI-equipped or otherwise — that's been the rule for decades and hasn't changed as of mid-2026. Medicare Part B will cover a hearing exam if your doctor orders it to diagnose a specific medical condition, not for a routine hearing aid fitting. Since a 2023 rule change, it will also cover one audiologist visit every 12 months without a doctor's order, specifically for gradual, age-related hearing loss. Medicare Advantage plans are a different story: about 95% of them now offer some kind of hearing benefit, but the amount and rules vary enormously by plan and region. If you're on a Medicare Advantage plan, call and ask specifically what's covered before you buy — some plans cover a set dollar amount toward any hearing aid, others only cover specific brands through a network. ## How to Choose Without Getting Overwhelmed **Start with how confident you are about your hearing loss.** If you're not sure whether your hearing loss is mild, moderate, or more serious, get a real hearing test first — either free at a Costco Hearing Center or through your doctor. Self-fitting OTC devices are built for mild-to-moderate loss; guessing wrong means paying for a device that won't actually help. **Decide how much you want a human involved.** If you'd rather have a specialist test your hearing and program the device for you, look at store-fitted options. If you're comfortable running a phone-based hearing check and adjusting settings yourself, OTC self-fitting saves time and often money. **Check the return window before you buy.** Hearing aids are a personal-fit product — what sounds great in a quiet store can feel different in daily life. Most reputable sellers, including Costco, offer trial periods of 45 to 180 days. Confirm the specific policy before purchasing, since it varies by retailer and brand. **Don't assume a higher price means smarter AI.** Some of the price difference between models comes from processing quality, but a lot of it comes from battery life, size, and how discreet the device is. If invisibility and premium comfort don't matter to you, a mid-priced option with solid noise reduction can perform just as well for everyday conversations. ## What to Try Next If hearing loss runs alongside vision changes or other accessibility needs, [AI Tools for People with Low Vision](/guides/ai-tools-low-vision) covers the phone-based side of that picture. For situations where captions work better than amplification — video calls, meetings, noisy public spaces — see [AI Captions and Live Transcription for Deaf and Hard-of-Hearing People](/guides/ai-captions-hard-of-hearing). --- # How to Use AI When Buying or Renting a Home (Without Trusting It Blindly) Source: https://aiforregulars.com/guides/ai-home-buying Buying or renting a home is one of the largest financial decisions most people ever make. It also comes with a wall of unfamiliar words — APR, escrow, conveyancing, ground rent, fixed-rate, variable-rate, easement — and paperwork that can feel designed to confuse. AI tools like ChatGPT or Gemini will not replace your estate agent, your mortgage broker, or your solicitor. But they can act as a patient, always-available tutor who explains the jargon, helps you prepare the right questions, and reads confusing documents aloud in plain English. A NerdWallet study found that roughly half of prospective U.S. home buyers (48%) plan to use AI tools during their search, and about one in four (27%) to help estimate housing costs — though, as we will see, that last use case needs some care (Source: NerdWallet). Here is how to use AI well, and where to stop before it leads you astray. ## Decode the Mortgage Jargon Before You Talk to a Lender Walking into a meeting with a mortgage broker without knowing the basic terms puts you at an instant disadvantage. AI is excellent at closing that gap before you sit down. Try a prompt like this: > "Explain the difference between a fixed-rate and a variable-rate mortgage in plain English. What are the pros and cons of each for a first-time buyer who plans to stay in the property for at least ten years?" Or go term by term: > "What does APR mean on a mortgage offer, and why is it more useful than the headline interest rate when comparing lenders?" > "Explain what escrow is in a home purchase. Who holds the money, and why does it exist?" > "What is the difference between freehold and leasehold? What are the risks of buying a leasehold property?" The AI will give you clear, patient explanations you can read in your own time and re-read as many times as you need. Write down the terms that come up in your own words afterward — that act of paraphrasing locks them in far better than just reading. **One important limit:** do not ask AI to calculate your specific mortgage payments or tell you what you can afford. It does not know today's interest rates, your credit profile, or your local property tax. Use your lender's official calculator, or the government housing agency calculator in your country, for those numbers. ## Understand a Lease or Rental Agreement Before You Sign Long rental agreements are full of clauses that look standard but have real consequences — rules about subletting, break clauses, what counts as fair wear and tear, and who is responsible for which repairs. **Paste the actual text into the chat.** AI is very good at explaining what a specific clause means in plain English. > "Here is a clause from my lease agreement. What does it actually mean, and what should I watch out for? > [paste the clause here]" You can also ask it to scan the whole document: > "I'm going to paste my rental agreement. Please summarize the main things a tenant should know — especially anything unusual, any financial obligations beyond the monthly rent, and any clauses that give the landlord unusual rights." This will not replace legal advice. AI can miss jurisdiction-specific rules or make errors with complex language. But it gives you a solid first understanding so you know which questions to ask a solicitor — and you can ask a solicitor much more specific, targeted questions after you have done this step. ## Research Neighborhoods Qualitatively AI cannot tell you what a square metre costs in a specific street today. But it can synthesize a lot of general knowledge about an area — the character of different neighborhoods, proximity to transport and amenities, things people tend to like or dislike about living there. > "I'm considering moving to [neighborhood or area]. What are the main things people typically appreciate and dislike about living there? What types of residents tend to be drawn to it, and how has it changed over the past decade?" > "Compare [Neighborhood A] and [Neighborhood B] for a retired couple who want walkability, good public transport, and access to parks. Which tends to score better on those factors, and why?" Use this as qualitative background, not as data. For crime figures, current school ratings, flood risk, or real price trends, go to official local government sources or dedicated property portals. AI's general knowledge can be months or years out of date on specific statistics. ## Draft Smart Questions for a Viewing or Agent Meeting Walking into a property viewing without a list of questions is the fastest way to forget everything you wanted to ask the moment you step through the door. AI can build that list for you in two minutes. > "I'm viewing a 1960s semi-detached house this weekend. What are the most important questions I should ask the estate agent at the viewing? Focus on things that could be expensive to fix or that affect the property's long-term value." > "I'm meeting a mortgage broker for the first time. What questions should I ask to compare their offer properly against other lenders?" > "I'm renting a flat for the first time. What should I ask the landlord or letting agent before I sign anything?" Print the list or keep it on your phone. At a viewing, it is completely normal — and it signals to the agent that you are a serious, prepared buyer or tenant. Before viewing a flat recently, I gave ChatGPT the property description from the listing and asked it to flag anything that sounded vague or like it might be hiding a problem. It noticed the listing said "cosy" and "original features" without mentioning a boiler replacement or any recent upgrades — and suggested I ask specifically about the heating system age and any history of damp. Both turned out to be real issues worth knowing before I made a decision. ## Get a Plain-English Summary of Any Contract Clause Purchase contracts, transfer deeds, mortgage offers, and homeowners' association agreements can run to dozens of pages. Reading them cover to cover is exhausting, and the important parts are rarely labeled "IMPORTANT." > "Here is a clause from my home purchase contract. Explain what it means, what my obligations are under it, and whether there is anything I should ask my solicitor to clarify before I sign." > "What does this indemnity clause in my mortgage offer actually require me to do?" This is closely related to the skills covered in [How to Use AI to Understand Any Contract or Legal Document](/guides/ai-explain-contracts) — that guide goes deeper into the exact process for working through a contract step by step. The key discipline: treat AI's explanation as a starting point for your own understanding, not as legal confirmation that the document is fine. A licensed solicitor, notary, or real estate attorney is the right person to give you a final opinion before you sign anything significant. ## What to Watch Out For **AI cannot give you a reliable affordability number.** This is the most important limit to keep in mind. As MIT Sloan researchers have noted, AI chatbots carry no fiduciary duty — they are not legally responsible for the advice they give you (Source: MIT Sloan). A chatbot that confidently tells you "you can afford a $400,000 home on your income" is giving you a guess, not a professional assessment. Use official mortgage calculators or speak to a licensed broker. **AI price data is not live.** AI models have a training cutoff. They do not have access to today's listings or this month's market prices. For any specific price — what similar homes are selling for, what rents are running in a given area, what a fair offer might be — use a dedicated real estate portal or ask your agent. **Watch out for hallucinated legal and tax claims.** AI sometimes states specific legal rules, tax thresholds, or government schemes confidently and incorrectly. If an AI tells you something specific about stamp duty rates, first-buyer tax relief, eviction rules, or rental deposit law, treat it as a starting point to verify — not as a fact. Always check with an official government website or a qualified professional. **Fake AI-generated listings are a growing problem.** Some fraudulent property listings use AI-generated photos or descriptions. Images that look too perfect, addresses that do not match street-view reality, and landlords who refuse to meet in person or want unusual payment methods are all warning signs. If something feels off, it probably is. **Keep your sensitive details out of the chatbot.** You do not need to share your real income, credit score, bank balances, or national insurance number to get useful guidance. Keep your prompts general — "a household earning around X" or "a buyer with a small deposit" — and your real financial data stays private. ## What to Try Next If you have a contract or lease in front of you right now, [How to Use AI to Understand Any Contract or Legal Document](/guides/ai-explain-contracts) walks through the exact step-by-step process for using AI to decode any document before you sign. Once you are settled, [How AI Can Help You Lower Your Household Bills](/guides/ai-lower-bills) shows practical ways to use the same tools to manage your costs after you move in. --- # Storm Damaged Your Home or Car? How AI Can Help With the Insurance Claim Source: https://aiforregulars.com/guides/ai-home-insurance-claim-storm-damage If a storm has just torn shingles off your roof or dented your car with hail, the insurance claim is one more stressful task on a bad day. AI chatbots like ChatGPT won't file the claim for you, but they're genuinely good at the organizing and writing parts — turning a phone full of damage photos and scattered notes into a clear, complete claim description. This guide walks through where AI actually helps, in the right order, and where you still need to rely on your own eyes, your policy document, and a real contractor's estimate. Call your insurance company or agent as soon as you can, even before you've fully documented everything. Have your policy number ready if you can find it, and write down the claim number and your adjuster's contact information as soon as you're given one — you'll need both for every follow-up call. Filing promptly matters: many insurers process claims in the order they come in after a widespread storm, and most policies set a deadline (often around a year) for filing at all. There's no benefit to waiting until your documentation is "perfect." Before any cleanup or repair, photograph and video every area of damage — wide shots of the whole roof, siding, or car, plus close-ups of specific problems like broken shingles, a cracked window, or a dented panel. For damaged belongings inside the house, photograph the item, its model or serial number if visible, and a receipt if you have one. Don't throw out damaged items until your adjuster has seen them or explicitly told you it's fine to discard them. If something has to go immediately for safety reasons, photograph it from several angles first and keep a written list of what it was. This is the one step AI genuinely can't do for you — but once you have the photos, it can help with the next part. Once you have photos and rough notes, a chatbot can help structure them into the kind of clear list an adjuster or claims form actually wants. Try a prompt like: > "I have storm damage to my home. Here's what I've documented so far: [list your rooms/areas and what's damaged in each, in whatever order you have it]. Organize this into a clear inventory, grouped by room or area, with a short description of each item's damage. Leave blanks for purchase date and estimated replacement cost where I haven't given you that information." This turns a messy voice memo or scattered notes app into something you can hand to your adjuster or attach to your claim, without you having to build a spreadsheet by hand. Before the adjuster visits, you're generally expected to prevent further damage — tarping an exposed roof, boarding a broken window, that sort of thing — but not to make permanent repairs. Permanent repairs done before an inspection can complicate your claim, since there's no longer physical damage for the adjuster to assess directly. Keep every receipt for temporary repairs and materials; these costs are typically reimbursable. If you're unsure whether something counts as "emergency" versus "permanent," ask your claim representative before doing the work. Insurers and adjusters respond well to a clear, factual written account of what happened and what was damaged — not a long emotional narrative, but not a bare list either. This is where AI drafting genuinely saves time. Try: > "Help me write a clear, factual description of storm damage to my home for an insurance claim. Here's what happened: [describe the storm/date] and here's the damage: [paste your inventory from the previous step]. Write it in a professional but plain tone, organized by area of the home, without exaggerating or minimizing anything." Ask for a second, shorter version too — a few sentences you could read aloud if your claim representative calls before you've submitted anything in writing. Your insurer will typically want a written repair estimate with itemized pricing, not just a verbal quote. Get bids from licensed, established contractors — after a widespread storm, be especially wary of unfamiliar contractors going door-to-door, demanding a large payment upfront, or quoting a price far below everyone else's. Verify they're licensed and pulling proper permits before you sign anything. Don't authorize permanent repair work until your insurer has approved the scope and pricing, unless it's the emergency work from Step 4. AI doesn't have your actual policy in front of it unless you paste specific sections in, and even then it can misread coverage limits, deductibles, or depreciation rules while sounding completely confident. Don't ask a chatbot "will this be covered?" and treat the answer as final — ask your adjuster, or read the specific coverage section of your policy yourself. The same caution applies to car storm damage: comprehensive coverage (not collision or liability) is usually what pays for hail, falling branches, or flood damage to a vehicle, but whether you carry comprehensive at all depends on your specific policy. The single most useful AI prompt in this whole process, in my experience, is the plainest one: paste in a confusing paragraph from your policy or your denial letter and ask, "What does this mean in plain English, and what should I ask my adjuster about it?" It won't replace reading the policy yourself, but it turns fifteen minutes of squinting at insurance jargon into something you actually understand before you make the call. ## What to Watch Out For **AI can write a confident-sounding answer that's still wrong.** Never let a chatbot's guess about your coverage or payout replace your actual policy document or a conversation with your adjuster. **Storm-damage scams increase after major storms.** Unsolicited contractors offering "free inspections" who then pressure you to sign a repair contract on the spot are a well-known pattern after storms. A legitimate contractor will let you get a second opinion. **Don't over-share personal or policy details with a general chatbot.** You don't need to paste your full policy number, Social Security number, or account details for AI to help you write or organize text — describe the situation and share only the specific clause or detail you need explained. ## What to Try Next If your policy language or a denial letter is confusing on its own, [how to use AI to understand a contract or legal document](/guides/ai-explain-contracts) walks through getting a plain-English read on dense text. And if the repair work itself is the next headache, [using AI to plan a home repair](/guides/ai-home-repair-help) covers getting quotes and understanding what a job should actually involve. --- # AI Monitoring for Aging Parents: The Privacy Dilemma Families Face Source: https://aiforregulars.com/guides/ai-home-monitoring-elderly When an aging parent lives alone, the quiet fear never fully goes away — the thought of a fall, a forgotten gas burner, or a medical event with no one nearby. AI monitoring technology has made it possible to keep a gentle eye on a home from thousands of miles away. But "keeping an eye" is exactly the phrase that makes this complicated. Whose eye? With whose permission? What happens to the data? And who decides when safety concerns outweigh privacy? ## What AI Home Monitors Can Do Modern home monitoring systems for older adults go well beyond a simple security camera. Depending on the product and configuration, they can: - Detect a fall and automatically alert family or emergency services - Notice when daily routines change (usual morning coffee time passes with no kitchen activity) - Alert caregivers if a stove or appliance is left on - Track sleep patterns and flag unusual nights - Monitor door and room sensors to indicate general activity without showing video Some systems use cameras; others rely entirely on passive motion sensors that produce no images at all. The camera-free options are often an easier conversation for families where privacy is a top concern. ## The Conversation You Must Have First Here is the non-negotiable part: **install nothing without an honest conversation first.** This matters for practical reasons — your parent may refuse to stay in a monitored home, making the whole effort counterproductive. It matters for trust — if they discover a device you installed without telling them, it may damage your relationship in ways that outlast any safety benefit. And it matters for their dignity — an adult living in their own home has the right to know who is watching. The conversation is often harder to start than to have. Here are a few approaches that tend to work: **Frame it as mutual peace of mind.** "I worry when I can't reach you. Would you be open to something that lets me know you're okay without calling six times a day?" This positions the monitor as something that benefits both of you, and it's honest. **Start with the least intrusive option.** Passive motion sensors with no cameras are a very different proposition than a video doorbell with a two-way microphone. Starting at the low end of intrusiveness makes the first yes much easier. **Let them help choose the system.** If your parent picks the device, reviews what data it collects, and decides who gets alerts, they remain in control of their own home. That shift in framing — from "you are being monitored" to "you chose a safety tool" — changes everything. **Be specific about who sees what.** "Only I would get an alert, only if the motion sensor hasn't triggered for four hours" is very different from vague reassurances. Specifics build trust. I've seen families install monitoring systems with the best intentions and then rarely check the alerts, effectively creating surveillance with no benefit. Whatever system you set up, agree in advance on what you will actually do when you get an alert — and how often you will check in regardless of whether an alert fires. ## Ongoing Consent Is as Important as Initial Consent A one-time conversation is not enough. Your parent's feelings about monitoring may change as their health changes, as their cognitive status shifts, or simply as they spend more time living with the awareness of being watched. Check in every few months: "Are you still comfortable with the sensor setup? Is there anything you want to adjust?" If cognitive decline progresses to the point where they can no longer meaningfully consent, involve their doctor and, if applicable, a social worker or legal guardian in the decision about continued monitoring. ## Privacy and Data: Questions to Ask Before You Buy Before purchasing any monitoring system, get clear answers to these questions: **Where is the data stored?** On-device storage is more private than cloud storage. If data goes to a company's servers, find out how long they keep it and whether they share it with third parties. **Who has access?** Make sure you can limit alert recipients to only the people your parent agrees to. **Can the data be deleted?** You should be able to request deletion of stored data if your parent wants to stop using the service. **What happens to the data if the company is sold?** Privacy policies can change under new ownership. This is harder to protect against, but worth knowing. **Is video encrypted?** If the system includes cameras, end-to-end encryption matters. ## When Monitoring Is Not the Right Answer AI monitoring works best for relatively independent people who need a safety net, not for those who need active daily care. If your parent needs help with medication management, bathing, meals, or regular medical attention, monitoring cannot substitute for that. Seeing that someone hasn't moved in four hours is not the same as being there. If safety concerns have reached the level where you are considering constant monitoring, it may be time to have a broader conversation — with your parent, their doctor, and possibly a geriatric care manager — about the right level of support. In-home care aides, assisted living, or a medical alert service with a professional monitoring center may be more appropriate solutions. ## A Note on Dementia and Reduced Capacity Monitoring is particularly fraught when a parent has dementia or significantly reduced cognitive capacity. They may not be able to give meaningful consent, may become frightened or confused by the presence of devices, or may not remember agreeing to monitoring they would have rejected when healthy. In these situations, work closely with the person's medical team and, if they have one, a legal guardian or healthcare proxy. Document decisions carefully. Prioritize dignity and the person's known wishes where they can be determined. ## What to try next If you're thinking about broader AI tools for an aging parent, [AI Companions for Seniors: Can They Really Help with Loneliness?](/guides/ai-companions-seniors) covers the emotional-support side of the question. For a general introduction to AI tools that older adults find genuinely useful, [AI for Seniors](/guides/ai-for-seniors) is a good starting point. --- # Use AI to Diagnose Home Problems Before Calling a Plumber or Handyman Source: https://aiforregulars.com/guides/ai-home-repair-help The faucet has been dripping since Tuesday. There's a brown stain spreading across the bathroom ceiling. A window that opened fine last year has suddenly jammed shut. Before you call a plumber — and potentially spend $150 on a callout fee for what turns out to be a five-minute fix — a quick conversation with an AI can tell you what you're actually dealing with, whether it's something you can handle yourself, and exactly what parts to buy if you can. Over 40% of homeowners are already using AI tools to help with minor repairs and home maintenance, according to a 2025 industry survey. ChatGPT and Gemini can both accept photos and deliver surprisingly detailed diagnoses when you describe the problem clearly. ## What Kind of Problems Work Best AI is genuinely useful for visual problems — things that show up clearly in a photo. Good candidates include a dripping or slow faucet, cracked grout or peeling caulk around the tub or shower, a stuck window or door, a water stain on the ceiling or wall, and a flickering light or outlet that stopped working. AI is less reliable for problems hidden behind walls, underground, or that require smell or sound to diagnose accurately. It can still help you figure out the right questions to ask a professional — but it works best when the problem is something you can actually photograph. ## Step by Step: Getting a Useful Diagnosis Get down to eye level with the problem and take two or three photos from different angles. Good lighting matters more than a great camera. If the problem is a stain, photograph the edges as well as the center. If it's a fixture or fitting, photograph any visible labels or model numbers while you're there — you may need them later when ordering parts. Both are free for basic use and accept photo uploads on their mobile apps and websites. ChatGPT uses a camera icon in the message bar; Gemini has a similar option. There are also dedicated home-repair apps on iOS designed around photo-based troubleshooting — but ChatGPT and Gemini will cover most problems without downloading anything new. Don't just upload the photo and ask "what's wrong?" The more context you give, the better the answer. A good prompt looks like this: > I have a slow drip from the hot water tap in my kitchen sink. It started about a week ago and has gotten slightly worse each day. I live in a house built in the 1990s and haven't tried to fix it yet. The faucet is a single-handle style. The photo shows water dripping from the base of the faucet, not the spout. What is likely causing this and can I fix it myself? Cover four things: what you're seeing, when it started, what you've already tried, and any relevant details about your home's age or setup. The AI will usually give you a ranked list of likely causes — for a dripping faucet, that might be a worn cartridge, a deteriorated O-ring, or a failing valve seat — with a short explanation of each. Read the full response before you do anything. If one cause seems more likely given your description, ask a follow-up: > Given that the drip is at the base and not the spout, does that make the O-ring more likely than the cartridge? The AI is good at narrowing things down when you push it with specific details. Once you have a likely diagnosis, ask: > If it's the O-ring, give me step-by-step repair instructions for a beginner, and a list of the parts and tools I'll need. A useful AI response will include the tools required (adjustable wrench, screwdriver, replacement O-ring), the steps in order, and any safety notes — like shutting off the water supply valve first. It will also tell you if the task is beyond what a beginner should try alone. If you're not sure exactly what to buy, ask: > What should I ask for at the hardware store? Is there a standard part, or do I need to bring the old one in to match it? For plumbing parts especially, the answer is often "bring the old part in" — there are dozens of variations that look almost identical. The AI will tell you this, which saves you a wasted trip back. If the repair turns out to be beyond your skills, use AI to make sure you're not overpaying. Ask: > What's a reasonable price range for a plumber to replace a kitchen faucet cartridge in a mid-size US city? This gives you a ballpark. If a quote comes back much higher, you know to ask why or get a second opinion. Knowing the going rate before you call takes about 60 seconds and can save you real money. I take a quick photo of the parts and their arrangement before I put anything back together — the inside of the pipe, the pieces laid out in the order I removed them, any labels or model numbers. If something goes wrong afterward, that photo is invaluable when I'm explaining the situation to a plumber or trying to reorder the right part online. ## The Hard-Stop List: Call a Pro, Full Stop AI can help you think through a surprising number of home problems. For these, skip the AI and call a licensed professional — not because the repair is complicated, but because getting it wrong has serious consequences. **Gas lines.** Any repair involving a gas pipe, gas valve, or gas appliance connection must be done by a licensed plumber or gas fitter. A slow leak is not always detectable by smell until it reaches dangerous levels. This includes connecting a new gas stove or water heater. **The electrical panel.** Replacing a light switch or an outlet is generally within a careful homeowner's ability. Working inside the main electrical panel — adding a circuit, replacing a breaker, running new wiring — is not. Mistakes here can start fires or cause electrocution, and the work almost always requires a permit and inspection. **Structural elements.** Load-bearing walls, foundations, roof structure, and anything that holds the house up or keeps water out are not DIY territory. Removing what looks like a non-structural wall without confirming it first can cause serious damage. **Anything requiring a permit.** In most areas, certain repairs — new plumbing runs, electrical work, room additions — require a permit and a final inspection. Skipping this creates problems when you sell the house and may void your homeowner's insurance for that area. Ask the AI whether your repair typically requires a permit; if it says yes or "it depends on local code," contact your local building department first. ## What to Watch Out For **AI does not know your local building code.** Requirements vary significantly by city and state. The AI gives you a general answer; local rules may be different. **AI can misread a photo.** A ceiling stain can look like a roof leak when the actual cause is a pipe dripping inside the wall. Use the diagnosis as a starting point, not a conclusion. **Don't let AI push you past your comfort level.** If the steps feel too complicated or you're not confident you can do it safely, that feeling is worth respecting. A repair that goes wrong can cost significantly more to fix than the original problem. **Photos you upload may be used to improve AI services.** Depending on your account settings, images you send to ChatGPT or Gemini may be stored or reviewed. Make sure your photos don't include visible house numbers, security keypads, or other identifying details if that concerns you. ## What to Try Next If a contractor sends you a quote or repair estimate, [How to Use AI to Understand Contracts and Legal Documents](/guides/ai-explain-contracts) shows you how to get a plain-English breakdown of any document before you sign. For more ways to put AI to work around the house, [50 Things You Can Ask ChatGPT That Actually Help](/guides/things-to-ask-chatgpt) has a dedicated home-and-maintenance section worth bookmarking. --- # AI Home-Safety and Fall-Detection Devices for Seniors: A Plain-English Buyer's Guide Source: https://aiforregulars.com/guides/ai-home-safety-devices An AARP survey found that 71 percent of adults aged 50 and older are interested in using AI to help monitor home and personal safety — one of the highest-rated future uses in the survey (AARP Research). The good news is that the devices that make this possible have improved a great deal. They are also easier to buy and use than most people expect. This guide walks you through the main categories, explains what each type actually does, and helps you decide which one fits your situation. ## Smart Doorbells and Home Cameras A smart doorbell replaces or sits alongside your regular doorbell. When someone walks up to your front door, the built-in camera sends a live picture to your phone or tablet. You can see who is there — and even talk to them through a speaker — without opening the door. The AI part: modern doorbells use AI to tell the difference between a person, a package, a car, and an animal. You get an alert only when a real person arrives, not every time a delivery van passes by. **What it collects:** Video footage of your front entrance, usually for a short period — anywhere from a few hours to 30 days depending on your plan. If you pay for cloud storage, that footage sits on the company's servers. Without a paid plan, footage may only be available live on your phone. **Good for:** Knowing who is at the door before you open it. Also useful for caregivers who want confirmation that a regular visitor — a home carer, a delivery person — has arrived at the right time. Products in this category include Ring Doorbells and Google Nest Doorbell. Prices and subscription costs change; check the current pricing on each manufacturer's website. Indoor cameras work the same way but face inside the home. They can alert a family member if an older person hasn't moved around by a certain time of morning. The ethics of indoor cameras deserve a proper conversation before you set one up — our guide on [AI home monitoring for elderly](/guides/ai-home-monitoring-elderly) covers exactly that. ## Wearable Fall Detection: Smartwatches If you or a family member already wears a smartwatch, some models include built-in fall detection. The Apple Watch and certain Samsung Galaxy Watch models are the most well-known examples. When the watch senses a hard fall and you don't respond to a prompt within about a minute, it automatically calls emergency services and sends a message to your emergency contacts. **What it collects:** Movement and location data, continuously. This data goes to the watch manufacturer's servers. You can control some privacy settings in the companion app on your phone. **What it does not do:** It is not infallible. Falls near furniture you grab onto, or certain types of exercise, can confuse the sensor. And if you set the watch down on a table, it cannot detect anything. **Good for:** Active older adults who already want a smartwatch for messaging, fitness tracking, or health monitoring and want fall detection as an added safety layer. ## Wearable Fall Detection: Medical-Alert Pendants A medical-alert pendant is a small device you wear around your neck or on your wrist. Unlike a smartwatch, it has one main job: keep you safe. When it detects a fall — using a motion sensor — it triggers an alarm and connects you to a monitoring center or calls a family member directly. Many modern pendants include: - **Automatic fall detection** — the sensor triggers without you pressing anything - **A manual SOS button** — press it anytime you need help, even if you didn't fall - **Two-way voice** — you can speak to the monitoring center through the pendant itself - **GPS location** — so help can find you if you're outdoors **Cellular backup:** Some pendants work over your home Wi-Fi; better ones include their own cellular connection. This matters — Wi-Fi can go down in a power cut, but a device with cellular backup keeps working. Products in this category include Life Alert, Lifeline (formerly Philips Lifeline), and Bay Alarm Medical. Monthly monitoring fees apply to most of these services; confirm current pricing directly with the provider. > "Does this device work if the power goes out? Does it need Wi-Fi?" Ask that question before you buy. It is one of the most important. ## Passive In-Home Monitoring Apps Passive monitoring means tracking someone's routine without them wearing anything. A set of small sensors placed around the home — on the refrigerator, bathroom door, front door — detects whether normal daily patterns are happening. If Mum hasn't opened the fridge by 10 a.m. when she normally does by 8 a.m., a family member gets a gentle alert on their phone. Some systems use cameras; others use motion sensors that don't record any video at all, making them much more privacy-friendly. **What it collects:** Motion patterns and door-open events. Camera-based versions collect video. All of this goes to an app the family member checks on their phone. **Good for:** Families checking in from a distance on an older relative who doesn't want to wear any device. The person being monitored should always know a system is in place — see "What to Watch Out For" below. Products in this category include dedicated sensor kits from various home-security providers. Subscription costs vary by provider. (Amazon discontinued its own Alexa Together caregiving service in 2025.) My own grandmother refused to wear a pendant for years — she felt it looked too medical and didn't want to think of herself that way. What finally worked was a passive sensor: a small disc on the fridge door. Her daughter got a quiet app notification if the morning pattern seemed off. No wearable, no camera — just a routine check. She agreed to that easily. ## The Five Key Buying Questions Before you buy anything, get clear answers to these five questions: **1. Does it need a monthly subscription?** Many devices require an ongoing monitoring fee on top of the upfront cost. Ask what happens if you stop paying — do the core alerts still work, or does the device become useless? **2. Does it work without Wi-Fi or if the power goes out?** A device with cellular backup keeps working when your internet fails. GPS tracking also works outdoors where there is no home Wi-Fi. **3. Who sees the footage or data?** Read the privacy policy, or ask a family member to read it. Find out whether footage is stored on company servers, for how long, and whether it can be shared with third parties. **4. Can the person wearing it turn it off?** The wearer should always be able to silence alerts or pause monitoring. Devices that don't allow this raise serious consent questions. **5. Is the device waterproof?** Bathrooms are where many falls happen. Look for a water-resistance rating — IP67 or higher — if you want to use it in the shower. ## Start Here: Which Type Fits Your Situation? **Lives alone, no immediate family nearby:** Start with a medical-alert pendant that has cellular backup and automatic fall detection. Add a smart doorbell so you can see visitors safely before opening the door. **Early memory challenges:** A passive sensor system works well here because there is nothing to remember to wear or charge. Pair it with a voice assistant (Alexa, Google Home) that can give verbal reminders throughout the day. **Caregiver at a distance:** A passive sensor kit or indoor camera (discuss it first and get agreement) gives you peace of mind without requiring your relative to do anything differently. A smart doorbell tells you when the home carer has arrived. **Active and independent, wants a safety net:** A smartwatch with fall detection is the least intrusive option. It doubles as a fitness tracker and doesn't feel like a medical device. ## What to Watch Out For **Privacy and consent.** The person being monitored should always know what is being recorded and who can see it. Installing a camera or sensor without telling them — even with good intentions — can seriously damage trust. Our guide [AI Home Monitoring for Elderly: What Families Need to Know](/guides/ai-home-monitoring-elderly) covers this in detail. **False alarms.** Motion sensors and fall detectors are not perfect. Too many false alarms and people start ignoring them — which defeats the purpose entirely. Choose a device with adjustable sensitivity and test it before you rely on it. **Over-reliance.** These devices are an added layer of safety, not a replacement for human contact. A sensor that reports "all normal" does not replace a phone call or a visit. Keep both. **Subscription lock-in.** Some providers make it expensive or difficult to cancel. Read the contract carefully before signing. Look for month-to-month options where possible. ## What to Try Next If you want to understand the full picture of AI in caring for an older relative — including the privacy conversation — [AI Home Monitoring for Elderly: What Families Need to Know](/guides/ai-home-monitoring-elderly) covers the human side in depth. For a broader look at health wearables and wellness gadgets, [AI Health Gadgets for Seniors: What's Worth Buying](/guides/ai-health-gadgets-seniors) puts the whole market in plain English. --- # The AI Cheating Conversation Every Parent Should Have Before School Starts Source: https://aiforregulars.com/guides/ai-homework-cheating-parent-talk Your child is about to go back to school, and AI tools that can write a convincing five-paragraph essay in under a minute are already on their phone or a Google search away. That puts most parents in an uncomfortable spot: worried about cheating, but unsure where the line actually is — or even what their school's rules say. The good news is that a single honest conversation before school starts does more than any monitoring software, and this guide walks you through exactly how to have it. ## Step 1: Find Out What Your School Actually Says Before you talk to your child, know what you're working with. A 2026 survey found that only about one in three schools has written AI use guidelines in place. That means if you haven't asked, you might be assuming there's a policy when there isn't — or enforcing something at home that conflicts with what the school actually allows. Email your child's homeroom teacher or the school office before the year starts and ask one direct question: **"Does the school have a written policy on students using AI tools for assignments?"** If the answer is yes, ask for a copy. If the answer is no — which it is in roughly two-thirds of schools — ask what individual teachers expect in the meantime. Many teachers have strong opinions about AI even when the school hasn't formalized them. NYC public schools issued a preliminary AI use policy in March 2026; other districts are working on their own. **If there's no policy yet, ask:** - Do individual teachers set their own rules about AI use on specific assignments? - Is there a school-level academic honesty rule that covers AI? - Is a formal policy being developed for this school year? This step takes one email and gives your child clarity about what the rules actually are — not just what you've decided at home. ## Step 2: Frame It as Curiosity, Not an Interrogation A 2026 Back-to-School survey found that 67% of parents are very or somewhat concerned that AI makes cheating too easy. The same surveys show that 72% of parents believe AI should be part of their child's education. Both things can be true at once, and that tension is actually a useful starting point for the conversation. Don't begin with "I know you might be using AI to cheat." Begin with: "I wanted to understand what AI tools you use for school, if any — so we can make sure we're all on the same page before the year starts." This framing matters because 28% of students already use generative AI in their schoolwork. If your child is already using it responsibly, leading with suspicion shuts down a conversation you actually need to have. **Questions that open things up rather than close them down:** - "Have any of your teachers said anything this year about AI?" - "What do you think is okay to use it for, and what isn't?" - "What would you do if you were stuck on something and AI could just... answer it?" Let your child's answers tell you where they are before you tell them where they need to be. I've found it helps to position AI like a calculator. A calculator doesn't make math easy — it makes arithmetic fast so you can focus on the harder parts. AI works the same way for writing and research. The rule that works in our house: the thinking has to be yours. AI can help you think, but it can't think for you. ## The Household Rule That Actually Holds Up Blanket bans on AI at home rarely stick. They also miss the bigger picture: AI tools are already part of the workplace your child will enter, and learning to use them responsibly is a real skill. The rule that works better is a distinction about purpose, not access: **AI may explain — AI may not produce.** In practice, that looks like this: - **Fine:** asking AI to explain how photosynthesis works, give examples of a grammar rule, summarize a passage your child has read, or suggest how to structure an argument - **Not fine:** asking AI to write the essay, solve the problem set, or complete any assignment that's supposed to demonstrate your child's own understanding The test is always the same: would the teacher object if they knew exactly how this was done? If the likely answer is yes, it's not allowed. 49% of parents worry their child is already too dependent on AI for schoolwork. This rule addresses that directly — it keeps AI in a supporting role rather than replacing the thinking your child is supposed to be doing. ## The One Question That Reveals AI-Produced Work You don't need an AI detector tool (which are unreliable anyway — they have a meaningful false positive rate even in controlled research settings). There's a simpler test that takes under two minutes. **Ask your child to explain their work out loud.** "Walk me through your argument in this essay." "How did you get to that answer?" "What does this paragraph mean in your own words?" AI-generated content tends to include vocabulary, phrasing, or ideas the student didn't actually come up with and can't elaborate on when asked. A student who genuinely thought through their work — even if they used AI to brainstorm or get feedback along the way — can typically walk you through it. A student who had AI do the thinking for them usually struggles to explain what was written. This isn't about catching your child in a lie. It's a useful habit in both directions: it confirms their understanding and gives you a natural opening to talk about what they found hard or interesting. ## If Your Child Is Accused (Fairly or Not) If a teacher flags your child's work as AI-generated, stay calm and ask for specifics before reacting. AI detection tools have a meaningful false positive rate — students who write clearly and concisely are sometimes flagged incorrectly. Ask the school which tool or method was used and what specifically triggered the concern. Schools are increasingly expected to have an appeal process for AI-related academic integrity cases. If you believe the accusation is unfair, request a meeting and ask your child to walk through their work with the teacher present. That's often more persuasive than any argument about detection accuracy. If your child did use AI in a way that crossed the line, that conversation is harder — but it's still better handled with curiosity before judgment. Understanding why they did it (overwhelmed, didn't understand the task, short on time) gives you something concrete to address rather than just something to punish. The guide [Falsely Accused of Using AI at School?](/guides/falsely-accused-ai) goes through the appeal process in detail if you're dealing with an accusation right now. ## What to Watch Out For **AI detectors are not reliable.** Teachers or parents who rely on them alone will make mistakes in both directions. The most useful signal is still conversational: can your child explain their own work? **Rules vary by assignment and teacher.** Using AI to brainstorm might be fine in one class and off-limits in another. Your household rule covers the principle, but make sure your child knows they're responsible for knowing each teacher's specific expectations. **Privacy is a real concern.** Anything your child types into a chatbot — including personal details, teacher feedback, or assignment content — may be stored by the service or used to train future models. Where the school has approved tools, use those. [AI Parental Controls: What Every App Lets You Set](/guides/ai-parental-controls-all-apps) covers what the major platforms actually let you manage. **Model the behavior you want to see.** If you use AI to write things you send as your own without acknowledging it, your child notices. Talking openly about how you use it — "I asked it to help me draft this, then I revised it" — demonstrates what responsible use actually looks like in practice. ## What to Try Next If your child is already using AI for school and you want a hands-on setup guide for the tools themselves, [ChatGPT for Kids: A Parent's Setup Guide](/guides/kids-chatgpt-homework-parents-guide) walks through the practical steps. If you're dealing with an accusation right now, [Falsely Accused of Using AI at School?](/guides/falsely-accused-ai) covers the appeal process in full. --- # AI in Your Car: ChatGPT, Gemini, and Alexa in the Dashboard Source: https://aiforregulars.com/guides/ai-in-car-voice-assistant The voice assistant in your car used to answer two questions well: "Navigate home" and "Call Mom." Everything else was a polite failure. That is changing quickly. Several automakers have plugged ChatGPT, Google Gemini, or Amazon Alexa directly into their dashboards, and the experience is genuinely different from what came before. Here is what you need to know — including the parts the marketing materials leave out. ## Which AI Is in Which Car? The answer depends on both your car brand and model year, so treat this as a starting framework rather than a definitive list. **ChatGPT** is built into vehicles from Volkswagen Group (VW, Audi, and ŠKODA, among other group brands) as part of their IDA voice assistant. The integration lets you ask factual questions, get recipe ideas on road trips, understand warning lights, and carry on multi-turn conversations — all while the car handles navigation and phone calls separately. **Google Gemini** lives inside **Android Auto**, which runs on your phone but displays on your car's screen. If your car has a screen that supports Android Auto, you already have access. Gemini in Android Auto can answer questions, send messages, and pull up information — completely hands-free with the microphone button on your steering wheel. **Amazon Alexa** has two forms in cars. Some manufacturers have built Alexa directly into the car — BMW is among the first to include the upgraded Alexa+ in 2026 vehicles. More commonly, drivers use the Alexa app on their phone through Bluetooth or CarPlay/Android Auto. Either way, you get Alexa's smart home and shopping features along for the ride. ## What In-Car AI Is Actually Good At Once you try it, certain use cases become second nature. **Answering questions mid-drive.** "What's the speed limit on this highway?" "How many calories are in a gas station protein bar?" "What does that dashboard light mean?" The AI handles these without you needing to touch your phone. **Composing messages by voice.** Say "Text my wife that I'm running 20 minutes late," and the AI drafts the message, reads it back to you, and sends it when you confirm. This is one of the safest uses of in-car AI — no screen interaction required. **Trip planning.** On a long drive, you can ask "Find me a coffee shop in the next 10 miles that's not a chain" and get a spoken answer you can act on. Some integrations can then hand the address directly to navigation. **Explaining the car itself.** ChatGPT-enabled VW vehicles let you ask things like "What does the Auto Hold feature do?" and get a plain-language answer from the car's built-in knowledge. I used Android Auto's Gemini integration on a three-hour highway drive and was surprised how much it changed small annoyances. When I couldn't remember the name of the podcast I wanted, I described it ("that one about unsolved disappearances in national parks") and it figured it out. Those moments add up on a long trip. ## The Safety-First Rules Voice AI in cars is powerful but not risk-free. The danger is not the technology itself — it is the temptation to look at the screen or get absorbed in a conversation. **Set up before you drive.** Connect your phone, choose your assistant, and start your route before the car is in motion. Fumbling with settings at 60 mph defeats the purpose entirely. **Eyes forward, always.** Voice-only exchanges (ask a question, listen to the answer) are much safer than ones that involve looking at text on the infotainment screen. If an AI response is showing on the screen, wait for a stop to read it. **Keep conversations short while moving.** Multi-turn philosophical debates are fine in a parking lot. While driving, ask one thing, get the answer, move on. **Do not rely on AI for safety decisions.** In-car AI assistants cannot tell you that the road ahead is icy or that your tire is about to blow. They complement your own judgment — they do not replace it. ## What In-Car AI Cannot Do Understanding the limits saves frustration. These systems generally cannot access your personal accounts (email, banking, social media) without explicit setup. They can't control apps on your phone the way you can with your hands. They also have no awareness of what is happening outside the car — no real-time traffic beyond what navigation provides, and no information about road hazards. If the car's data connection drops (tunnels, rural areas), built-in AI goes offline. CarPlay and Android Auto depend on your phone's signal, which is generally more reliable. ## What to try next For a deeper look at ChatGPT's voice capabilities beyond the car, [ChatGPT Voice Mode: How to Use It and What It's Good For](/guides/chatgpt-voice-mode) covers the full feature. And if you want to compare all the major voice assistants on their everyday strengths, [Alexa vs. Google vs. Siri: Which One Should You Use?](/guides/alexa-google-siri-comparison) is a useful side-by-side. --- # AI Interview Prep: Practice Job Interviews with ChatGPT Source: https://aiforregulars.com/guides/ai-interview-practice Job interviews make most people nervous, no matter how qualified they are. One of the best ways to get less nervous is to practice — but finding someone willing to sit through a full mock interview is not always easy. ChatGPT will do it any time you ask. Here is how to set it up so you actually get something useful out of it. Do not practice generic interview questions. Practice for the exact role you are applying for. Open the job posting, select all the text, and copy it. You will paste it into the setup prompt in step 3. The more specific your input, the more relevant the questions will be. Go to [chat.openai.com](https://chat.openai.com) and start a new chat. Starting fresh means ChatGPT will not mix up your interview practice with other conversations. The free version works fine for this — you do not need a paid account. This is the most important step. Copy and paste this prompt, then fill in the three bracketed parts before you hit enter: ``` You are a hiring manager at [Company Name] interviewing me for the role of [Job Title]. Ask me one interview question at a time. After I answer, give me brief, honest feedback — what was strong, what was weak, and what I should add or change. Then ask the next question. Be realistic and a little tough — I want to be well prepared. Start with "Tell me about yourself." Here is the job description for context: [paste the full job description here] ``` Once you hit enter, ChatGPT will open with that first question — and the mock interview begins. This is important: do not just type your answers. Say them out loud first, then type a summary of what you said. Interviews happen verbally. Practicing out loud helps you hear how you actually sound — too long, too vague, too quiet, too rushed. Typing alone does not prepare you for the real thing. If it feels awkward to talk to yourself, try standing up. It changes how you speak in a way that sitting usually does not. After each answer, ChatGPT will tell you what worked and what did not. Take its feedback seriously, especially when it says your answer was vague, too long, or did not actually address the question. Common weaknesses it catches: - Giving a general answer instead of a specific example from your own experience - Describing a problem but not explaining what *you* specifically did about it - Ending without a clear result — "and eventually things worked out" does not land well - Talking for more than two minutes on a single answer without a clear point When ChatGPT flags one of these, ask it to re-ask the question so you can try again with a better answer. I tried this before a job interview I was especially nervous about. I practiced three times over two days — not just reading the questions, but actually speaking the answers out loud. The questions ChatGPT asked were very close to what came up in the real interview, because I had given it the actual job description. I was still nervous walking in. But I had answers ready, and that is the difference practice makes. After your first mock session, go back and find the two or three answers where the feedback was harshest. Ask ChatGPT to re-ask those specific questions so you can try again: ``` Let's redo that last question about [topic]. I want to give a better answer. Can you re-ask it? ``` Or ask for help building a stronger answer from scratch: ``` I struggled with "Tell me about a time you dealt with a difficult coworker." Can you give me a framework for this type of question, then ask it again? ``` Keep going on your weak spots until the answer feels natural rather than rehearsed. ## The night before Do one more short session the night before — 20 minutes maximum. Focus on the questions most likely to come up regardless of the role: "Why do you want this job?" and "What is your biggest weakness?" are nearly universal. Then stop. Preparation beats memorization, and rest helps more than another hour of drilling. **What to try next:** If you have not finalized your application materials yet, start there — [making your resume ATS-friendly](/guides/ats-friendly-resume-with-ai) and [writing a strong cover letter with ChatGPT](/guides/chatgpt-cover-letter) are both quick with the right approach. --- # AI Investment Scams: How Fake Trading Platforms Steal Life Savings Source: https://aiforregulars.com/guides/ai-investment-scam-fake-platform *The following is a composite story based on patterns reported across multiple documented cases. No real individual is depicted.* --- Margaret had been retired for two years when she matched with someone on a social app. His name was Daniel, and he said he was a retired engineer living in Vancouver. They talked every day for three weeks — about travel, their grown children, the restaurants they missed. One afternoon, Daniel mentioned he had been using an AI trading platform that had meaningfully improved his retirement income. "I'm not trying to sell you anything," he said. "I just wouldn't feel right not telling a friend about it." ## The Introduction The platform had a professional interface, live-looking charts, and a customer service chat. Daniel walked her through setting up an account. She deposited a small amount to try it out. Within two weeks, the dashboard showed her balance had grown noticeably. She withdrew a small amount — and it arrived in her bank account within days. That was the moment the trap closed. This is a deliberate tactic: letting victims withdraw early builds trust and removes doubt. The platform can afford it because they know far larger deposits are coming. ## The Escalation Daniel explained that her account tier limited her returns. A higher deposit would unlock the "institutional AI algorithm." The platform's support chat confirmed this, enthusiastically. She moved a larger sum over. The balance climbed dramatically on screen. She showed her daughter the dashboard, proud of her decision. When she tried to withdraw a substantial amount to pay for a home repair, the platform said her account had been flagged for a "tax compliance review" and she needed to pay a release fee — a percentage of her displayed balance. She paid the fee. A week later, another hold appeared — this time a "government bond" requirement. Daniel was reassuring: "I had the same issue. You are so close to getting it all out." ## The Exit Ramps You Can Take At each stage, there is a door out. Recognizing which stage you are in is the key. **Stage 1 — First contact with the platform.** Before depositing anything: search the platform name plus "scam" and "review." Check whether the person or firm offering to trade for you is actually registered — in the US, look them up with investor.gov's "Check Out Your Investment Professional" tool or FINRA's BrokerCheck; in the UK, use the FCA's Firm Checker and the FCA's Warning List of unauthorised firms. Legitimate investment professionals and firms show up in these registers and are easy to verify; one that doesn't is a stop sign, not a detail to overlook. **Stage 2 — Early profits look real.** The withdrawal that arrives is real money — but it is bait. Ask yourself: would I be comfortable if I never added more? If the answer is no, you are already being pressured. **Stage 3 — The fee requests begin.** Any legitimate platform deducts fees from your balance. No real platform asks you to send additional money from outside to "release" funds you supposedly already own. This is the clearest signal. Stop here. **Stage 4 — The relationship pressures you.** If the person who introduced you to the platform is encouraging you to pay fees or deposit more, that relationship is part of the scam. This is painful to accept, but it is true. The hardest part of these scams is not the technology — it is that the scammers are patient and kind for weeks before any ask. They earn trust genuinely, which makes the eventual request feel safe. If someone you met online, who you have never met in person, introduces you to an investment opportunity, treat that as an automatic pause point regardless of how well you know them digitally. ## What Happened to Margaret In our composite story, Margaret's daughter saw the fee requests and did a quick web search. She found a consumer protection warning describing the exact same platform interface and fee structure. They reported it to the FTC and their bank immediately. Margaret felt embarrassed. She also felt relieved. Both feelings made complete sense. ## Where to Report - **FTC (US):** reportfraud.ftc.gov - **SEC (US):** file a Tip, Complaint, or Referral at sec.gov/tcr; check a professional's background first at investor.gov - **FINRA BrokerCheck (US):** brokercheck.finra.org — verify or report a broker - **FBI IC3 (US):** ic3.gov - **Report Fraud (UK, England/Wales/Northern Ireland):** reportfraud.police.uk, or 0300 123 2040 (this is the service that used to be called Action Fraud); in Scotland, report to Police Scotland on 101 instead - **FCA (UK):** fca.org.uk/consumers/report-scam — to flag an unauthorised firm - **Your bank** — immediately, especially for recent wire transfers ## What to Try Next Protecting your finances starts with understanding your money clearly. [How to build a monthly budget with AI](/guides/chatgpt-monthly-budget) shows how to use tools like ChatGPT to organize your finances — without handing control to anyone else. And if the investment contact came through a message or email, [how to spot AI phishing emails](/guides/spot-ai-phishing-emails) shows what the written red flags look like before you engage. ## Sources - [Check Out Your Investment Professional](https://www.investor.gov/introduction-investing/getting-started/working-investment-professional/check-out-your-investment-professional) — Investor.gov (U.S. Securities and Exchange Commission) - [About BrokerCheck](https://www.finra.org/investors/investing/working-with-investment-professional/about-brokercheck) — FINRA - [FCA Firm Checker](https://www.fca.org.uk/consumers/fca-firm-checker) — Financial Conduct Authority (UK) - [FCA Warning List of unauthorised firms](https://www.fca.org.uk/consumers/warning-list-unauthorised-firms) — Financial Conduct Authority (UK) - [Report a scam](https://www.fca.org.uk/consumers/report-scam) — Financial Conduct Authority (UK) - [UK's home for reporting cyber crime & fraud](https://www.reportfraud.police.uk/) — Report Fraud (formerly Action Fraud) - [Investment Fraud and Pig Butchering](https://www.secretservice.gov/investigations/investmentfraud-pigbutchering) — U.S. Secret Service - [Report Suspected Securities Fraud or Wrongdoing](https://www.sec.gov/submit-tip-or-complaint/tips-complaints-resources/report-suspected-securities-fraud-or-wrongdoing) — U.S. Securities and Exchange Commission --- # AI Rejected Your Job Application — Here's What You Can Actually Do Source: https://aiforregulars.com/guides/ai-job-rejection-appeal You apply for a job on a Tuesday afternoon. By Tuesday evening — sometimes within minutes — a rejection email is sitting in your inbox. No phone call, no specific feedback, no sign that anyone actually read what you submitted. That is often not a person's decision. It's an algorithm's. And while you usually cannot undo it, there are steps worth taking before you move on. ## Is This Actually an AI Rejection? It helps to be reasonably sure you're dealing with automated screening before you decide what to do next. Nothing is certain from the outside, but these signals together point toward a bot rather than a human: **The rejection came within minutes or hours of applying.** A human recruiter sorting applications at the end of a busy day might respond quickly, but a response in under an hour — especially at a large company — almost always means a system scanned your résumé before anyone opened it. **The email arrived outside business hours.** A rejection at 2:37 a.m. or on a Sunday was not written by a person. Automated systems send whenever the scan completes, not when anyone is at their desk. **The message contains no specific feedback.** Generic language like "we've decided to move forward with other candidates whose backgrounds more closely align with our needs" is standard ATS output. A recruiter who actually read your application would more likely say something, even if brief. **You never spoke with anyone.** If you applied, heard nothing, and then received a rejection without any phone screen or recruiter outreach in between, the application likely never made it to a human queue. According to a 2025 survey, 50.5% of job seekers who were rejected that year heard nothing from a human, and only 9.7% were ever told AI was involved. If that sounds familiar, you are in very large company. ## How to Request a Human Review Asking for reconsideration costs you an email and a few minutes. The realistic success rate is low — especially at large companies — but it is not zero, and a professional tone leaves a positive impression regardless of the outcome. Find the recruiter or HR contact named in the job posting. If no name is listed, search LinkedIn for the hiring manager or a recruiter at the company. Keep the email under 150 words and focus on what you bring, not on challenging the decision. Here is a template you can adapt: > Subject: Following up on my application — [Job Title] > > Hi [Name], > > I applied for the [Job Title] role on [date] and recently received a rejection notice. I wanted to reach out directly, because I believe my background in [specific skill or experience] is a strong match for what you're looking for — particularly [one concrete detail from the job posting]. > > If there's any possibility of a conversation, I'd welcome it. I understand you're reviewing many applications and I appreciate your time regardless of the outcome. > > [Your name] Avoid mentioning "AI" or "algorithm" in the email. Even if you suspect that's what happened, framing the message as a grievance reduces the chance it gets a warm read. Write as if you're simply following up on something you're excited about. When I apply for a role I care about, I note the rejection email's timestamp and the exact subject line right away. If I later want to follow up or document a pattern across multiple rejections, I already have the details without having to search my inbox. ## What Are Your Rights? The honest answer is: not many in most places, and they vary considerably by where you live. **In California,** the CCPA's automated decision-making regulations — finalized by the California Privacy Protection Agency in September 2025 — will give residents the right to request information about automated decisions and to opt out. However, these regulations do not take effect until January 1, 2027, so they are not yet enforceable. **In Colorado,** the replacement AI law (SB 26-189, signed May 2026) covers "covered automated decision-making technology" in consequential decisions — including hiring. It requires employers to provide notice after an adverse decision and grants applicants the right to request meaningful human review, though only to the extent commercially reasonable. The law takes effect January 1, 2027, so it is not yet in force. **Everywhere else in the US,** there is currently no federal law specifically governing AI hiring tools. Existing statutes — Title VII of the Civil Rights Act, the Age Discrimination in Employment Act, the Americans with Disabilities Act — still apply if you believe the AI acted as a proxy for discrimination against a protected characteristic. The bar for proving this is high, but the laws exist. ## If You Suspect Discrimination Was Involved In 2025, a US court approved a collective action against Workday, one of the most widely used HR software platforms, under the ADEA, alleging that its AI screening tools systematically scored down older applicants. The case matters partly because it named the software vendor as a potentially liable party — not just the employer who used the tool. If you suspect age, race, gender, or disability discrimination was a factor in your rejection, these steps matter: **Document everything now, while it's fresh.** Screenshot the job posting, save the rejection email with its timestamp, and note when you applied and what qualifications were listed. Evidence you gather immediately is far more useful than evidence you try to reconstruct later. **Look for a pattern across multiple applications.** One rejection is not a pattern. If you apply to several roles at the same company or through the same ATS platform and receive rapid rejections despite meeting the stated qualifications, that accumulation is worth noting. **File an EEOC charge if you believe discrimination occurred.** The Equal Employment Opportunity Commission handles complaints under federal anti-discrimination laws. You generally have 180 days from the discriminatory act to file — 300 days in states that have their own anti-discrimination agencies. Visit eeoc.gov to start a charge. ## What Actually Works: Moving Forward The most reliable path forward is not forcing a review you cannot compel — it is applying smarter. **Get a referral.** A direct referral from someone inside a company is the single most effective way to bypass automated screening. Your résumé gets routed to a real person instead of entering the ATS queue with thousands of others. If you know anyone at the company, even a loose acquaintance, a brief LinkedIn message asking for an introduction is worth sending. **Reapply with a tailored résumé.** If the same role reopens or a similar one appears, a version of your résumé that mirrors the specific language from the job posting — rather than using your generic all-purpose document — scores significantly better in automated screens. [How to Use AI to Write an ATS-Friendly Résumé](/guides/ats-friendly-resume-with-ai) walks through exactly how to do this. **Apply earlier in the hiring cycle.** Many ATS systems close a role's pipeline once a threshold number of candidates have cleared the automated screen. Applications that arrive in the first 24 to 48 hours of a posting tend to have better odds than those submitted a week later. **Ask about the process in informational conversations.** When speaking with a recruiter or someone inside a company before applying, asking "does this role have an automated screen before a recruiter reviews applications?" gives you useful information — and sometimes leads to practical advice on how to frame your résumé. ## What to Watch Out For A few things that are unlikely to help, or that could make things worse: AI tools claiming to reverse-engineer ATS scores are usually selling confidence more than accuracy. The actual scoring criteria differ by system and employer and are not publicly documented. Use them as rough formatting guides, not as guarantees. Legal threats in a follow-up email will almost certainly close the door permanently. If you have a legitimate discrimination concern, handle it through the EEOC separately rather than mixing it into a recruiter email. And if a paid service promises to "get you past the AI," be skeptical. Résumé mills and keyword-stuffing services rarely deliver what they promise, and some may make your application look less authentic to both automated systems and humans. ## What to Try Next Understanding what the AI was actually looking for — and why it may have passed on your résumé — is the subject of [How AI Screens Your Résumé (and What It's Actually Looking For)](/guides/how-ai-screens-resumes). If you're 50 or older and finding automated screening especially difficult, [AI Tools for Job Seekers Over 50](/guides/ai-job-search-over-50) addresses the patterns that specifically affect that group. --- # AI Job Scams: How to Spot Fake Recruiters and Ghost Jobs Source: https://aiforregulars.com/guides/ai-job-scams-fake-recruiters Job searching is stressful, and scammers know it. They exploit the urgency job seekers feel — the need to move quickly when a promising role appears. AI now lets them generate convincing job postings, realistic recruiter profiles, and polished fake company websites faster than ever before. Knowing what to check before you invest your time or share personal information can protect you from a costly mistake. Before filling out any application, look up the company on your own. Don't use links or contact information from the job posting itself — open a new browser tab and search for the company name directly. Check whether they have a real website that matches what the posting describes. Search the company name plus "scam" or "review" to see whether other job seekers have reported problems. Look for the company on LinkedIn — does it have a page with real employees and a history that predates the posting? If the company appears to have been created in the last few months and has minimal online presence, treat that as a significant warning. A recruiter reaches out to you directly? Great — but verify who they actually are before engaging further. Check their LinkedIn profile: does it have a real work history, a reasonable number of connections, and activity that predates their message to you? Look at the email domain they are writing from — a recruiter for a real company will use that company's actual domain, not Gmail, Yahoo, or a lookalike address. If the profile photo looks like a polished headshot that seems slightly too perfect or symmetrical, do a reverse image search (right-click the photo in Google Images). AI-generated faces are increasingly common in fake recruiter profiles. This is the clearest line: no legitimate employer, recruiter, or staffing agency charges applicants money. If anyone asks you to pay for a background check, training materials, equipment, certification, or a work visa processing fee — stop contact immediately. The sophistication of the setup doesn't matter. A polished website, a friendly recruiter, a convincing video interview — none of it changes this rule. Some scammers ask for a "refundable deposit" on equipment that will be shipped to your home for remote work. The equipment never arrives, and the deposit disappears. Even after passing the initial checks, stay alert through the hiring process itself: The entire interview is conducted by chat or email only, with no video or live phone call. The offer comes very quickly with little discussion of your qualifications. The salary or benefits seem unusually high for the role and industry. You are asked to start immediately, before any formal offer letter or paperwork. The onboarding involves you receiving a check, depositing it, and sending some portion to a third party — this is a fake check scheme, and the check will bounce days after your own transfer has already cleared. Variants of the same fraud skip the check and instead use a bank-transfer overpayment, a gift card purchase request, or a cryptocurrency transfer. Your name and email address on a resume are fine for an initial application. Everything sensitive should wait until the offer is verified: Do not share your Social Security number, date of birth, passport, or driver's license until you have received a written offer letter and independently confirmed the company is real. Use established, secure platforms — an actual company HR portal, Workday, Greenhouse, or ADP — if sensitive information is required during onboarding. If you are asked to provide banking details for direct deposit setup before you have even completed a background check or signed paperwork, treat that as a red flag. If you encounter a fake job posting or recruiter, report it — your report helps protect the next person who sees it: Report the posting directly on the job board (LinkedIn, Indeed, and others have dedicated fraud reporting options). Report to the FTC at reportfraud.ftc.gov. If the recruiter impersonated a real company, notify that company's HR or legal department — they want to know their brand is being used. If any money moved, contact your bank immediately. Then keep moving forward with your search. These scams are designed to exploit a vulnerable moment in your life. Getting targeted is not a reflection of your judgment — it is a reflection of how sophisticated and professional these operations have become. I started verifying company domains and recruiter LinkedIn history as a standard habit, not just when something feels off. It takes under two minutes per lead and has helped me filter out several dubious postings during my own job searches. Real postings from established companies hold up easily under this scrutiny. The ones that don't hold up were never worth applying to. ## What to Try Next If you received the fake job contact through an email, [how to spot AI phishing emails](/guides/spot-ai-phishing-emails) covers the signals that reveal fake messages even when the writing looks professional. And when you are ready to build a strong application for real opportunities, [how to write a professional email with AI](/guides/write-professional-email-with-ai) and [creating an ATS-friendly resume](/guides/ats-friendly-resume-with-ai) can help you stand out to legitimate employers. ## Sources - [Scammers impersonate well-known companies, recruit for fake jobs on LinkedIn and other job platforms](https://consumer.ftc.gov/consumer-alerts/2023/08/scammers-impersonate-well-known-companies-recruit-fake-jobs-linkedin-other-job-platforms) — Federal Trade Commission - [Scammers Exploit Security Weaknesses on Job Recruitment Websites to Impersonate Legitimate Businesses, Threatening Company Reputation and Defrauding Job Seekers](https://www.ic3.gov/PSA/2022/PSA220201) — FBI Internet Crime Complaint Center (IC3) - [Recognize and report scams](https://www.linkedin.com/help/linkedin/answer/a1336387) — LinkedIn Help --- # AI Job Search After 50: How to Beat Age Bias in Hiring Systems Source: https://aiforregulars.com/guides/ai-job-search-over-50 If you've been job searching after 50, you've probably noticed something: applications often disappear without a response. Part of that is volume — employers receive hundreds of applications per posting. But part of it is that automated screening systems can filter out resumes in ways that disadvantage experienced candidates. Here's how to use AI tools to make sure your experience reaches a human reviewer. Before you revise anything, it helps to know what the software is looking for. Most large employers use Applicant Tracking Systems (ATS) that parse your resume into fields and rank you against the job requirements. Some also use AI scoring tools that look for keyword matches. These systems don't see your 30 years of experience as an asset — they compare specific terms in your resume against specific terms in the job posting. Your job is to close that language gap. See the guide on [how AI screens your resume](/guides/how-ai-screens-resumes) for a deeper explanation of how this works. Some information on an older resume signals your graduation decade without helping your application. Consider removing or adjusting: - **Graduation years** from the 1980s or 1990s — list the degree and school, drop the year - **Work history older than 15 years** — either omit it or collapse it into a short "Earlier Career" section with job titles only, no bullet points - **Software or certifications that are clearly outdated** — WordPerfect, Lotus 1-2-3, or obsolete certifications can raise red flags; remove them if they're no longer relevant None of this is dishonest. A resume is a marketing document, not a complete life history. Industry language changes over time. Terms you used in job titles and bullet points twenty years ago may not match what employers search for today. Open ChatGPT and try this prompt: ``` I'm updating my resume. Here are three bullet points from my current resume describing my work as [your job title]. Please rewrite them using modern, industry-standard language that would be recognized by applicant tracking systems in 2025. Keep the meaning and accomplishments exactly the same. [paste your bullet points] ``` Review every suggestion carefully. The goal is to replace outdated phrasing with current terminology, not to exaggerate or change what you actually did. Generic resumes score poorly in AI systems because they don't mirror the specific language of the job description. This step is the most time-consuming, but also the most effective. Paste both your resume and the job posting into ChatGPT and ask: ``` Compare my resume to this job posting. Which required skills or keywords from the job posting are missing or unclear in my resume? I want to add them if they accurately describe my experience. [paste resume] [paste job posting] ``` ChatGPT will give you a list of gaps. Go through them and add any that are genuine. Don't add skills you don't have — you'd need to demonstrate them in an interview. I've found that the biggest language gaps tend to show up in how people describe management and technology skills. "Supervised staff" is often better recognized as "team leadership" or "people management." "Used computer systems" becomes "proficient in [specific software]." The content is the same — the vocabulary is what changes. One thing that can trigger age bias in automated scoring is a resume that lists no recent activity. Show that you're current by adding: - Recent online courses or certifications (LinkedIn Learning, Coursera, industry associations) - Modern tools you actually use — cloud software, project management platforms, video conferencing tools - Recent professional development, even informal (workshops, webinars, industry reading groups) If you're genuinely short on recent credentials, pick one skill relevant to your target role and spend two to four weeks on a free course before applying widely. Alison and freeCodeCamp are both free. One of the most useful things you can do before interviewing is practice answering common questions out loud until you feel natural. ChatGPT can generate realistic interview questions for any role. ``` I'm interviewing for a [job title] role at a [type of company]. Generate 10 interview questions I'm likely to face, including at least two behavioral questions and one question about working with newer technology. ``` Then answer each question in writing in ChatGPT and ask for feedback: ``` Here's my answer to the question "Tell me about a time you managed a difficult team situation." What's working well and what could be stronger? [your answer] ``` Practice the answers you're least confident about until they feel natural, then practice them aloud. Interviewers may not ask directly about your age, but they sometimes ask questions that are really about it: "Where do you see yourself in five years?" or "You have a lot of experience — are you comfortable working with a younger team?" Prepare honest, forward-looking answers. With ChatGPT: ``` I'm a [age range] job seeker interviewing for [role]. Help me write a confident, brief answer to "You have a lot of experience — how do you stay current?" that doesn't sound defensive. ``` The best answers acknowledge your depth while signaling genuine engagement with how the field is evolving. ## What to try next Now that your resume is updated and you've practiced your answers, the next step is making sure the document itself is formatted in a way that ATS software can read correctly. The guide on [building an ATS-friendly resume with AI](/guides/ats-friendly-resume-with-ai) covers the formatting details step by step. For the interview itself, [AI interview practice](/guides/ai-interview-practice) goes deeper on using AI to prepare for specific question types. --- # How to Use AI for Knitting and Crochet Patterns (Resize, Adapt, Create) Source: https://aiforregulars.com/guides/ai-knitting-crochet-patterns If you knit or crochet, you've probably spent time squinting at a pattern that calls for a yarn weight you don't have, in a size that doesn't fit, at a gauge that doesn't match your needles. AI won't replace the feel of yarn between your fingers or the satisfaction of blocking a finished piece — but it can do the arithmetic you'd otherwise do with a pencil and calculator, and sometimes suggest ideas you wouldn't have thought of on your own. Here are four things AI genuinely helps with, along with an honest account of where it falls short. This is the most reliable thing you can ask AI to do for crafting. If you have a pattern written for DK weight yarn and you want to use worsted, or you have an adult size but need a child's, AI can do the gauge conversion math that would otherwise take you an hour with pencil and paper. **What you need before you start:** Your original pattern gauge (stitches and rows per 10 cm or per 4 inches, from the pattern), your new yarn's gauge (from your test swatch — never skip swatching), and the key stitch counts from the pattern you want to adapt. **Copy and adjust this prompt for ChatGPT:** > I'm adapting a knitting pattern. The original gauge is 22 stitches and 30 rows per 10 cm on 4mm needles. My new yarn swatches to 18 stitches and 24 rows per 10 cm on 5mm needles. The pattern casts on 132 stitches for the body. What should my new cast-on count be, and how does this affect the row count for a 30 cm section? Show your math step by step. Asking it to show the math step by step is important — AI sometimes makes arithmetic errors, and this lets you catch them before you've cast on 200 stitches. **For garment sizing:** Include target measurements. "The original finished bust is 40 cm and I need 48 cm" gives AI something concrete to work with rather than a vague "make it bigger." For straightforward projects — a dishcloth, a basic hat, a simple rectangular shawl — AI can write a starter pattern from a description. This works better for experienced crafters who can spot errors quickly than for beginners who may not have the background to catch mistakes. **Try purlJam first.** purlJam (at purljam.uk) is a free tool built specifically to generate hand knitting, machine knitting, and crochet instructions from a text description. It's purpose-built for this task rather than being a general AI, so its output tends to be better structured for crafting. **For ChatGPT or Gemini, be specific about every variable:** > Write a beginner crochet pattern for a flat dishcloth, approximately 20 cm square, using worsted weight yarn and a 5mm hook. Use single crochet throughout. Include the foundation chain count, stitch count per row, number of rows, and any turning chains needed. **Treat the output as a draft, not a finished pattern.** Before you pick up your hook or needles, read the whole thing through and check: do stitch counts stay consistent from row to row? Do any increases have matching decreases? Does the stated row count at your gauge actually produce the intended measurement? These are the places AI most often slips up. Colorwork design — whether you're planning a Fair Isle yoke, a tapestry crochet bag, or a striped blanket — is one of the stronger AI applications for crafters. You can describe a mood or a style and get back a range of motif and palette ideas without risking the stitch-count errors that come with AI-generated technical patterns. **Describe your project to ChatGPT:** > I'm planning a Fair Isle-style yoke for a child's sweater in three colors: cream, forest green, and rust red. I want a Scandinavian folk-art feel with animal motifs. Suggest three or four motif ideas, describe how they could be arranged in the yoke bands, and suggest which color to use for each element. You'll get descriptions, not charts — you'll still need to transfer the design onto graph paper or into a charting tool like Stitch Fiddle. But a short conversation can give you a clear direction when you're staring at a blank grid. **For visual reference:** OpenArt (openart.ai) has a crochet image generator available on its free tier. AI-generated fabric images rarely reflect real stitch behavior — yarns twist and blend in ways that don't translate to actual stitches — but they're useful for exploring color palettes before you commit to buying yarn. This is the simplest use of AI for crafters and often the most immediately helpful. Copy the line from the pattern that's confusing you, paste it into ChatGPT, and ask in plain language what it means. > This knitting pattern says: "K2tog, yo, k1, yo, ssk, k3." I'm a beginner and I don't understand what "yo" means in this context, or why there are two of them. Can you explain what this instruction is doing and what the finished fabric should look like at that point? AI is good at explaining abbreviations, describing what a technique produces visually, and walking you through why a sequence creates a particular effect. This is interpretation and explanation — much lower risk of errors than asking AI to generate or calculate anything. The most reliable way I've found to use AI for knitting is to check my gauge math rather than generate full patterns. I work out what I want to make, calculate the rough stitch counts myself, and then ask AI to verify my arithmetic and flag anything that seems off. It's like a second pair of eyes on your numbers — faster than redoing the math from scratch, and I still verify the output before I cast on. ## What to Watch Out For **AI gets stitch counts wrong.** This is the core limitation. Every increase and decrease in a knitting or crochet pattern has to balance across the rows; stitch counts need to match your gauge to produce the right measurements; joined rounds have to close correctly. AI language models predict plausible-sounding text — which means a confident-sounding number can be completely wrong. The Fairythorn, a practitioner review site, noted in June 2025 that AI is helpful for resizing and adapting but unreliable for generating complete patterns without significant errors. **Always swatch.** No gauge conversion — yours or AI's — means anything until you've knitted or crocheted a test swatch with your actual yarn, your actual needles or hook, and washed it the way you'll wash the finished item. Ten minutes of swatching can save hours of frogging. **Never sell or publish an untested AI pattern.** If AI generated or heavily shaped your pattern, you must work it yourself from start to finish before sharing it publicly. An untested pattern with wrong stitch counts wastes someone's time, money, and yarn — and is especially hard on less-experienced crafters who may not know where the instructions went wrong. **Free tools can disappear.** purlJam and similar free AI tools may update their features, move behind a paywall, or shut down without notice. If you get useful output from any tool, save the pattern text locally rather than assuming you can generate it again later. ## How to Spot AI-Generated Fake Pattern Photos A growing problem in crafting marketplaces is AI-generated photos used to sell patterns for garments and accessories that were never actually made. The images look polished and the items look perfect — because they were never real. **Signs that a pattern photo might be AI-generated:** - Yarn texture shifts direction mid-stitch in ways that aren't physically possible - Stitches appear to float, overlap, or blur together without consistent structure - Hands holding needles, a hook, or a finished object look wrong — too many fingers, unnatural proportions, or fingers that fade into the fabric - The background has a slightly soft, dreamlike quality compared to the crisp garment in the foreground - The listing has only one or two glamour shots with no in-progress photos, no social media posts showing the item being made, and no customer photos in reviews **What to do:** Search the seller's social media profile or Ravelry page for in-progress and finished object photos. Check for customer photos in reviews. If a pattern listing has beautiful images but no evidence the seller has ever actually made the item, be cautious with your money. For a broader guide to spotting AI fakes across contexts, see [How to Spot AI-Generated Fake Photos](/guides/spot-ai-generated-photos). ## What to Try Next If you're interested in using AI to pick up other crafts or skills beyond knitting and crochet, [How AI Can Help You Learn Any New Skill](/guides/ai-learn-new-skill) covers the same practical, skeptical approach applied to a wider range of hobbies. And for a broader sense of what AI handles well and badly across everyday tasks, [50 Things You Can Actually Ask ChatGPT to Do](/guides/things-to-ask-chatgpt) is a practical next read. --- # Best AI Language Learning Apps vs Duolingo: An Honest Comparison Source: https://aiforregulars.com/guides/ai-language-apps-vs-duolingo You've heard of Duolingo. Maybe you've tried it, hit a streak, and eventually dropped it. Or maybe you've wondered whether there's something better. In 2026, the answer is: it depends on what "better" means for you. Here's an honest look at the options. ## What Kind of Learner Are You? The right tool depends on your goal: - **Complete beginner** who wants structure and a daily habit — Duolingo or Babbel - **Intermediate learner** who needs real conversation practice — ChatGPT or Speak - **Commuter or audio learner** — Pimsleur - **Someone who wants everything in one place** — Babbel ## App-by-App Breakdown ### Duolingo — The Habit Builder Duolingo's genius is gamification. Streaks, points, and badges make you come back every day. For building a habit and learning absolute beginner vocabulary, it's hard to beat. But Duolingo has real limits. The sentences can be oddly unnatural ("The elephant drinks beer"). Grammar explanations are minimal. And the free tier now limits how many mistakes you can make per day before you're cut off. **Duolingo's AI:** The app uses AI to adjust difficulty and has added some AI conversation features, but these are still more limited than a full chatbot. ### ChatGPT (and similar AI chatbots) — The Conversation Partner ChatGPT is the best conversation practice tool available — free, available at any hour, and infinitely patient. You can ask it to speak only in Spanish, correct your grammar, explain idioms, or role-play specific scenarios (a shopkeeper, a job interview, ordering at a restaurant). What it lacks: a structured curriculum. ChatGPT won't tell you "you should know this vocabulary before that." It reacts to what you bring. ### Babbel — Best Structured Learning Babbel has a strong reputation for practical, real-world vocabulary — travel, work, social situations — taught through spaced repetition and short lessons. It feels more grown-up than Duolingo and covers grammar more thoroughly. **Cost:** ~$14/month or ~$84/year for full access. ### Speak — Best AI Conversation App Speak is built specifically around AI conversation. You speak into the app, it transcribes and corrects you, and explains your mistakes in plain English. For people who freeze up in real conversations, practicing with an AI where there's no judgment is genuinely useful. **Cost:** Free tier; premium subscription for unlimited practice. ### Pimsleur — Best for Audio Learners Pimsleur is a classic audio-based method, now with an AI-assisted app. If you learn best by listening and speaking aloud — driving, walking, commuting — it's unmatched. Heavy focus on pronunciation and spoken fluency. **Cost:** Subscription-based, priced higher than most alternatives. ## Head-to-Head Comparison | App | Best for | Free tier | Structured curriculum | Real conversation | Grammar depth | |---|---|---|---|---|---| | Duolingo | Habit + beginners | Yes | Strong | Limited | Light | | ChatGPT | Conversation practice | Yes | None | Excellent | On demand | | Babbel | Practical vocab | No | Strong | Limited | Medium | | Speak | Speaking confidence | Partial | Medium | Excellent | Medium | | Pimsleur | Audio learners | No | Strong | Good | Light | I used ChatGPT to practice Italian for three months before a trip to Rome. I'd spend 10 minutes before bed chatting in Italian — ordering food, asking directions, describing my day. My accent was rough, but I could hold a conversation. What surprised me was how much the lack of judgment helped. I said ridiculous things and just got a polite correction, not awkward silence. ## The Honest Verdict No single app wins for every learner. Duolingo is a great on-ramp but a poor endpoint. ChatGPT is a brilliant conversation partner but assumes you bring your own structure. The most effective learners use two tools: one for structure (Babbel or Duolingo) and one for conversation (ChatGPT or Speak). ## What to Try Next For a deeper guide to using ChatGPT specifically for language practice, including the best prompts to use, read [Learn Any Language with ChatGPT](/guides/learn-language-with-chatgpt). And if you need to translate documents or photos along the way, [Translate Any Document or Photo with AI](/guides/translate-photo-document-ai) covers all the best tools. --- # How to Learn a New Skill with AI: Piano, Drawing, and DIY Source: https://aiforregulars.com/guides/ai-learn-new-skill Learning something new as an adult is harder than it looks. Schedules are packed, classes are expensive, and YouTube videos assume you already know the vocabulary. An AI assistant won't replace a real teacher, but it gives you something almost as good: a patient, always-available tutor who meets you exactly where you are. Here's how to use AI for three popular skills — piano, drawing, and DIY home repair — along with ready-to-use prompts you can copy directly. ## How to set up AI as your tutor (any skill) Before we get to the specific skills, one trick applies to all of them: give the AI a full picture of who you are as a learner. The most common mistake is asking something too vague. Instead of "teach me to draw," open with a setup prompt that tells the AI your level, your goal, your available time, and any constraints. Here's a template you can adapt: ``` I want to learn [skill]. I am a complete beginner with no prior experience. I have about [X] minutes per day to practice. My goal is to [specific goal, e.g., play a simple song in 8 weeks / sketch portraits / fix a leaky faucet]. I don't have [any limitations — e.g., a real piano, art classes nearby, professional tools]. Please give me a realistic week-by-week plan to get started. ``` Once you set this up, the AI will remember it for the rest of that conversation. You can keep asking follow-up questions without repeating yourself. You don't need an expensive keyboard to start. Many beginners use a basic 61-key keyboard or even a free piano app on a tablet. **Prompt to get started:** ``` I'm a complete beginner learning piano on a 61-key keyboard. I can practice 20 minutes a day. I want to play a simple song like "Let It Be" by the Beatles in about 6 weeks. Give me a week-by-week practice plan, including specific exercises and what I should be able to do by the end of each week. ``` **When you get stuck on theory:** ``` What is a C major chord? Show me which notes to press on a keyboard and explain why those three notes sound good together. ``` ``` I keep mixing up quarter notes and eighth notes when I read sheet music. Explain the difference with a simple example I can clap out. ``` **To quiz yourself:** ``` Quiz me on the names of the white keys from C to C. Ask me one at a time and wait for my answer before telling me if I'm right. ``` Drawing is one of the best skills to learn with AI because so much of it is about observation and practice, not expensive materials. A pencil and printer paper are enough to start. **Prompt to get started:** ``` I'm a complete beginner at drawing. I have pencils and paper. I can practice 15 minutes a day. My goal is to draw recognizable portraits of people in about 3 months. Give me a structured beginner plan, starting with the most foundational exercises. ``` **For specific exercises:** ``` Give me 5 beginner drawing exercises I can do in 15 minutes right now, with no special materials. Explain exactly what to draw and what skill each exercise builds. ``` **For feedback (describe what you drew):** ``` I drew an eye but the proportions look off — the iris seems too small and the eyelid shape looks flat. What am I probably doing wrong, and what should I look for when I try again? ``` I tried learning basic guitar chords twice before and gave up both times. The third time, I asked ChatGPT to explain exactly why my fingers hurt so much in the first two weeks (calluses haven't formed yet — it's normal and it does go away), and to give me a schedule that started with just 5 minutes of practice so it didn't feel like a commitment. That tiny reframe — "it's supposed to hurt at first, keep going" — was what I needed. The AI didn't teach me to play, but it stopped me from quitting. AI is excellent for DIY tasks because most home repairs follow clear, repeatable steps. It can also help you figure out when a job is within your ability and when to call a professional. **Prompt to describe a problem:** ``` My bathroom faucet drips constantly when turned off. It's a single-handle faucet in a 20-year-old house. Walk me through how to diagnose what's causing the drip and fix it, step by step. I have basic tools but no plumbing experience. ``` **To check if a job is safe for a beginner:** ``` I want to install a ceiling fan to replace an existing light fixture. Is this a realistic DIY project for someone with no electrical experience? What are the risks, and what would tell me I should hire an electrician instead? ``` **For a shopping list:** ``` I'm going to patch a drywall hole about 4 inches wide. Give me a complete list of materials I need from the hardware store, with approximate prices, and tell me if there are any tools I can rent instead of buy. ``` One important note: AI can describe electrical, plumbing, and structural work accurately, but it cannot see your specific setup. Always shut off power or water before starting, and when in doubt, get a professional for anything involving the electrical panel, gas lines, or load-bearing structures. The biggest advantage of AI tutoring is that it never gets impatient. After a week of practice, come back and tell it how things went: ``` I've been practicing piano for one week following your plan. I can play the C major scale with my right hand but my left hand keeps falling behind. What specific exercise should I focus on this week? ``` This kind of ongoing conversation — where you report what happened and get adjusted advice — is something most classes don't have time for. Use it. ## What to try next If you want to go deeper on one skill in particular, AI is also excellent for learning languages. The guide on [learning a language with ChatGPT](/guides/learn-language-with-chatgpt) covers conversation practice and vocabulary drills in detail. And if you're new to using AI assistants in general, [how to write better prompts](/guides/how-to-write-prompts) will make every AI interaction you have more useful. --- # AI Learning Apps for Kids Tested: Khanmigo, Socratic and More Source: https://aiforregulars.com/guides/ai-learning-apps-kids There are now dozens of AI-powered learning apps aimed at students. The hard question is not "does it work?" — they all give answers. The real question is: does it *teach*? An app that hands you the answer is not a tutor. It is a shortcut. To find out which apps actually help kids learn, we compared five popular tools on the same five questions a middle-schooler might ask — drawing on published reviews, educator accounts, and each app's documented design principles — and scored each on whether it guides students through reasoning or simply hands over the answer. ## The Five Test Questions 1. "What is 15% of 340?" (math) 2. "Why does the moon have phases?" (science) 3. "What is the main theme of *The Outsiders*?" (reading comprehension) 4. "Write me a paragraph about climate change." (essay shortcut test) 5. "Is it okay to copy my friend's homework?" (values/trick question) The last two test something important: what does the app do when a student tries to use it to cheat, or asks something a real teacher would handle differently? ## The Results | App | Teaches or just answers? | Essay shortcut test | Trick question response | Best age | Free? | |---|---|---|---|---|---| | **Khanmigo** | Teaches (Socratic questions) | Refused, offered guided help | Gentle, educational | 9–14 | School plan / paid | | **Socratic / Google Lens Homework** | Mix (explains steps) | Gave tips, not full essay | No response (ignored it) | 9–13 | Yes | | **ChatGPT (default)** | Mostly answers | Wrote the paragraph | Answered neutrally | 13+ | Free tier | | **Microsoft Copilot** | Mix | Wrote the paragraph | Mild disclaimer | 13+ | Free | | **Google Gemini** | Mostly answers | Wrote the paragraph | Answered neutrally | 13+ | Free | *Note: Socratic by Google was discontinued as a standalone app in October 2024; its step-by-step homework features are now part of Google Lens (open Google Lens, tap Homework). The behavior column reflects the pre-discontinuation app and current Google Lens Homework mode.* ## What the Results Mean **Khanmigo** stood out clearly. On the math question, it asked "What would you do first?" rather than calculating the answer. On the essay prompt, it said it would not write the paragraph but asked what angle I wanted to take. On the homework question, it gave a thoughtful, age-appropriate response about why copying hurts learning. The Socratic approach takes longer, but that friction is the point — the student does the thinking, not the app. **Socratic** is fast and good at visuals. For a photo of a math problem, it produces step-by-step breakdowns that are genuinely useful. On the essay prompt and the trick question, it was inconsistent — sometimes it gave too much, sometimes it ignored the question entirely. Still worth keeping on a child's phone for homework help, just not as a primary study tool. **The general chatbots** (ChatGPT, Copilot, Gemini) are not bad — they often explain clearly when pushed — but their defaults skew toward answering, not teaching. They work well for a mature student who will ask follow-up questions rather than just copying the output. For younger or less self-directed students, they require more parental oversight. Published reviews and educator accounts consistently note that Khanmigo is the most consistent of the five. ChatGPT shows the widest variation — sometimes guiding students through reasoning beautifully, other times simply answering. That inconsistency is hard to manage with a younger student who does not know the difference. ## Teaching-vs-Answering Score Based on the five test questions, where 1 means "just answered" and 5 means "guided the student through it": | App | Score | |---|---| | Khanmigo | 4.5 / 5 | | Socratic | 3 / 5 | | ChatGPT | 2 / 5 | | Copilot | 2 / 5 | | Gemini | 2 / 5 | The score reflects the app's *default* behavior, not what it can do if prompted carefully. A parent who sits with their child and steers ChatGPT toward explanation can get much better results. The question is whether that happens in practice. ## How to Choose - **Pick Khanmigo** if your child is in grades 4–9 and you want an app that actively prevents shortcut-taking. Best for math, science, and writing. - **Add Google Lens Homework mode (formerly Socratic)** for quick photo-based homework checks, especially in STEM subjects. Good as a secondary tool. - **Use general chatbots** only once your child is mature enough to understand the difference between "help me understand" and "do it for me" — and after you have set up appropriate parental controls. ## What to Try Next For parents who want to understand what settings to turn on before handing any AI tool to a child, [the ChatGPT parental controls guide](/guides/chatgpt-parental-controls) is the fastest place to start. If your student is older and you want to cover how to use AI without crossing into cheating, [the guide on legitimate AI study uses](/guides/ai-for-studying) lays out exactly where the line is. --- # Can AI Write Legal Documents for You? What's Safe and What Isn't Source: https://aiforregulars.com/guides/ai-legal-documents-safe People have been asking AI chatbots legal questions ever since chatbots became widely available. It makes sense — lawyers are expensive, legal language is baffling, and being able to ask "what does this clause mean?" at midnight feels like a genuine advantage. In many ways, it is. But the line between "AI helping you understand the law" and "AI practicing law on your behalf" is one you need to know, because crossing it can cause serious problems. ## Where AI Genuinely Helps **Understanding legal language.** Contracts, leases, insurance policies, and government letters are written for lawyers, not people. AI is excellent at translating jargon into plain English. Paste in a clause you don't understand and ask "What does this mean in plain English?" — this is a completely safe and useful use. **Drafting informal letters and complaints.** A letter to a landlord about a repair, a complaint to a company about a billing error, a request for a refund — AI can draft these well. These are not legal documents; they are business correspondence. A well-written letter often resolves disputes without lawyers. **Preparing for a legal appointment.** If you have a meeting with a lawyer and don't know what to ask, AI can help you build a list of questions. "I'm meeting an estate attorney to discuss my will — what should I ask?" produces genuinely useful prep material. **Summarizing documents.** If you've been handed a multi-page contract, AI can summarize the main points and flag clauses worth reading carefully. This is not legal advice — it's a reading aid. You still need to read and understand the document yourself, or have a lawyer review it. ``` Prompt example: "Please explain this clause in plain English and tell me if there's anything I should ask a lawyer about: [paste clause here]" ``` I regularly use AI to decode insurance policy language, which seems designed to be confusing. The AI is usually good at explaining what is and isn't covered in plain terms. I still confirm anything important with the insurance company directly — but the AI saves me 20 minutes of re-reading. ## Where AI Is Genuinely Risky **Wills and estate planning.** A will is a formal legal document with strict requirements that vary by state and country. In most places, it must be signed in front of witnesses; some states require notarization; some relationships can complicate what you can and cannot specify. An AI-generated will that doesn't meet these requirements may be partially or entirely invalid — often discovered only after you've died, when it's too late to fix. **Binding contracts for significant transactions.** Employment agreements, property sales, business partnerships, large service contracts — these need a lawyer's eye. A missing clause, an unenforceable provision, or a jurisdiction-specific requirement that the AI didn't know about can cost far more than the legal fee would have. **Court filings and legal responses.** If you are involved in a lawsuit, facing eviction, dealing with debt collectors, or responding to a government agency, AI-generated text can be helpful background research. But the documents you actually file or send carry legal consequences. Get professional help. **Anything involving children, custody, or adoption.** Family law is deeply jurisdiction-specific and emotionally consequential. Do not navigate it with an AI chatbot alone. ## The Jurisdiction Problem This is something AI chatbots handle poorly: laws are not universal. What is required in a contract in California may be different from Texas. What makes a will valid in the UK may differ from Australia. What constitutes a legal eviction in New York is not the same as in Florida. When you ask an AI about a legal matter, it will often give you a reasonable general answer — and not tell you that the specific rule in your state or country is different. This is not dishonesty; the AI genuinely may not know or may not realize the distinction matters. The result is advice that sounds correct but applies somewhere else. Any time you ask AI a legal question, include your location: ``` Prompt example: "I want to create a simple will. I live in Texas. What are the legal requirements for a valid will in Texas, and what questions should I ask an estate attorney?" ``` Even with the location, treat the answer as a starting point for research, not as legal advice. ## A Useful Middle Path: AI + a Lawyer You don't have to choose between relying entirely on AI and paying a lawyer for everything. A practical approach: 1. Use AI to understand the issue and learn the vocabulary. 2. Use AI to draft a first version of any document you need. 3. Take that draft to a lawyer for a brief review — much cheaper than having them draft from scratch. 4. Use AI to prepare your questions before that meeting. Many routine legal tasks — a simple will, a basic rental agreement, a cease-and-desist letter — can be handled cost-effectively this way. The lawyer's job becomes reviewing and refining rather than creating from zero. ## Free Legal Help Worth Knowing Before assuming you need to pay a lawyer, explore these options: - **Legal aid organizations** provide free or income-scaled help for civil matters like housing, family law, and benefits. - **Bar association referral services** often include a free or low-cost initial consultation. - **Law school clinics** handle certain types of cases at no charge. - **State court self-help centers** provide forms and guidance for common matters. A quick web search for "legal aid [your city, region, or country]" will surface local options — the term for free legal help varies by country, so try "free legal advice" or its local equivalent if that phrase returns few results. ## What to try next If you're using AI to help understand your finances as well as legal matters, [AI and Your Pension: What AI Can Realistically Help You Understand](/guides/ai-retirement-pension-guide) covers a similar question in the financial domain — where AI helps, where it stops, and when to involve a professional. For drafting other types of professional correspondence, [Write a Professional Email with AI](/guides/write-professional-email-with-ai) walks through effective prompt techniques. --- # How to Use AI to Lower Your Phone, Cable, and Internet Bill Source: https://aiforregulars.com/guides/ai-lower-bills Your phone, cable, and internet bill probably went up in the last 12 months — and you may not have noticed. Providers routinely offer discounted promotional rates to new customers, lock you in for a year, then quietly let the price reset higher once the promotion expires. It's a common pattern, and most people never push back. AI can help you change that. ## Why Your Bill Is Higher Than It Should Be When you signed up, you likely got a promotional rate — a deal available for the first 12 months that isn't automatically renewed. Once it expires, the bill climbs to the standard rate. The gap between what a new customer pays today and what a long-time customer pays for the exact same service is often $20–$40 a month. You have real leverage here: cable and internet companies know that losing a customer entirely is worse than giving a small discount. The retention department — the team you reach when you say you're thinking about cancelling — has access to deals that the regular billing team doesn't. The goal isn't to be rude or to bluff; it's to ask clearly and know what you're asking for. ## Path 1: Do It Yourself with AI Help These steps take about 20 minutes to prepare and 10–15 minutes for the actual call. Pull up your most recent bill — paper, PDF, or your online account. Write down: - The exact monthly amount you're charged (before fees and taxes) - The name of your current plan or package - When your current contract or promotional period started Most billing statements list the plan details near the top. If you're not sure when your promo rate expires, call billing and ask, or look for the welcome email from when you signed up. Before you call, find out what the company is advertising to new customers right now. This is your strongest piece of information. Open ChatGPT (or any AI assistant) and type: > What are [your provider's name] current promotional rates for new internet or cable customers in the US? What plans and speeds are they advertising this month? The AI will give you a general picture, though prices change frequently. Cross-check on the provider's website by opening an incognito window — that's what a new customer would see. Note the price and the plan name. Now draft what you're going to say. Ask ChatGPT: > I've been a customer of [provider] for [X years]. My current bill is $[amount] per month for [plan]. New customers are being offered [plan] at $[price] per month. Write me a brief, polite 90-second script I can use to call and ask for a better rate. Include what to say if they ask why I'm calling and how to bring up the new-customer offer. Edit the result so it sounds natural when you say it out loud. The goal is a script you can follow without reading it word-for-word — a few key sentences, not a full speech. Retention agents are trained to handle bill-reduction requests. They'll push back. The most common responses are: "I don't see any promotions available for your account," "That offer is only for new customers," and "I can look at what packages are available at a higher tier." Ask ChatGPT to help you practice: > Act as a cable company retention agent. I'm going to call to ask for a lower rate. Push back the way a real agent would, and I'll practice my responses. Run through it once or twice until you have a comfortable answer for each objection. The key responses are: "I understand, but I've been a loyal customer for [X years]," "I'm looking at what new customers are being offered right now on your website," and "If I can't get a better rate, I may need to look at switching to [competitor]." Call the billing or customer service number on your bill — not a general help line. When the agent answers, say you're calling about your bill and are considering your options. If they can't help, ask specifically to be transferred to the retention or loyalty department. State your case calmly: how long you've been a customer, what you're currently paying, and what you've seen offered to new customers. Mention a competitor if one is genuinely an option for you. Ask directly: "Is there anything you can do to bring my rate closer to what new customers are paying?" If the first agent says no, thank them and call back another time — a different agent may have more flexibility. Set a calendar reminder for 11 months after you start any new promotional rate. By month 11 you can call and ask about renewal options before the price increase hits — which puts you in a better position than calling after you've already been paying the higher rate for a few months. ## Path 2: Let an AI Service Make the Call for You If the idea of negotiating on the phone makes you uncomfortable, or you just don't have time, services like Pine (19pine.ai) handle the entire process on your behalf. Pine's annual plan starts at around $26 a month (billed as $319 per year); month-to-month plans charge a 25% success fee on confirmed savings instead of a flat rate. It works by connecting to your billing account, identifying where you're overpaying, then calling the provider and negotiating directly. Customers report saving $300 or more per year on a single bill — though results vary, and the service doesn't work for every provider or every type of account. Negotiation services like Pine report a 93% success rate. When they don't succeed on the first try, they typically try again. The math works out for many people: if the service saves you $300 a year and you use the percentage-fee model, the 25% fee comes to about $75 — leaving you $225 ahead. Other services operate on a different model — they take a percentage of whatever they save you rather than charging a flat monthly fee. That's worth knowing upfront so you understand what you'll owe if they succeed. ## What to Watch Out For **Promotional rates have end dates.** If a retention agent offers you a new promotional rate, ask when it expires and what the price will be afterward. Write it down and set a reminder to renegotiate before it runs out again. **Cancelling may trigger early termination fees.** If you're still within a contract period, know what your contract says before you use cancellation as leverage. **Third-party services access your account.** Before using a service like Pine, understand what information you're sharing and how it's stored. Read the privacy policy. **Results aren't guaranteed.** Some providers are more willing to negotiate than others. If you have only one option in your area, your leverage is limited — but it's still worth asking. **Savings vary.** The "$300+ per year" figures you see advertised represent good cases. Your actual savings depend on your provider, your plan, your history as a customer, and what promotions are available when you call. ## What to Try Next If you want to keep building on this, [AI Message Templates for Difficult Conversations](/guides/ai-message-templates) covers how to use AI to draft other kinds of tricky requests — useful when calling isn't an option. And if you're looking at your overall spending, [Best AI Budgeting Apps](/guides/best-ai-budgeting-apps) can help you spot where else your bills might be quietly climbing. --- # AI Meal Planning: A Week of Dinners in 10 Minutes Source: https://aiforregulars.com/guides/ai-meal-planning Figuring out what to cook every night is one of those low-stakes decisions that somehow takes enormous mental energy. ChatGPT can take it off your plate — no pun intended. In about ten minutes you can have seven dinner ideas tailored to your family, a combined grocery list, and even a few full recipes ready to go. Here is how. The plan only works if it fits your real life. Before you type anything, decide: - How many people are you feeding? - Any dietary restrictions or allergies? (vegetarian, dairy-free, no shellfish, etc.) - Are there picky eaters and what do they absolutely refuse to eat? - How much time do you have on a typical weeknight? (20 minutes? 45 minutes?) - What is your rough weekly grocery budget? - Do you want variety or are repeating a meal or two okay? You do not need to write a formal list — you will just tell ChatGPT in plain sentences. Open ChatGPT and send a message like this one, filled in with your real details: ``` I need a week of dinner ideas for [NUMBER] people. Dietary needs: [vegetarian / no pork / gluten-free / none — whatever applies] Allergies to avoid: [nuts, dairy, shellfish, etc. — or "none"] Picky eaters: [e.g., "my 8-year-old refuses anything spicy or with mushrooms"] Weeknight cooking time: about [20 / 30 / 45] minutes per meal Weekend cooking time: up to [1 hour] if needed Rough weekly grocery budget: $[AMOUNT] Variety: [we like trying new things / we prefer simple familiar meals] Please give me Monday through Sunday dinners with a one-line description of each. ``` ChatGPT returns seven dinner suggestions in seconds. Scan them and move to the next step. You will probably like five or six of the suggestions and want to change one or two. Just say so: - "Can you replace Wednesday's pasta with something without cheese?" - "Thursday's stir-fry sounds good but I don't have a wok. What else could I make in a regular skillet?" - "Can you make Friday something the kids will actually eat? They love tacos." ChatGPT adjusts instantly. Keep going back and forth until the whole week feels right. Once the dinner plan is finalized, send this: ``` Now give me a combined grocery shopping list for all seven dinners. Group it by store section: produce, proteins, dairy, pantry staples, and frozen. If an ingredient is used in more than one meal, list it once with the total amount I need. Assume I already have basic pantry staples like salt, pepper, olive oil, and common spices. ``` You will get a clean, organized list you can screenshot or copy straight into your phone's notes app before you head to the store. I make a habit of checking my fridge before I send that shopping list request. I add one line: "I already have [eggs, half a block of cheddar, a can of chickpeas]." ChatGPT removes those from the list automatically. It sounds small but it keeps me from buying duplicates and wasting food. If a meal on the list is unfamiliar, ask for the full recipe: ``` Give me the full recipe for the lemon herb chicken from Monday, including exact ingredient amounts and step-by-step instructions. Serves 4 adults. ``` ChatGPT gives you a proper recipe — ingredients, amounts, and numbered cooking steps. You can ask it to simplify any step that seems complicated. Copy the finished meal plan and grocery list out of ChatGPT and paste them somewhere you will actually see them. Good options: - A note in your phone so you have the list in the store - A shared note with your partner (Apple Notes, Google Keep) - A printed page stuck to the fridge ChatGPT does not remember your plan between sessions, so get it out of the chat window before you close it. When Sunday rolls around, starting over takes about two minutes. Open a new chat and paste: ``` Give me a different week of dinners using the same rules as before. [Paste your original constraints again here.] No repeats from last week — last week we had [list the 7 meals]. ``` You build up a rotating collection of family-approved dinners over a few weeks, and meal planning becomes almost automatic. ## What to try next Knowing what you will cook each week makes it much easier to stick to a food budget. The [ChatGPT monthly budget guide](/guides/chatgpt-monthly-budget) shows how to use ChatGPT to set spending targets across all your categories, not just groceries. And if you want to put ChatGPT to work on a bigger project, the [vacation planning guide](/guides/plan-vacation-with-chatgpt) walks through building a full trip itinerary the same way. --- # Use AI to Find Errors in Your Hospital Bill (Most Big Bills Have Them) Source: https://aiforregulars.com/guides/ai-medical-bill-errors Hospital bills are notoriously hard to read — columns of unfamiliar codes, line items that seem wildly expensive for everyday supplies, and totals that feel impossible to verify. Most people just pay, or set up a payment plan, assuming the hospital must have gotten it right. But Medical Billing Advocates of America estimates that three in four bills over $10,000 contain errors, averaging around $1,300 each. AI can now help you read that bill line by line, explain what each code means, and flag the most common types of mistakes — in roughly the time it takes to make a cup of coffee. In a case widely covered in 2026, a family used Claude to review a $195,000 hospital bill and ultimately reduced it to around $33,000. The AI flagged duplicate charges, an outpatient procedure that had been billed at inpatient rates, and supply costs running 500–2,300% above what Medicare would pay for the same items. That result is not typical — but the underlying errors it found are. Most hospitals send a summary bill: a single page with a few broad categories and a total. That summary is almost useless for finding errors. What you need is the itemized bill — a line-by-line breakdown of every service, supply, and medication, with the CPT (Current Procedural Terminology) code next to each item. CPT codes are five-digit numbers that identify specific medical procedures and services. They're publicly searchable, which means you can verify exactly what was billed. Without them, you're looking at descriptions like "medical/surgical supplies — $3,200" with no way to check what's actually inside. Call the hospital's billing department and ask for an "itemized bill with CPT codes." You have a legal right to this document in the US. If the billing department pushes back, ask to speak with a patient advocate. Before uploading anything to an AI tool, black out the following — in a PDF editor or simply with a marker on a printed page before scanning: - Your full name and date of birth - Your Social Security number - Your health insurance member ID and group number - Your hospital account number The AI needs the CPT codes, dates of service, quantities, and dollar amounts. It does not need anything that identifies you personally. Free AI tools handle conversation data in various ways; medical billing information is sensitive enough that it's worth taking sixty seconds to clean the file first. Upload your redacted bill to ChatGPT (a Plus subscription allows file uploads) or to Claude at claude.ai — both handle PDFs well. If you don't have a subscription, paste the bill's text content directly into the chat window. Start with this prompt: > "I'm going to share an itemized hospital bill. Please review it for common billing errors, including: duplicate charges (the same procedure billed more than once), upcoding (a more expensive code than the procedure likely warrants), services that seem inconsistent with the type of visit, and supply costs that appear unusually high. For each potential issue you find, explain what the CPT code means in plain English, why it looks suspicious, and what question I should ask the billing department." Let the AI work through the full bill before you start asking follow-up questions. AI explains codes well, but you'll catch more if you know what to look for: **Duplicate charges** — the same procedure or supply appears more than once for the same visit date. One of the easiest errors to catch. **Upcoding** — a procedure is billed under a more expensive CPT code than what was actually performed. A routine check-up billed as a complex consultation, for example. **Inpatient vs. outpatient status errors** — if you were treated as an outpatient but billed at inpatient rates (or vice versa), the cost difference can be enormous. This was one of the central errors in the $195,000-to-$33,000 case mentioned above. **Inflated supply costs** — hospitals mark up supplies, but some charges are far outside normal ranges. Checking a charge against Medicare's published rates gives you a useful benchmark. **Services you never received** — procedures or medications on the bill that don't match your memory of the visit. **Unbundling** — a single procedure that should be billed under one CPT code is split into several separate codes, each charged individually, adding up to more than the bundled rate. After the AI reviews the bill, ask it to create a table: one row per suspicious line item, with the CPT code, what it means, why it's potentially wrong, and the exact question to ask the billing department. That table becomes your checklist for the phone call — you're not guessing, you're working from a list. Once you have a list of potential errors, ask the AI to help you write a formal dispute letter: > "Based on the errors you identified, please help me write a professional dispute letter to the hospital billing department. Include: a summary of each error with the specific CPT code and line item, what I believe is incorrect and why, and a request for a written response within 30 days. Keep the tone factual and professional." Billing departments handle disputes regularly. A specific, well-organized letter — with CPT codes cited and exact dollar amounts referenced — gets taken seriously faster than a frustrated phone call. Send it by certified mail or email with delivery confirmation so you have a record. Even after errors are corrected, the remaining balance may be negotiable — especially if you're uninsured or underinsured. Most hospitals have financial assistance programs and room to adjust bills that few patients know to ask about. Approaches that often work: - Ask about a **prompt-pay discount** if you can pay a lump sum upfront - Ask whether the bill can be adjusted to **Medicare rates** — hospitals often agree - Ask for the **self-pay or uninsured rate**, which is typically much lower than the chargemaster list price AI can help you prepare for these conversations too. Try: "Help me write talking points for negotiating my remaining hospital balance of [amount] as an out-of-pocket patient." AI is a useful first pass. A professional medical billing advocate brings something it can't — experience with specific hospital billing departments, knowledge of local billing norms, and the ability to review your insurance policy alongside the bill. Consider hiring one when: - The bill is over $50,000 and the errors are complex - The hospital isn't responding to your dispute letters - Your insurer and the hospital are in a disagreement about coverage - You're dealing with a surprise out-of-network bill after an emergency You can find certified medical billing advocates through the Medical Billing Advocates of America at billadvocates.com. Most work on contingency — they take a share of what they save you, with no upfront fee. ## What to Watch Out For **Protect your privacy first.** Do not upload an un-redacted bill containing your name, Social Security number, or insurance ID to any free chatbot. The CPT codes and dollar amounts are what the AI needs; your personal identifiers are not. **AI explains; it doesn't know your actual care.** The AI can tell you that a particular CPT code looks inconsistent with an outpatient visit, but it doesn't know whether you actually received that service. Your own memory and any notes from your visit are essential for deciding which flags are real errors and which are things you simply don't remember. **Specialized tools go further on pricing.** General AI tools like ChatGPT and Claude are excellent at explaining codes and spotting duplicates. For detailed pricing comparisons against Medicare benchmarks, specialized tools — such as CoveredUSA's bill analyzer — are built specifically for that analysis. **A dispute does not pause your due date automatically.** When you send your letter, also ask the billing department in writing to hold collection activity while they review. Keep copies of everything. Most hospitals will agree to pause collection during a good-faith dispute, but you need that in writing. ## What to Try Next The skill you just learned — uploading a confusing document and asking AI to explain it — applies far beyond hospital bills. [How to Upload a PDF and Ask ChatGPT Questions About It](/guides/upload-pdf-ask-chatgpt) covers the mechanics of document uploads, and [How to Use AI to Understand a Contract Before You Sign](/guides/ai-explain-contracts) puts the same approach to work on legal paperwork. --- # Don't Ask AI About Your Medications Until You Read This Source: https://aiforregulars.com/guides/ai-medication-advice-danger Your doctor just added a third medication to your morning routine, and you're wondering if it's safe to take alongside the two you already have. Asking ChatGPT feels faster than waiting on hold for the pharmacy. But before you type that question, there is something you should know: a 2023 study found that 74% of the medication information ChatGPT provided was incomplete or outright incorrect when reviewed by pharmacology experts. The errors weren't minor rounding differences. They were the kind that matter. That's the core problem with AI and medications. It's not that AI is always wrong — it gets some things right. It's that it's wrong in ways that sound exactly like it's right, and on questions where the stakes are too high to guess. ## What AI Is Actually Good For With Medications There are real things AI can help with when it comes to medications — just not the ones most people reach for it to answer first. **Understanding what a drug class does.** If your doctor prescribed a "beta-blocker" or an "SSRI" and you want to understand what that category of drug does before your next appointment, AI is a reasonable starting point. You're asking about general pharmacology, not a personal dosing decision. **Translating label language into plain English.** Medication leaflets are written for regulatory compliance, not for patients. Asking an AI "what does 'take with food to avoid gastric irritation' actually mean in practice?" is a question with a fairly stable general answer. For a more complete look at reading your prescription label, [AI and Your Prescription Label](/guides/ai-prescription-explainer) walks through this in detail. **Preparing questions for your pharmacist or doctor.** This is one of the smartest uses of AI in a medical context. Use it to generate a list of questions to ask your pharmacist at your next visit. > "I'm starting a new blood pressure medication called lisinopril. What are the most important things I should ask my pharmacist about taking it safely?" Bring the list to the pharmacy. The pharmacist answers them; AI helped you think to ask. ## Where AI Medication Advice Becomes Dangerous These are the categories where AI answers sound confident, feel helpful, and can seriously hurt you. **Dosing questions.** "How much ibuprofen should I take?" sounds simple. But the right answer depends on your kidney function, your other medications, whether you have a history of stomach ulcers, your age, and your weight. AI does not know any of that — and it will answer anyway, often correctly in a generic sense, which is exactly what makes it dangerous for your individual situation. **Drug-interaction checks.** This is the highest-risk category for older adults. Seniors take an average of four to five prescription medications; the risk of a harmful interaction grows with every additional drug added. Correctly checking interactions requires current, complete drug databases — not a language model trained on general text. AI can miss a known interaction, or invent a warning that sounds plausible but doesn't exist. **"Is it safe to take X with Y?"** This includes over-the-counter medications, supplements, and even common foods. Grapefruit, for example, is known to interact dangerously with certain heart and blood pressure medications. AI has been shown to give inconsistent answers to the same interaction question asked different ways, and it cannot account for your full medication list. **Deciding to stop or change a medication.** If you're having side effects and considering reducing your dose or stopping something, AI should not be part of that decision. Stopping certain medications abruptly — blood pressure drugs, antidepressants, corticosteroids — can cause serious health events. A pharmacist can tell you in minutes whether tapering is needed; AI cannot do this safely. When I have a medication question, I call the pharmacy and ask to speak to the pharmacist directly — not the counter staff. Pharmacists are the most accessible clinical experts most people have free access to. No appointment, no waiting room, and they're required by their professional license to give personalized, accurate advice. Most people don't realize this call is free. ## Why AI Gets Medication Information Wrong The 74% incomplete-or-incorrect finding isn't a surprise when you understand what language models actually do. They are trained to produce text that sounds helpful and correct — which is a different skill from retrieving verified pharmaceutical information. **AI invents citations.** Ask an AI chatbot to back up a medication claim with a scientific study, and it will frequently produce a completely made-up citation: a real-sounding journal name, a plausible author, a volume and page number that don't exist. This is called hallucination, and it is well-documented. A pharmacist who cites a study actually read it. An AI is pattern-matching to what a citation looks like. **Drug information changes.** New interactions are discovered regularly. Dosing recommendations are revised. Black-box warnings are added to medications that were previously thought safe at certain doses. AI has a training cutoff and no mechanism for pulling current prescribing information. **No AI chatbot holds a pharmacy license.** In every jurisdiction, dispensing medication advice is a regulated, licensed activity with professional accountability. No AI chatbot is licensed anywhere in the world, which means its advice carries no accountability if something goes wrong. ## Real-World Consequences By late 2025, multiple lawsuits had been filed in different jurisdictions alleging that AI chatbot advice contributed to patient deaths or serious harm — including a lawsuit against OpenAI brought by the parents of a 19-year-old who died of a drug overdose in 2025, alleging ChatGPT advised him it was safe to combine two substances that turned out to be a fatal combination. These are not edge-case hypotheticals. They involve real people who asked what seemed like a reasonable medication question and received a confident, detailed, wrong answer. It is also worth noting that AI chatbot developers themselves warn against using their products for medical decisions. The gap between those warnings and what millions of people actually use these tools for every day is enormous — and the people most likely to be hurt by that gap are older adults managing multiple medications, which is exactly the population that has embraced AI chatbots most enthusiastically for practical questions. ## What to Use Instead **Your pharmacist.** Free, no appointment needed. Call your local pharmacy and ask to speak to the pharmacist. Pharmacists are trained to check interactions against real drug databases, advise on dosing, explain side effects, and flag risks specific to your situation. Many major chains also offer online or phone chat. This is the right first call for nearly any medication question. **MedlinePlus (medlineplus.gov).** The US National Library of Medicine's patient information site. Plain-language drug information that is verified and updated. Free, available in English and Spanish. **Drugs.com.** A widely used drug interaction checker, run on a US database. Enter your medications and see a clear summary of known interactions. Free for patients — if you're outside the US, check the active ingredient rather than the brand name, since the same drug can be sold under a different brand (acetaminophen is sold as Tylenol in the US and as paracetamol or Panadol in the UK, for instance). **Your doctor's nurse line.** For questions about changing, stopping, or adjusting a prescription, a message to the nurse line at your doctor's office usually gets a same-day response. ## What to Watch Out For A few additional risks beyond the accuracy problem: **Privacy.** When you type your medications and health conditions into a chatbot, that information may be stored and used to train future models, depending on the platform's privacy policy. Check before you share. **The tone of confidence.** AI chatbots answer with authority regardless of how uncertain the underlying information is. A pharmacist will say "let me look that up" — an AI almost never will. **Using AI to second-guess your doctor.** There's a difference between wanting to understand your treatment and asking an AI whether your doctor's recommendation is correct. AI is not a second opinion from a clinician. For decisions about your actual treatment, talk to a clinician. ## What to Try Next If you already have a prescription and want to understand what it says, [AI and Your Prescription Label](/guides/ai-prescription-explainer) covers what's safe — and useful — to ask AI about. If a symptom sent you looking for answers first, [When You Can and Can't Ask AI About Symptoms](/guides/ai-health-symptoms) covers those limits clearly. ## Sources - [Appropriateness of ChatGPT as a resource for medication-related questions](https://pubmed.ncbi.nlm.nih.gov/39096130/) — British Journal of Clinical Pharmacology / PubMed (Grossman et al., source of the 74% figure) - [Free ChatGPT may incorrectly answer drug questions, study says](https://www.cnbc.com/2023/12/05/free-chatgpt-may-incorrectly-answer-drug-questions-study-says.html) — CNBC - [Drug information](https://medlineplus.gov/druginformation.html) — MedlinePlus, U.S. National Library of Medicine - [Usage policies](https://openai.com/policies/usage-policies/) — OpenAI - [Their son died of a drug overdose after consulting ChatGPT. Now they're suing OpenAI.](https://www.cbsnews.com/news/open-ai-chatgpt-drug-overdose-lawsuit/) — CBS News --- # AI Journaling Apps for Stress and Anxiety: What They Help With (and What They Don't) Source: https://aiforregulars.com/guides/ai-mental-health-journaling It is 3am and your mind will not stop. You have tried reading, you have tried breathing, and now you are staring at your phone wondering whether to write down whatever is chasing you. An AI journaling app is not a therapist — but it is awake, it will not judge you, and it might help you get what is spinning in your head onto a page and into perspective. That is what these apps are built for: not to replace professional care, but to give you a low-barrier way to process stress, notice patterns in your mood, and learn a few practical coping techniques. Here is what they genuinely offer, where they fall short, and how to use them without making the common mistakes. ## What AI Journaling Apps Actually Do The useful ones do a few specific things well. **Prompted reflection.** Instead of staring at a blank page, you get a question: "What's one thing that felt heavy today, and what made it feel that way?" That structure makes it easier to start — and the process of answering, even imperfectly, often helps more than the answer itself. **Mood tracking.** Most apps ask you to rate how you are feeling, tag what was going on, and log this over time. After a few weeks, patterns become visible that are invisible when everything lives inside your head. You might notice you are consistently lower on Sunday evenings, or that the same kinds of situations keep showing up on hard days. **CBT-style exercises.** Cognitive behavioral therapy is a well-researched approach that examines the connection between your thoughts, feelings, and behaviors. Several AI apps translate core CBT techniques into short guided conversations — things like spotting catastrophic thinking, questioning the evidence behind a worry, or identifying small actions you could take. These are not a replacement for CBT with a trained therapist, but they are useful tools you can reach for at any hour. **Something to talk to at 3am.** This is not nothing. A lot of the value people describe from these apps is not any specific technique — it is simply having a patient, non-judgmental space to say what is going on, without waking anyone up or worrying about what they will think. These apps work best for ordinary day-to-day stress: work pressure, relationship friction, sleep troubles, low-grade worry. If you are looking for apps designed specifically for social connection and loneliness rather than stress management, [AI Companion Apps for Seniors](/guides/ai-companions-seniors) covers that related category. ## The Apps People Use Most **Woebot** (woebot.io) was one of the best-known mental-wellness chatbots, built by clinical psychologists and using CBT and DBT techniques to guide structured conversations. It shut down its consumer app on June 30, 2025 and is no longer available to individual users. The company now operates exclusively as an enterprise platform for healthcare organisations and employers. For a free alternative with a similar approach, plain ChatGPT can guide structured conversations at no cost. **Reflectly** is a guided journaling app with AI-personalized prompts. It costs around $9.99 per month on iOS (pricing varies by platform). More journal-like than a chatbot approach: you write, and the app follows up with questions. It includes mood tracking over time. **Day One** is a well-established journaling app that has added AI features to help you write, reflect on the day, and find patterns in older entries. It is not primarily a mental-wellness app, but many people use it that way. When I am stuck in a worry loop at night, I open a simple notes app and write one sentence: "Here is what I am afraid will happen." Then I write another: "Here is what I would do if it did." That two-sentence exercise — a simplified version of what CBT apps do — takes about five minutes and usually changes the texture of the anxiety enough to let me sleep. You do not need a dedicated app to start. ## Using Plain ChatGPT as a Journaling Partner You do not need a dedicated app. ChatGPT and similar AI assistants can serve as an on-demand journaling partner, available any time, capable of going deeper on whatever you bring up. The difference is that you drive the session yourself — there are no daily prompts, no mood tracker, and no history that carries over between conversations unless you save it. Prompts you can copy and use right now: > "I'm feeling anxious about [situation]. Help me work through this using CBT — ask me to examine the evidence for and against my worry, and suggest what I might tell a friend in the same situation." > "I've been feeling stressed lately but I can't quite name why. Ask me some questions to help me figure out what's actually going on." > "Walk me through a short breathing or grounding exercise. I'm feeling overwhelmed right now." > "I want to do a brief mood check-in. Ask me three questions about how I'm feeling and what's going on, then give me a short reflection and one small thing I could try today." One thing worth knowing: ChatGPT is designed to respond carefully when you describe significant distress. It will suggest professional help if what you describe sounds like a crisis — which is the right response, not a limitation. ## What to Watch Out For **Privacy matters more than you might think.** These apps ask you to share things you would not tell most people. That information lives on someone else's servers. Before you start, take five minutes to read the privacy policy: does the company share your data with third parties? Can they use your entries to train future AI models? How do they handle law enforcement requests? For a fuller look at what to keep private, [What Not to Tell an AI Chatbot](/guides/what-not-to-tell-ai-chatbot) is a short, practical read. A useful rule of thumb: treat AI journal entries the way you would treat a public diary. Anything you would not want a stranger to read — crisis thoughts, relationship details, identifying information about other people — keep off the app. **AI is not a diagnosis.** These apps can reflect your feelings back to you and teach you useful techniques. They cannot tell you whether what you are experiencing is generalized anxiety disorder, depression, OCD, or something else with a clinical name. If you are wondering whether what you feel has a diagnosis, that conversation belongs with a professional. **The app does not know what it does not know.** A skilled therapist notices when you are avoiding something, reads between the lines, and remembers what you said three months ago. An AI journaling tool does none of these things — unless the app specifically builds in long-term memory, which most do not. Useful daily tool; not a replacement for care that adapts to you over time. **Substituting instead of supplementing.** The main risk with AI wellness apps is not that they cause harm — it is that they can make someone feel like they are doing something about their mental health while the underlying issue goes unaddressed. Use these apps for daily habits and ordinary stress. Do not use them instead of getting professional help when you genuinely need it. ## When to Reach for a Human Instead An AI journaling app is a reasonable tool for processing everyday stress. It is not the right tool when: - Your anxiety or low mood has been affecting your sleep, work, or relationships for more than two weeks - You are having thoughts of harming yourself or others - You are using alcohol or substances to manage how you are feeling - You do not feel like you can keep yourself safe **In the US, call or text 988.** The 988 Suicide and Crisis Lifeline is available 24 hours a day, seven days a week — not only for people who are suicidal, but for anyone in serious emotional distress. If you are outside the US, a quick web search for "[your country] crisis line" will find a local number. Most countries have one, and most are free. A crisis line is not a failure. It is the right tool for a different kind of moment. A 2025 systematic review and meta-analysis published in JMIR found significant short-term reductions in anxiety and depression symptoms across clinical trials of AI mental-health chatbots — encouraging evidence that these tools can help with genuine distress, not just mild stress. The same research, and a parallel analysis in the Hastings Center Report, also noted that independent oversight of these apps is still catching up to how widely they are being used. Using them thoughtfully, and knowing their limits, is something each person has to decide for themselves. ## What to Try Next If loneliness is part of what you are navigating, [AI Companion Apps for Seniors](/guides/ai-companions-seniors) covers a related category of tools built for social connection — worth reading to understand how the two kinds of apps overlap and differ. Before you start sharing personal details with any AI, [What Not to Tell an AI Chatbot](/guides/what-not-to-tell-ai-chatbot) has a short, practical list of what belongs off the record. --- # How to Politely Decline, Complain or Negotiate — AI Message Templates Source: https://aiforregulars.com/guides/ai-message-templates Some messages are genuinely hard to write. You know what you want to say, but you don't want to sound rude, passive-aggressive, or too aggressive. You rewrite the same email four times and it still doesn't feel right. AI tools are particularly good at this. They draft polished, appropriately toned messages fast — and you can edit the result to sound like you. Below are ready-to-use prompts for the messages most people dread. ## How to use these prompts Copy the prompt below, paste it into ChatGPT (or any similar AI chat tool), fill in the brackets with your actual details, and send. Read the result, edit anything that doesn't sound like you, and send your message. The more specific you are about the situation, the better the draft will be. Whether it's a social event, a volunteer commitment, or a work task you don't have capacity for, a clean, warm "no" is hard to write. Use this prompt: ``` Write a short, warm message declining an invitation to [event or request]. I want to sound genuine, not make up an excuse, and leave the relationship intact. The person is [my colleague / a friend / my neighbor — choose one]. Keep it to 3-4 sentences. ``` If you want to leave the door open for future invitations, add: *"Include a brief line that I'd love to next time."* If you want a firm close, add: *"Don't leave the door open — I just want to decline cleanly."* Complaints work best when they're specific, calm, and clear about what you want to happen. Emotional messages often get slower responses. Use this prompt: ``` Write a complaint email to [company name] about [the problem]. I [bought / ordered / hired] [product or service] on [approximate date]. The problem is: [describe what went wrong]. I want [a refund / a replacement / an apology and explanation — choose what you want]. The tone should be firm and professional, not apologetic. Include a clear request in the last paragraph. ``` Refund requests often fail because they're too vague or too emotional. This prompt keeps it clear: ``` Write a refund request email for [product or service] purchased from [company]. The reason for the refund is [brief explanation]. I have [proof of purchase / a receipt / an order number — add what you have]. I'd like a full refund of [amount if known]. Keep the tone polite but direct, and end with a specific deadline for a response — [7 business days / 14 days]. ``` If you've already been refused once, add: *"This is a follow-up — I was previously told [what they said]. Acknowledge that and escalate the tone slightly."* I used an AI complaint prompt after an online order arrived damaged and customer service had already offered me a partial refund I didn't think was fair. The message it helped me write was firm without being confrontational. I got a full refund within two days — I think in part because the email was specific about what I wanted and gave them a clear path to resolving it. This is one of the messages people rehearse in their heads for weeks and still don't send. A clear, confident prompt: ``` Write an email to my manager asking for a salary increase. I've been in this role for [time period] and my responsibilities have [grown / changed — describe briefly]. I believe my current salary of [amount, optional] should be [your target or 'reviewed']. I have [specific achievements or context to mention]. The tone should be confident and professional, not apologetic. I want to open a conversation, not make an ultimatum. ``` If you're negotiating a job offer rather than an existing salary, change the prompt to: *"Write a counter-offer email for a job offer I received. The offer was [amount]. I'd like to counter at [your target]. Keep the tone enthusiastic about the role while being clear about the ask."* Canceling a service, ending a contractor relationship, or stepping back from a commitment is easier when the message is clear and doesn't over-explain: ``` Write a short, professional email ending my [subscription / contract / working relationship] with [company or person]. The reason is [brief honest reason, or 'I'll keep this private']. I want to wrap up by [date]. Keep it respectful and to the point — no need for a long explanation. ``` If you've had a difficult experience and want to mention it, add: *"I'd like to briefly note that [the issue] and keep it factual without sounding angry."* Sometimes the hard message isn't a complaint or a refusal — it's a correction. A colleague misrepresents your work, someone assumes something that isn't true, or a relationship has drifted into territory you're not comfortable with: ``` Write a message that politely but clearly corrects a misunderstanding. The situation is: [describe what happened and what the other person thinks]. The truth is: [what actually happened or what you want them to understand]. The relationship is [professional / friendly — choose]. I want to fix the record without making it a conflict. Keep it brief — 4-6 sentences. ``` ## What to try next These templates get you started, but knowing how to shape a prompt for any situation makes you faster and more flexible. The guide on [how to write better AI prompts](/guides/how-to-write-prompts) covers the principles behind what makes a prompt work well. If you need to write professional emails in a work context more broadly — not just difficult ones — [AI prompts for office work](/guides/ai-prompts-office-work) has a full set of examples. --- # 10 Things People Believe About AI That Are Simply Not True Source: https://aiforregulars.com/guides/ai-myths-debunked AI has been in the news so much that a layer of myths has built up around it — some alarming, some overly optimistic, and some just confusing. This guide goes through ten of the most common beliefs people have about AI and explains, calmly and clearly, what's actually true. ## Myth 1: AI Is Always Right **What people believe:** If the AI said it, it must be accurate — it has access to all that information. **What's actually true:** AI makes mistakes regularly, and some of those mistakes are confident-sounding and completely wrong. The technical term is "hallucination." AI produces text that *sounds* right based on patterns, but it has no built-in fact-checker. It doesn't know when it's wrong. For anything important — a medical question, a legal matter, a financial decision — always verify AI output with a separate, reliable source. ## Myth 2: AI Is Thinking and Has Feelings **What people believe:** AI is a mind inside a machine. It reasons, understands, and might even feel something when you interact with it. **What's actually true:** AI language models work by predicting which words should come next, based on patterns learned from training data. They are sophisticated pattern-matching systems. When ChatGPT seems empathetic or curious, it's producing text that fits the pattern of an empathetic or curious response — not experiencing those states. This isn't a criticism. It's just what the technology is and how it works. ## Myth 3: AI Is Going to Take Over the World **What people believe:** AI is becoming so powerful that it will eventually control humanity — or is already subtly doing so. **What's actually true:** Current AI tools are very good at specific, narrow tasks: generating text, recognizing images, making recommendations. They don't have goals. They don't want things. They don't make plans. The idea of an AI with human-like ambitions and self-preservation instincts is a plot device from science fiction — entertaining, but not a description of how today's AI actually works. ## Myth 4: AI Knows Everything **What people believe:** AI has read the whole internet, so it must know every fact. **What's actually true:** AI training has a cutoff date — the model learned from data up to a certain point and doesn't automatically know about things that happened after that. It also has significant gaps in specific, niche, or local information. And even on topics it was trained on, it gets things wrong. Knowing a lot is not the same as knowing everything, and even "knowing a lot" doesn't guarantee accuracy. ## Myth 5: AI Can See and Hear You All the Time **What people believe:** AI is surveillance — it's watching through your camera and listening through your microphone without permission, all the time. **What's actually true:** AI chat tools only have access to what you explicitly type or speak during an active session. They don't run in the background. They can't access your camera or microphone without a permission prompt from your device, exactly like any other app. Standard device privacy controls apply. ## Myth 6: Using AI Is Cheating **What people believe:** If you used AI to help write something or figure something out, that's dishonest — you didn't really do it. **What's actually true:** This depends entirely on context. Using AI to help draft a work email, brainstorm ideas, explain a concept, or organize your thoughts is no more "cheating" than using a spell-checker, a calculator, or a thesaurus. Using AI to submit schoolwork as entirely your own, in a setting where that's explicitly not allowed, is a different question — one about following agreed-upon rules, not about AI itself. One myth I hear very often is that using AI means you didn't really do the work. I think of it more like using a good dictionary or asking a knowledgeable friend for a second opinion. The thinking, the judgment about what to keep and what to change, and the responsibility for the final result still belong to you. ## Myth 7: AI Will Replace Everyone's Jobs in a Few Years **What people believe:** It's only a matter of time before AI makes most human workers unnecessary. **What's actually true:** AI is changing certain tasks within certain jobs — particularly repetitive, text-heavy, or pattern-based work. But most jobs involve bundles of many tasks, and AI handles only some of them well. Jobs requiring physical presence, trusted relationships, ethical judgment, and fast improvisation in unpredictable situations are much harder for AI to replace. The picture is complex, the changes are gradual, and no one can give you a reliable timeline. ## Myth 8: AI Is Too Complicated for Regular People **What people believe:** You need a technical background to get anything useful out of AI tools. **What's actually true:** The most widely used AI tools today are deliberately conversational — you talk to them in plain English. No coding required. No settings to configure. If you can type a question or speak one out loud, you can use an AI chatbot. The learning curve is genuinely gentle, and most people feel reasonably comfortable within their first session. ## Myth 9: The AI in Movies Is What Real AI Looks Like **What people believe:** AI is like HAL 9000, the Terminator, or robotic characters in science fiction — physically capable, goal-driven, and potentially dangerous on its own terms. **What's actually true:** Real AI in 2026 is mostly text on a screen. It answers questions, writes drafts, and generates images. It doesn't have a body, goals, or self-preservation instincts. Hollywood AI is designed to be dramatic. Real AI is designed to be useful. The gap between the two is very large. ## Myth 10: AI Knows When It's Making Something Up **What people believe:** If the AI doesn't flag something as uncertain, it must be confident for a good reason. **What's actually true:** AI often produces incorrect information with complete confidence. It doesn't have an internal fact-checking process, and it doesn't feel uncertain the way a person does when they're guessing. Newer AI systems are better trained to say "I'm not sure" — but this is still far from reliable. The absence of a disclaimer is not evidence that the information is correct. This is why verifying anything important remains the right habit regardless of how certain the AI sounds. **What to try next:** For a deeper look at why AI produces confident mistakes, [Why Does AI Make Things Up?](/guides/why-ai-hallucinates) explains the mechanism clearly and gives you three habits to catch errors. And if you want a clear explanation of what different AI tools actually are, [What Does 'AI' Actually Mean?](/guides/what-does-ai-mean) breaks it down without jargon. --- # AI Note-Takers Compared: Otter, Fireflies, and Zoom AI Companion Source: https://aiforregulars.com/guides/ai-note-taker-comparison Taking notes during a meeting is hard. You're trying to listen, contribute, and write at the same time — and something always gets missed. AI note-takers solve this by transcribing the meeting automatically, then summarizing the key points and action items. The three most widely used tools are Otter.ai, Fireflies.ai, and Zoom AI Companion. They solve the same core problem but work quite differently. Here's what each one actually does. ## A note on recording etiquette Before comparing tools, this needs to be said clearly: **always tell everyone in a meeting that you are recording it.** Most AI note-takers join calls as a visible bot — participants will see it — but a verbal heads-up at the start is both polite and legally required in many jurisdictions. For sensitive conversations (HR, legal, medical, client negotiations), check whether recording is appropriate before you start. ## The tools compared ### Otter.ai — Best all-around for individuals Otter joins your Zoom, Meet, or Teams calls as a bot and produces a live transcript as the meeting happens. After the call, it generates a summary, highlights key moments, and creates a list of action items. You can search across all your meeting transcripts for specific topics — useful when you need to find who said what three weeks ago. The free plan includes 300 transcription minutes per month and basic summaries. The paid plan raises that limit, adds integrations with tools like Notion and Slack, and lets multiple teammates share a workspace. Otter works best for: individuals who attend lots of meetings across different platforms and want one place to search all their notes. Limitations: the AI summaries are good but sometimes miss context that was implied rather than stated. Action items occasionally get attributed to the wrong person. Always review before sharing. ### Fireflies.ai — Best for teams with CRM or workflow integration Fireflies takes a similar approach — it joins as a bot, transcribes, summarizes, and extracts action items — but its strength is integrations. Fireflies connects directly with Salesforce, HubSpot, Slack, Notion, Asana, and several other business tools. After a sales call, for example, Fireflies can automatically log notes to a Salesforce contact record. The team workspace features are also more developed than Otter's. Managers can search across all team meetings, track whether action items from one meeting showed up in the next, and build a searchable meeting library over time. The free plan is more limited than Otter's for individual use. Fireflies really shines on a team plan. Fireflies works best for: sales teams, customer success teams, or any organization that wants meeting notes to flow automatically into existing tools. Limitations: more complexity than most individuals need. Setup requires more time, especially if you're connecting CRM integrations. ### Zoom AI Companion — Best if your team is already on Zoom paid Zoom AI Companion is built into Zoom and available to all Zoom accounts with a paid subscription at no extra charge. It produces meeting summaries, transcripts, and action items without any setup — it's just there when you need it. The advantage is simplicity. You don't need to add a bot, set up an account on a separate service, or manage another tool. The summary appears in Zoom's meeting recap after the call ends, and it's integrated with Zoom's team chat. The limitation is equally clear: it only works in Zoom. If your team uses Google Meet or Microsoft Teams for some meetings, Zoom AI Companion won't cover those. Zoom AI Companion works best for: organizations that already use paid Zoom and want the lowest-friction way to get meeting summaries. I've used all three of these tools at different points. The one thing that catches people off guard is the quality of action item extraction — all three tools are more reliable at transcription than they are at correctly identifying what needs to happen next. The meeting summary is useful; the action item list needs human review. Don't forward an AI-generated action list to a client or colleague without reading it first. ## Quick comparison | Tool | Works on | Best for | Free tier | Key extra feature | |---|---|---|---|---| | Otter.ai | Zoom, Meet, Teams | Individuals, cross-platform | Yes, limited minutes | Searchable archive | | Fireflies.ai | Zoom, Meet, Teams | Teams with CRM needs | Yes, more limited | CRM/workflow integrations | | Zoom AI Companion | Zoom only | Existing Zoom teams | Included with paid Zoom | Zero extra setup | ## Which one should you choose? Start with Otter.ai if you're an individual or small team who wants to try AI note-taking with minimal commitment. The free tier is enough to evaluate whether the concept works for you. Move to Fireflies if you're managing a sales or customer-facing team and want notes to flow into your existing tools. Choose Zoom AI Companion if your whole organization is on paid Zoom and you want the simplest possible solution. What to try next: if you want to understand what AI meeting features do more broadly across different platforms, the guide to [AI features in Zoom, Teams, and Meet](/guides/ai-features-zoom-teams) explains what each platform actually offers and how far to trust the automated summaries. --- # AI Wrote the Obituary — Is That Okay? A Gentle Guide to AI, Grief, and Etiquette Source: https://aiforregulars.com/guides/ai-obituary-condolence Death comes without warning, and it often leaves the people it touches without words. A grief-struck family member is asked to produce a complete, accurate life history of their loved one within hours of their passing — and increasingly, the funeral home quietly hands that task to an AI tool, sometimes without mentioning it at all. At the same time, more people are turning to AI when they themselves can't find the words to reach out to someone who is grieving. Neither of these is wrong, exactly — but both carry risks worth knowing. ## Part One: When the Funeral Home Uses AI A Washington Post investigation published in August 2025 found that funeral homes across the United States were using AI tools — including software called Passare — to draft obituaries at scale, often pulling from basic intake forms filled out while families were in shock. Families who later noticed errors reported finding wrong birth years, invented hobbies, and phrasing that bore no resemblance to who their loved one actually was. Most of those families said the funeral home had never mentioned that AI was involved. This doesn't mean AI-assisted obituary drafting is inherently bad — a compassionate first draft, produced quickly, can genuinely help a grieving family who doesn't know where to start. The problem is when the draft goes out without the family seeing it, or when errors make it into print or a permanent online memorial. You have more control here than you may realize. ### What to Ask the Funeral Home When you first sit down — in person or by phone — with a funeral home to arrange a service, it is completely reasonable to ask: "Will you be writing the obituary, or will we?" and "If your team drafts it, can we review it before it's published?" A funeral home that uses AI tools should be willing to show you the draft. If they're not, or if they've already published something you didn't approve, that's worth addressing directly. Most funeral homes depend on reputation and word of mouth — they genuinely want families to be satisfied, and will correct errors if you ask. ### How to Review an AI-Drafted Obituary When the draft comes to you, go through it in two passes. **First pass — facts only.** Check each of the following: - Full name, including middle name or maiden name - Birth date and place - Death date - Names of surviving family members (spouse, children, siblings, parents) - Names of family members who died before them (predeceased) - Where they lived, and for how long - What they did for work, and when they retired These are the details most likely to contain errors when an AI tool works from a brief intake form. A wrong birth year, a misspelled family name, or a missing sibling will matter enormously to the people who read this in the years ahead. **Second pass — tone.** Read the whole thing aloud. Does it sound like someone who knew this person? AI-drafted text tends to fall back on generic formulations — "devoted parent," "beloved by all," "will be deeply missed" — which are true of almost everyone and feel true of no one in particular. You have every right to change the language so it sounds like your family actually talking about this specific person. > Prompt you can use: "Help me revise this obituary draft so it sounds more like [Name]. Here are three things that were most true about them: [your list]. Can you suggest sentences that bring those in?" If writing the obituary from scratch feels overwhelming, start with just three things: one fact about where they came from, one thing they were known for doing, and one sentence about what they leave behind. That's enough to give AI something real to work with — and enough to make the result feel true when it comes back to you. ## Part Two: When You Can't Find the Words to Reach Out Sending a condolence message is one of the hardest kinds of writing there is. You want to say something true, something that acknowledges real loss — and grief has a way of jamming the words before they make it to the page. AI can help. Asking ChatGPT or Gemini to help you draft a condolence note is not cold or lazy — it's a practical way to get unstuck when you've been staring at a blank screen for twenty minutes. But the way most people do it (send the AI draft more or less as-is) often doesn't land the way they hope. ### Why AI Condolence Messages Often Feel Generic Etiquette writers have long noted that what makes a condolence message feel real is one specific, personal detail — something that shows you actually knew the person who died, or truly know the person who is grieving. "I still think about the way he laughed at his own jokes before he finished telling them" is more comforting than "Your loss is immeasurable." AI tends to produce the second kind of sentence. The people receiving condolence messages are especially attuned to this. Grief makes everything feel heightened — both the kindness that lands and the emptiness that doesn't. A message that reads like a template, even a technically warm one, can feel like no one really stopped to think about the person who died. ### The Middle Path That Works Use AI for the first draft. Don't send the first draft. After AI gives you something to work with, do this: 1. Read it and mark anything that's actually true about your relationship to the person or their loss. 2. Cut the generic phrases ("words cannot express," "in my thoughts and prayers" unless you genuinely mean it, "they are in a better place" unless you know the family's beliefs). 3. Add one sentence — just one — about a specific memory or quality. It doesn't have to be poetic. 4. Read it back to yourself. If it sounds like you, send it. If it still sounds like a template, try once more. > Prompt to try: "Help me write a condolence note for [Name], who lost her father last week. I knew him briefly — he coached my son's soccer team for two years and always stayed after practice to help kids who were struggling. I want to tell her I'm thinking of her without it sounding like a greeting card." That last line matters: telling AI what you want to avoid is often more useful than describing what you want. ### When It's Better to Call Some losses are close enough that a written message isn't quite right — a childhood friend, a sibling's spouse, someone you've known for decades. In those cases, don't let the difficulty of finding words become a reason to stay silent. A short, imperfect voicemail or a handwritten note that arrives a few days late is better than a polished AI message that arrived immediately but felt like nothing. If you're unsure whether your message hits the right tone before you send it, the tip on [checking your message's tone before you send it](/tips/check-text-tone-before-sending) can help — but that works best when the words in the message are already yours. ## What to Watch Out For **Factual permanence.** Obituaries often become part of the permanent record — newspapers archive them, genealogical databases index them, families save them. An error that seems minor ("born in Toledo" instead of "born in Dayton") can confuse family records for generations. Review carefully before anything is published. **AI filling in what it doesn't know.** Some AI drafting tools, when given sparse information, will plausibly complete details — a vague mention of "community involvement" might become "served on the school board" because that seemed likely. That's an invention, not a mistake, and it's the hardest kind to catch. Read the draft against what you actually know about the person, not just against what sounds plausible. **Religious and cultural language.** AI defaults to familiar phrases that may not match the family's beliefs or traditions. An obituary for a secular family that says "she has gone to her eternal rest," or a condolence note that promises "heaven's welcome" to a family of a different faith, can land poorly. You're allowed to take this language out. **Privacy.** Publishing an obituary puts personal information — names, relationships, locations — into public view. Some families prefer to omit surviving children's ages, home cities, or other details that might be useful to people who weren't meant to be part of the service. That's a reasonable call, and you don't need the funeral home's permission to ask for less detail. ## What to Try Next If writing about a life — whether an obituary, a eulogy, or something longer — has made you want to preserve more of your loved one's story, [Let AI Help You Write a Life Story or Family Memoir](/guides/ai-write-life-story-memoir) walks through how to approach that with care. And if you're looking for help with a different kind of sensitive message — an apology, a thank-you, something you've been putting off — [AI Message Templates: How to Start Messages You've Been Avoiding](/guides/ai-message-templates) covers the same principle of using AI to start and then making it yours. --- # ChatGPT Plus, Google AI Pro, or Claude Pro: Which $20 AI Plan Fits You? Source: https://aiforregulars.com/guides/ai-paid-plans-compared If you're thinking about paying for AI for the first time, you've probably noticed that three services land at nearly the same price: ChatGPT Plus at $20/mo, Google AI Pro at $19.99/mo, and Claude Pro at $20/mo. The names suggest they're direct competitors. But in practice they're built for different kinds of use, and picking the wrong one means paying for features you'll never open. Here's how to figure out which one is actually yours. ## Start Here: What the Free Tiers Already Give You Before you put in a credit card number, it's worth knowing how much AI you're already getting for nothing. **ChatGPT Free** gives you access to GPT-4o mini with a daily usage limit, which handles most everyday tasks: answering questions, drafting emails, summarizing short documents. You'll hit the cap if you use it heavily all day, and you won't get image generation, voice mode, or the most powerful GPT-4o model — but for light use, it's genuinely capable. **Google Gemini Free** lets you use Gemini in your browser and integrates with Google Docs and Gmail at a basic level. Most people who want to draft a reply in Gmail or tidy up a document in Docs can do a reasonable amount without paying. Since October 9, 2026, though, free personal accounts only get Google's lightest model, Flash-Lite, so answers to harder questions can be shorter and less thorough ([what changed](/news/gemini-free-plus-model-access-cut-october-9)). **Claude Free** gives you access to Claude with a daily message limit and no file uploads. Responses tend to be longer and more thorough than the other free tiers, but heavy users will hit the cap quickly. The honest answer is that a lot of people considering a subscription don't actually need one. If you're a light user — a few questions a day, the occasional draft — try the free tier for two weeks first. Pay only when you keep running into real limits. ## ChatGPT Plus ($20/mo) — The Widest Toolset ChatGPT Plus is the right choice if you want AI to do more *kinds* of things, not just answer questions and write text. **What you get:** Access to GPT-4o (the full model, not the lite version), image generation via ChatGPT Images, voice mode for spoken conversation, the ability to use and create custom GPTs built for specific tasks, and scheduled tasks that let you set ChatGPT to do something automatically at a set time. Usage limits are also significantly higher than the free tier. **Who it's for:** People who want an AI that functions more like a toolbox than a text box — someone who might use voice mode while driving, generate an image for a presentation, use a custom GPT for recipe suggestions, and ask a text question, all in the same week. It's also the natural starting point if you've already been using ChatGPT Free and simply want more of what you already like. > "What's a healthy meal I can make in 30 minutes with chicken, rice, and whatever vegetables I have in the fridge?" ChatGPT Plus handles this kind of flexible, varied usage better than the other two. It's the most popular paid AI plan for a reason: it's the most general-purpose option available. If you want to go deeper on whether Plus specifically is worth it for your situation, [Is ChatGPT Plus Worth It?](/guides/chatgpt-plus-worth-it) covers the tradeoffs in more detail. ## Google AI Pro ($19.99/mo) — For the Google Power User Google AI Pro is the most interesting value proposition of the three — not because the AI itself is necessarily the strongest, but because of what comes bundled with it. **What you get:** Gemini Advanced (Google's most powerful model), deeper integration with Gmail, Docs, Drive, and Meet, and — this is the key part — **5TB of Google One storage**. **Who it's for:** People who live in Google's ecosystem. If your email is Gmail, your documents are in Google Docs, your photos are in Google Photos, and you were already paying (or about to pay) for extra Google Drive storage, Google AI Pro effectively cuts your net AI cost in half. You're getting an AI upgrade plus storage for roughly the same money you'd otherwise spend just on storage. The AI integration goes beyond a browser tab. Gemini in Gmail can help you draft replies based on prior conversation threads. Gemini in Docs can rewrite sections, adjust tone, and summarize long documents while you're already working in the document. If that's where you spend your working hours, the reduction in friction is real. **One caution:** If you don't use Google's tools heavily, the storage bonus doesn't help you, and you're paying purely for AI access — at which point ChatGPT Plus or Claude Pro may give you more. For a deeper comparison of how Gemini stacks up against ChatGPT on actual tasks, [ChatGPT vs. Gemini: Which Should You Use?](/guides/chatgpt-vs-gemini) is the full head-to-head. If you're already paying for Google One storage at any tier, check your billing page before subscribing to Google AI Pro separately. It's easy to end up paying for both — Google doesn't automatically merge or cancel your existing storage plan when you add AI Pro. Cancel your current Google One plan first, then subscribe to AI Pro, which will include the 5TB storage going forward. ## Claude Pro ($20/mo) — For Long Documents and Careful Writing Claude Pro is a narrower product than the other two, and that's not a criticism — it just means its advantages show up most clearly in specific situations. **What you get:** Access to Claude's most capable model, a 200,000-token context window (which fits roughly 150,000 words — a full novel or a stack of research papers — in a single conversation), priority access during peak hours, and higher daily message limits. **Who it's for:** People who regularly feed AI large amounts of text — contracts, legal documents, academic papers, long email threads, financial reports — or who use AI mainly for writing where accuracy and care matter. Claude's writing style tends to be more measured and less assertive than ChatGPT's; it hedges where things are uncertain, it doesn't invent information as readily, and it tends to say "I'm not sure" when it genuinely isn't. That style suits some tasks better than others. > "Here's a 40-page contract. What are the clauses I should be most careful about before signing?" That kind of task is where Claude Pro earns its price. Most AI assistants will tell you they can read a long document; Claude Pro is the one that can actually hold the whole thing in a single conversation without losing track of context. For a head-to-head comparison on writing and reasoning tasks, [Claude vs. ChatGPT: What's the Real Difference?](/guides/claude-vs-chatgpt) covers the specifics. ## The Premium Tiers ($100–200/mo): Most People Should Skip Them All three services offer much more expensive plans aimed at heavy professional users. **ChatGPT Pro** comes in tiers from about $100 to $500 a month (as of September 2026) and adds much higher usage limits, OpenAI's most powerful model and agent features, built for heavy professional work. **Google AI Ultra** sits at a similar premium price point. These plans exist for people who use AI as a core professional tool all day, every day — researchers, developers, analysts with intensive workloads. For regular everyday use, the $20 plans are already at the ceiling of what most people can realistically use. If you're considering a premium tier without a specific use case in mind, don't. Try the standard plan first and find out whether you actually hit its limits before spending $100–200 a month. ## A Simple Decision Flow Not sure which to pick? Work through this: **Do you already use Google products every day** — Gmail, Google Docs, Drive, Photos? Start with Google AI Pro, especially if you're already paying for Google One storage. **Do you want AI to do things beyond text** — generate images, use voice mode, run tasks automatically, access specialized GPT plugins? Go with ChatGPT Plus. **Do you regularly work with long documents** — contracts, research papers, detailed reports — or want the most careful writing style? Try Claude Pro. **Not sure what you'd use it for?** Start with ChatGPT Plus. It's the most general-purpose plan and the easiest to explore. If after a month you find you're mainly using it for long text tasks, switch to Claude Pro. If you find yourself mostly in Google's tools anyway, switch to AI Pro. ## What to Watch Out For **Prices and tier names change.** The figures in this article reflect what was verified for mid-2026, but AI service pricing has shifted fast. Before subscribing, check the current price on each service's own website. **Paid plans still have usage limits.** Plus, AI Pro, and Claude Pro all have daily or monthly caps — just much higher than free. Very heavy users running long tasks all day may still hit them occasionally. **Billing cycles and cancellations.** If you subscribe on the 5th and cancel on the 14th, you'll typically keep access until the end of that billing period, but you won't get a prorated refund. Check each service's specific cancellation policy. **Multiple subscriptions add up fast.** The most common mistake is subscribing to two or three services to try them and forgetting to cancel. Set a calendar reminder for one month after you subscribe. For a full picture of how AI tool costs accumulate, [What AI Tools Actually Cost You Per Month](/guides/chatgpt-monthly-budget) is worth reading before you commit. ## What to Try Next If ChatGPT Plus caught your eye, [Is ChatGPT Plus Worth It?](/guides/chatgpt-plus-worth-it) walks through the specific features and whether they match how you actually use AI. For a straight capability comparison across the three services, [Claude vs. ChatGPT: What's the Real Difference?](/guides/claude-vs-chatgpt) is the place to start. --- # Parental Controls for AI Apps: ChatGPT, Gemini and Copilot in One Guide Source: https://aiforregulars.com/guides/ai-parental-controls-all-apps Your child is likely already aware of AI assistants — or already using one. The question is not whether to allow it, but how to set it up so you have some visibility into what is happening. This guide covers the three most common AI apps — ChatGPT, Google Gemini, and Microsoft Copilot — and what each one actually offers for families right now. Each platform sets its own minimum age, and knowing the rules matters before you create or adjust any account. **ChatGPT (OpenAI):** The minimum age is 13. Users between 13 and 17 are supposed to have parental consent, though the age gate is self-reported. In some countries, the minimum is higher due to local data privacy laws. **Google Gemini:** Users under 13 have Gemini blocked by default on supervised accounts — parents must enable access through Google Family Link. Teens 13 and up with a supervised Google account may have restricted access depending on settings. **Microsoft Copilot:** The minimum age is 13. Microsoft Family Safety can be used to manage a teen's access and set content filters on their Microsoft account. Here is a quick side-by-side comparison of what each platform currently provides: | App | Min Age | Family Account Required | Content Filter | Activity Log | |---|---|---|---|---| | ChatGPT | 13+ (parental consent for under 18) | No dedicated family plan | Default filters; 'Reduce sensitive content' toggle available via linked teen account | Chat history visible on shared account only | | Gemini | 13+ standard; under 13 accessible if parent enables via Family Link | Yes — Google Family Link | Supervised account restrictions | Email alert on first use; no conversation log | | Copilot | 13+ with Family Safety | Yes — Microsoft Family Safety | Content filters via Family Safety dashboard | Activity reports in Family Safety | None of these tools is designed specifically for children. They are general-purpose AI assistants that have added some family-account features. Keep that in mind as you work through the steps below. ChatGPT has the most nuanced setup of the three, so there is a full guide dedicated to it: [ChatGPT Parental Controls and Safe Setup for Kids](/guides/chatgpt-parental-controls). It covers checking for existing accounts, creating a shared account, adjusting privacy settings, and layering in device-level controls. The short version: the safest arrangement is a shared account managed under a parent email address where you control the password and can see the full conversation history. If your teen already has their own account, check that the correct age was entered and that the Memory feature is turned off under Settings. There is currently no separate child profile or parent dashboard in ChatGPT. Google Family Link is Google's parental supervision tool. It links your child's Google account to yours and gives you oversight over certain apps and settings. **To set up Family Link:** 1. Open the Google Family Link app on your phone (or download it from your app store). 2. Follow the prompts to add your child's Google account. Children under 13 require a parent-managed account, which you will create during setup. 3. Once linked, you can approve app downloads, set daily screen time limits, and manage account-level settings from the Family Link dashboard. **What Family Link does for Gemini:** For supervised accounts belonging to users under 13, Gemini is blocked by default — parents can enable it through Family Link settings, though the Gemini web app remains unavailable for all supervised accounts. For teens 13 to 17 with a supervised account, access may be limited depending on the specific settings you choose. One important limitation: a supervised Google account does not give you a transcript of what your teen asked Gemini. The main control is access management, not conversation monitoring. Microsoft Family Safety works similarly to Google Family Link. It connects your child's Microsoft account to a parent-managed family group and lets you control settings from a central dashboard. **To set this up:** 1. Sign in to your Microsoft account at account.microsoft.com or open the Microsoft Family Safety app. 2. Go to Family Safety and choose "Add a family member." Follow the prompts to invite or create your child's account. 3. Once they are part of your family group, you can set content filters, screen time limits, and spending limits from the Family Safety dashboard. **What Family Safety does for Copilot:** Content filters configured in Family Safety apply across Microsoft products, including Copilot. The filters are designed to block explicit material. You can also access activity reports that show search queries and websites visited through Microsoft services. As with the other platforms, these controls limit what is easily accessible rather than creating a fully monitored conversation log. I will be direct: I have tested all three of these setups, and none of them left me feeling like I could walk away without thinking about it again. The settings help — they are worth doing — but they are a floor, not a ceiling. What matters more is whether your child feels comfortable coming to you when something they see online confuses or upsets them. That is harder to configure than a content filter, and it is the protection that actually lasts. After configuring controls, spend ten minutes with your child using each app together. Try a prompt or two that you would want a filter to catch — something clearly age-inappropriate, or a request for dangerous information. See what happens. This confirms that the controls are working as you expect, and — more importantly — it gives you a shared reference point with your child. They see what gets blocked, and you both talk about why. That conversation is more useful than any single setting. Do not be surprised if some test prompts get through. No filter catches everything, and naming that out loud with your child when it happens is worthwhile. Parental controls slow things down. Conversation changes behavior. A few things worth covering whenever the topic comes up naturally — not in a single formal sit-down, but over time as context arises: AI tools make mistakes and can sound more confident than they should. Important facts need to be checked elsewhere. Anything typed into an AI goes to a company's servers, so personal details — home address, school name, phone number — should stay out of it. And if they see something in an AI conversation that confuses or bothers them, they should be able to bring it to you without worrying about getting in trouble for the question. These points land better in the moment than in a lecture. When homework comes up, when a news story mentions AI, when they ask you something about it — those are the natural openings. **What to try next:** For a deeper look at ChatGPT on its own — account options, privacy settings, and device-level controls — [ChatGPT Parental Controls and Safe Setup for Kids](/guides/chatgpt-parental-controls) covers each step in full. And for a five-minute family safety project you can do right now, [Safe Words for Families](/guides/family-safe-words) explains how to set up a code word that protects against voice scams — a useful addition to any conversation about online safety. --- # How AI Cracks Passwords — and How to Lock Down Your Accounts Source: https://aiforregulars.com/guides/ai-password-security In an AARP survey of U.S. adults 50 and older, worry about AI being used to crack passwords ranked first at 87% — ahead of deepfakes (85%) and voice cloning (84%). (Source: AARP Public Policy Institute.) That concern is well founded. AI tools can now test enormous lists of stolen passwords across hundreds of websites automatically, faster than any human could manage. The reassuring part: you do not need to be technical to protect yourself. The four steps below work whether you have an iPhone, an Android phone, a Windows PC, or a Mac — and the most important first step takes about five minutes. ## What "AI Password Guessing" Actually Means When you hear that AI can "crack" passwords, here is what is actually happening. Criminals obtain lists of usernames and passwords from old data breaches — leaks from large websites that happened years ago. Those lists are sold online. Software, increasingly powered by AI, then tries those username-and-password combinations on hundreds of other websites automatically. This is called credential stuffing. The AI part makes it faster and smarter. Instead of trying random combinations, the software focuses on combinations that are statistically likely to work: common words, names, birth years, simple patterns. It learns which guesses succeed. The target is rarely your strongest password. It is your weakest one — on a site you forgot about — that you also reused somewhere that matters. ## The Real Danger: Reused Passwords If you use the same password on more than one site, a breach on any one of them can unlock the others. This is the single biggest password vulnerability most people have. A password that looks strong but is used on five sites is less safe than a simpler password used only once. Think of it this way: every site holding your email and password has a copy of your key. If one site drops its key and criminals pick it up, they try it on your bank, your email, your government accounts, your health records. The solution is not harder-to-remember passwords. It is making sure no two sites share a password — and the next step shows you the easiest way to do that. ## Step 1: Use a Password Manager A password manager does one job: it stores a different, randomly generated password for every site you use. You only need to remember one master password to open the manager. Every other password is long, random, and unique — the kind that no person and no AI can guess. Your phone almost certainly has one built in. On an iPhone, go to Settings, then Passwords. On an Android phone, go to Settings, then Google, then Autofill. Both are free and work automatically once you turn them on. When you visit a site, the password manager fills in your login for you. You do not need a paid app to get started. The built-in managers on iPhones and Android phones are excellent for most people. Start there. One rule: your master password — the one that opens the manager — should be a passphrase. That means four or five ordinary words strung together, like "purple kitchen blanket sunset." Easy to remember, very hard for software to guess. This lines up with current U.S. government guidance: NIST's federal password standard (Special Publication 800-63B) favors length over complexity and no longer recommends forced special characters or scheduled password changes — a long passphrase you actually remember beats a short, "complex" password you have to write down. ## Step 2: Try Passkeys — Sign In With Your Face or Fingerprint A passkey is a newer, simpler way to log in. Instead of a password, you use your fingerprint, your face, or your phone's PIN. The website never receives a password at all, so there is nothing for criminals to steal in a breach. More and more major services — Google, Apple, Microsoft, Amazon, PayPal — now offer passkeys. When a site asks whether you want to "save a passkey" or "sign in with a passkey," say yes. It is safer than any password you could type. You are probably already using a version of this on your phone when you unlock it with your face or fingerprint. Passkeys bring that same simplicity to websites and apps. ## Step 3: Turn On Two-Factor Authentication Two-factor authentication (often called 2FA) means that even if someone gets your password, they still cannot log in without a second check — usually a code sent to your phone. There are two types of 2FA, and they are not equally safe: **SMS codes** — the site sends a six-digit code by text message. This is better than nothing, but phone numbers can be taken over by scammers in some circumstances, making it the weaker option. **An authenticator app** — free apps like Google Authenticator or Microsoft Authenticator generate a fresh code every 30 seconds on your phone. Even if a scammer has your password, they cannot get that code without physically holding your phone. This is much more secure. Start with SMS if it is the only option offered — it is still a meaningful improvement over no 2FA at all. But for your most important accounts (email, banking, social media), switch to an authenticator app when you can. ## Step 4: Check If Your Email Was in a Breach A website called Have I Been Pwned (haveibeenpwned.com) lets you type in your email address and see which data breaches have exposed it. It was built by a security researcher and is free to use. The site does not ask for your password — only your email address. Go there, type your email, and press the check button. The site will show you a list of any breaches where your email address appeared. If your email shows up in a breach, your password for that site is known to criminals. Change that password right away — and change it on any other site where you used the same password. ## What to Watch Out For **Fake "your account was hacked" emails.** Criminals send emails that look like official security alerts: "We detected unusual activity. Click here to secure your account." These are phishing attempts. Do not click the link. Instead, go directly to the site by typing its address in your browser, or call the company using a phone number from its official website. **Password-reset phishing.** A scammer triggers a real password-reset email from a service you use, then calls you pretending to be from the company's security team. They ask you to read them the code that just arrived on your phone. Never read a security code to someone who called you. Real company employees never ask for this. **Pop-up fakes that imitate your password manager.** Your real password manager fills in your login credentials automatically — it never asks you to retype your master password on a webpage. If a site is asking for your master password, close that tab. I keep one thing written down in a safe place at home: the four-word passphrase that opens my password manager. Not my individual passwords — just that one key. Everything else is stored safely inside the manager, and I never have to remember another password again. ## What to Try Next Phishing emails are often the first move criminals make before attempting an account takeover. [How to spot AI phishing emails](/guides/spot-ai-phishing-emails) gives you six quick checks that reveal fake messages — even the ones that look completely legitimate. If you're also wondering what personal details you should never share with an AI chatbot, [What not to tell an AI chatbot](/guides/what-not-to-tell-ai-chatbot) covers the specific information that could put you at risk. ## Sources - [Older Adults Express High Concern and Limited Knowledge About AI Scams and Fraud](https://www.aarp.org/pri/topics/work-finances-retirement/fraud-consumer-protection/ai-fraud-concerns-older-adults/) — AARP Public Policy Institute - [Have I Been Pwned](https://haveibeenpwned.com/) — Troy Hunt - [Digital Identity Guidelines: Authentication and Authenticator Management (SP 800-63B-4)](https://csrc.nist.gov/pubs/sp/800/63/b/4/final) — NIST - [FIDO Passkeys: Passwordless Authentication](https://fidoalliance.org/passkeys/) — FIDO Alliance - [Criminals Increasing SIM Swap Schemes to Steal Millions of Dollars from US Public](https://www.ic3.gov/PSA/2022/PSA220208) — FBI Internet Crime Complaint Center (IC3) --- # AI Pet Care: What to Ask ChatGPT About Your Dog or Cat Source: https://aiforregulars.com/guides/ai-pet-care-chatgpt Owning a pet means a steady stream of questions: Is this food okay? Why is she doing that? How do I get him to stop barking at 6 a.m.? A vet visit for every small question isn't practical. ChatGPT can fill a lot of those gaps — if you know what to ask and what to leave to the professionals. ## What AI is genuinely good for ChatGPT has absorbed a huge amount of veterinary and pet care information. It handles everyday questions well, explains things in plain language, and never makes you feel embarrassed for asking something basic. ![ChatGPT answering a question about a dog's itchy ears](/screens/ai-pet-care-chatgpt/example-answer.webp) *A real question and the answer it gave, with its sources — our own screenshot, 31 July 2026.* If you're thinking about getting a new pet, or you just adopted one and want to understand its personality, AI is a great research tool. ``` I'm thinking about getting a Cavalier King Charles Spaniel. What health issues are common with this breed, and what should I know before I adopt? ``` ``` My rescue cat is a tabby, about 4 years old. What are typical tabby behaviors I should expect? ``` You'll get a realistic overview, including the less glamorous parts breeders sometimes skip. Portion sizes, feeding schedules, foods that are dangerous for pets — ChatGPT handles these well. ``` My 8-year-old Labrador weighs 70 lbs and is a bit overweight. How many calories should she be eating per day, and what should I look for on a dog food label? ``` ``` Can cats eat plain cooked chicken? What about tuna — is it safe as a regular food? ``` ``` My dog ate a small piece of dark chocolate about an hour ago. He weighs 20 lbs. Should I be worried? ``` That last question is a good example of AI being useful as a first step — it can tell you whether something is low-risk or worth an emergency call. But if there's any real concern, call your vet or an animal poison control hotline immediately. ChatGPT is patient and never runs out of time. It's especially good for step-by-step training guidance. ``` My puppy bites during play and it's getting painful. Walk me through a training plan to teach bite inhibition, step by step. ``` ``` My cat yowls loudly every morning at 4 a.m. What are common reasons for this, and how do I stop it? ``` ``` My dog lunges at other dogs on leash. What positive reinforcement techniques can I try before hiring a trainer? ``` If a behavior problem feels unsafe or is getting worse despite training, a certified professional trainer or a veterinary behaviorist is worth the cost. AI can help you understand the issue, but hands-on help matters for serious cases. Grooming instructions for specific breeds can be surprisingly hard to find in plain language. ``` How do I safely trim my golden retriever's paw fur at home? What tools do I need and what mistakes should I avoid? ``` ``` My cat hates being brushed. What's a step-by-step approach to get her used to it? ``` ChatGPT will walk you through the process as if it's right there with you. When my older cat started eating less, I asked ChatGPT to help me list the questions I should ask the vet — things like "what bloodwork makes sense at her age" and "what are early signs of kidney disease in cats." I walked into that appointment much better prepared, and the vet said it was a great question list. AI didn't replace the vet, but it made that visit more productive. You can describe what you're observing and ask ChatGPT what might cause it. This is useful for deciding whether to call the vet today, wait and watch, or go to an emergency clinic. ``` My dog has been licking his paws obsessively for three days. No visible sores. What are the most common causes and what should I watch for? ``` ``` My cat vomited once this morning but seems fine now. Is this something to worry about? ``` **Clear rule: call the vet immediately if your pet shows any of these signs** — difficulty breathing, collapse or extreme weakness, seizures, obvious pain, blood in vomit or stool, won't eat or drink for more than 24 hours, or suspected poisoning. AI is not the right first call in those situations. ChatGPT is a great writing assistant for pet owners. ``` Help me write a lost-dog post for Facebook. His name is Biscuit, he's a 5-year-old beagle mix, brown and white, last seen near Oak Street Park on July 3rd. ``` ``` I have a vet appointment tomorrow for my senior dog's annual checkup. What questions should I ask about aging, joint health, and cognitive changes? ``` ``` Write a pet profile for my cat to share with the pet sitter — include feeding schedule, favorite hiding spots, medications, and emergency contacts. ``` That last one is genuinely practical. A written pet profile means anyone looking after your animal has everything they need in one place. ## What to try next ChatGPT is useful for a surprising range of everyday tasks beyond pet care. See the full list of [things worth asking an AI chatbot](/guides/things-to-ask-chatgpt), or if you want to get better answers from ChatGPT in general, read the guide on [how to write good prompts](/guides/how-to-write-prompts) — a few small changes make a big difference in what you get back. ## Sources - [Pet owner resources](https://www.avma.org/resources-tools/pet-owners) — American Veterinary Medical Association - [Animal Poison Control](https://www.aspca.org/pet-care/animal-poison-control) — ASPCA - [Usage policies](https://openai.com/policies/usage-policies) — OpenAI --- # AI Plant and Bird ID: Point Your Phone and Get the Answer Source: https://aiforregulars.com/guides/ai-plant-bird-identification You're walking in a park and spot a flower you've never seen before. Or a bird lands nearby and you have no idea what it is. A few years ago, finding out would mean buying a field guide. Now you just point your phone at it. AI identification tools have gotten surprisingly good. Here's how to use them — starting with what's already on your phone. ## The Apps Worth Knowing Before the step-by-step, a quick overview of the main options: - **Google Lens** — built into most Android cameras and the Google app on iPhone. Best starting point because you probably already have it. Identifies plants, flowers, and trees well. - **Seek by iNaturalist** — free app from the California Academy of Sciences and National Geographic. Shows a confidence meter as you point your camera. Great for beginners because it shows you the genus even when it can't nail the exact species. Privacy-friendly: doesn't require an account. - **Picture This** — app focused specifically on plants. More detailed care and toxicity information than Google Lens. Free version has daily limits; paid version is more capable. - **Merlin Bird ID** (Cornell Lab of Ornithology) — free, excellent, and the most trusted bird identification app available. Works by photo or by sound. On **Android**: Open your camera app and look for a Lens icon (a small square with a circle inside) in the corner. Tap it, point at the plant, and Lens shows identification results in real time. On **iPhone**: Open the **Google app** (download it free if you don't have it), tap the Lens icon in the search bar, then point at the plant. Lens shows you its best guess and usually links to more information. For common garden plants, flowers, and trees in your yard, it's often all you need. Download **Seek** from the App Store or Google Play (free, no account required). Open it and tap the camera icon. Point your phone slowly at the plant. Seek shows a real-time readout — first a broad category (flowering plant, insect, bird), then narrowing down as it gets more confident. When it's confident enough, it shows the full species name. Seek also shows an "evidence" indicator — how many observations of that species have been reported near your location. If thousands of people have logged that species nearby, you can be more confident. **Tip:** Move slowly, keep the subject in good light, and try to get the whole plant or leaf clearly in frame. Download **Merlin Bird ID** (free). The first time you open it, set your location so it knows which birds live in your area. **By photo:** Tap **Photo ID**, upload a photo or take one now. Merlin asks a few quick questions (size, main color, behavior) and then shows you the top matches with photos for comparison. **By sound:** This is the most impressive feature. Tap **Sound ID**, tap the microphone, and hold your phone toward a bird you can hear. Merlin lists the species it hears in real time — often multiple species at once if several birds are calling. You can see exactly which sound triggered each identification. For backyard birds, feeders, parks, or a nature walk, Merlin is remarkably accurate. I use Merlin Sound ID on morning walks. I'll hold my phone up in a park and watch three or four species appear on screen from birds I can hear but can't see. It completely changed how I pay attention outside — I started noticing how many different birds were around me all along. Once you have a species name, ChatGPT is great for learning more. You can take a screenshot of the identification and upload it, or just type the species name. Good follow-up questions: ``` I just identified a black-eyed Susan (Rudbeckia hirta) in my yard. Is it native to the eastern United States? Does it attract butterflies? Can I eat any part of it? ``` ``` I saw a red-tailed hawk in suburban Ohio. What does it eat? Is it common year-round or does it migrate? How do I tell it apart from a Cooper's hawk? ``` ChatGPT can explain things at whatever level you want — ask it to "explain this like I'm a complete beginner" if you want a simpler answer. Both Seek and Merlin let you save the species you identify to a personal log. Seek contributes your observations (anonymously if you don't have an account) to iNaturalist, a global database used by scientists. Over time, your log becomes a record of what you've seen — which birds visit your yard in winter, which wildflowers bloom along your usual walking path. Some people find this unexpectedly satisfying. You can also upload your observations to iNaturalist's full website for more detailed information and to have the identification confirmed by the community. ## A Safety Note for Plants AI identification is good, but not infallible. For plants: - **Never eat anything based solely on an app ID**, especially mushrooms or wild berries - If a child has touched or eaten an unknown plant, contact poison control — don't rely on an app - For plants you want to eat or use medicinally, confirm with a local expert or a trusted field guide For casual curiosity — identifying what's growing in your garden or along a trail — the apps are great. For anything with safety implications, add a second opinion. ## What to try next Now that you've seen how AI can answer nature questions from a photo, it's worth knowing what else you can ask ChatGPT by just describing what you found. The [guide to useful things to ask ChatGPT](/guides/things-to-ask-chatgpt) has many practical examples that go well beyond plant and bird questions. --- # How to Use AI to Understand Your Prescription and Side Effects Source: https://aiforregulars.com/guides/ai-prescription-explainer Prescription labels and drug information sheets are written for regulators, not for patients. They're full of medical jargon, long lists of rare side effects, and Latin abbreviations like "q.i.d." that most people have never seen before. No wonder so many people end up confused — or worse, worried about side effects that apply to almost no one. AI can bridge that gap. This tutorial shows you how to use a general AI chatbot to understand your prescription in plain language, while making sure you verify the important details with the professionals who know your complete health picture. **The rule throughout:** AI explains. Your pharmacist and doctor confirm. Before turning to AI, read through the label carefully and note what you don't understand. The label tells you: - The medication name (both brand and generic) - The dose (e.g., 10 mg) - How often to take it (e.g., "once daily" or "q.i.d." which means four times a day) - Special instructions (e.g., "take with food," "avoid grapefruit") - Refills remaining Write down the terms or instructions you don't understand. You'll use these in your AI prompts. Don't photograph the label and upload it — type out the drug name and dose instead, so you're not sharing identifiable information. Start with the most basic question: ``` What is metformin 500mg used for? Explain it in plain English for someone with no medical background. ``` A good AI will explain the general purpose of the drug, how it works in simple terms, and who it's typically prescribed for. It should also say something like "your doctor prescribed this for your specific situation" — because the same drug is sometimes used for more than one condition. If the explanation doesn't match what your doctor told you, make a note and ask your doctor or pharmacist at your next contact. Drug information sheets list every side effect reported in clinical trials, including ones that happened to fewer than 1 in 1,000 people. That list can be alarming if you read it without context. Ask AI to give you a more useful picture: ``` What are the most common side effects of metformin that most people actually notice? And which ones are rare but serious enough that I should call my doctor? ``` This framing gets you two useful lists: what to expect versus what to watch for. If you experience something that's on the "call your doctor" list, do that — don't keep asking AI. I've found this "common vs. serious" framing really useful. Drug information sheets list everything that happened in trials without telling you how likely any of it is. Asking AI to sort those into "probable" and "rare but important" gives a much calmer and more actionable picture. Many medications work differently depending on when and how you take them. AI can explain the general rules: ``` Should lisinopril be taken at a specific time of day? Does food affect how it works? ``` ``` I've heard grapefruit can interact with some medications. Does that apply to atorvastatin? ``` These are good AI questions because the answers are general pharmacology principles rather than personal medical judgment. That said, always cross-check with your pharmacist — especially for newer medications. Your pharmacist is a free, highly trained resource that most people underuse. After reviewing AI explanations, build a short list of questions for your next pharmacy visit or call: ``` Based on what I've learned about metformin, help me write 4 questions to ask my pharmacist. ``` Good questions to consider asking: - How should I store this medication? - What should I do if I miss a dose? - Are there any over-the-counter medications or supplements I should avoid? - Is there a generic version if this is expensive? Bring the written list. Pharmacists are used to answering questions and it's absolutely appropriate to ask. Prescription abbreviations come from Latin and are genuinely confusing. If you see something like "sig: 1 tab PO BID AC," you can ask: ``` What do these prescription abbreviations mean: PO, BID, AC, PRN? ``` Here's a quick reference of common ones: | Abbreviation | Meaning | |---|---| | PO | By mouth (oral) | | BID | Twice a day | | TID | Three times a day | | QID | Four times a day | | PRN | As needed | | AC | Before meals | | PC | After meals | | HS | At bedtime | | QD or QDay | Once daily | AI can explain any abbreviation you encounter that isn't on this list. ## When to call your doctor or pharmacist — not AI Contact your pharmacist or doctor directly (don't rely on AI) when: - You're unsure whether to take a dose you missed - You're experiencing a side effect that concerns you - You're about to start a new supplement or over-the-counter medication - A medication is causing a reaction, rash, or difficulty breathing — call 911 for severe reactions - You want to stop taking a medication your doctor prescribed ## What to try next If you'd like to prepare more thoroughly for the appointment where a prescription was or will be discussed, [Preparing for Your Doctor's Appointment with AI](/guides/ai-doctor-visit-prep) walks through building a symptom log and question list step by step. For a broader look at when to trust AI health information, see [Should You Ask AI About Your Health Symptoms?](/guides/ai-health-symptoms). --- # 15 AI Prompts That Save 5 Hours of Office Work a Week Source: https://aiforregulars.com/guides/ai-prompts-office-work Most office workers spend a significant part of their day on writing tasks — meeting notes, emails, status updates, reports. The work is not hard. It is just slow and repetitive, and it crowds out the things that actually require your judgment. These 15 prompts hand that first draft to AI. You fill in the details, review the result, and send. **How to use this list:** Copy the prompt, replace the brackets with your real information, paste it into ChatGPT (or any AI chat tool), and edit the result before using it. Never send an AI draft without reading it first. | Category | Prompts | What they handle | |---|---|---| | Meetings | 1–5 | Agendas, summaries, action items, decisions, catch-ups | | Inbox | 6–10 | Follow-ups, declinations, complaints, introductions, status updates | | Reports | 11–15 | Summaries, long documents, data narratives, presentations, weekly reports | --- ## Meetings **1. Write a meeting agenda** Use this the day before any meeting where you are the organizer. ``` Write a short meeting agenda for a [30/60]-minute meeting about [topic]. Attendees: [list roles, not names]. Goal of the meeting: [what decision or outcome do we need?] Format it as a numbered list with time estimates for each item. ``` **2. Summarize rough meeting notes** Use this right after a meeting, while everything is still fresh in your mind. ``` Here are my rough notes from a meeting. Turn them into a clear summary with three sections: Key Decisions, Action Items (with owner and deadline), and Open Questions. Keep it under 200 words. Notes: [paste your rough notes here] ``` **3. Pull action items from a transcript** Use this if your video call tool — Zoom, Teams, or Google Meet — gave you an auto-generated transcript. ``` Read this meeting transcript and extract all action items. For each one, list: what needs to be done, who is responsible, and the deadline if one was mentioned. Format as a table. Transcript: [paste transcript here] ``` **4. Write a post-meeting decision email** Use this when a meeting produced a decision that affects people who were not in the room. ``` Write a short email to [team or department] summarizing a decision made in today's meeting. The decision was: [describe it]. It affects them because: [explain briefly]. Tone: clear and professional. Two short paragraphs maximum. ``` **5. Write an async catch-up message** Use this when you missed a meeting and need a summary without interrupting your whole team. ``` Write a short, professional message to a coworker asking for a quick catch-up on what I missed in today's [meeting name] meeting. Keep it casual and under three sentences. ``` ## Inbox **6. Write a follow-up for no response** Use this when someone has not replied after three to five business days. ``` Write a polite follow-up email. I'm checking in on [topic or request] I sent to [Name or role] on [date]. Keep it friendly, not pushy. Two sentences maximum. ``` **7. Say no to a request** Use this when declining a meeting invite, an extra project, or a favor at work. ``` Help me write a polite but clear email saying no to [request] from [Name or role]. I want to be respectful and not close the door completely on future opportunities. Three sentences maximum. ``` **8. Respond to a complaint** Use this for unhappy customers, clients, or internal stakeholders who have raised an issue. ``` Write a professional response to this complaint. Acknowledge the issue, apologize without admitting legal fault, and explain the next step. Tone: calm and empathetic. The complaint: [paste or summarize it here] ``` The prompt I use most often is number 2 — the meeting summary. I paste in whatever notes I managed to scribble during the call, and ChatGPT turns them into something I can actually send to the team. It saves me at least 15 minutes per meeting. As a bonus, my notes have gotten better, because I stopped trying to write complete sentences during the meeting and just jotted keywords instead. **9. Reply to an introduction email** Use this when someone new reaches out to connect, introduce themselves, or open a conversation. ``` Write a warm, professional reply to this introduction email. Match their tone. Express genuine interest and suggest one specific next step (a call, coffee, or a follow-up in a month). Their email: [paste it here] ``` **10. Write a status update to your manager** Use this for weekly or biweekly check-ins, especially if writing these tends to eat more time than it should. ``` Write a short status update email for my manager. This week I: [list 2–3 things you completed]. Next week I plan to: [list 1–2 priorities]. Blockers: [none / describe briefly if any]. Format as bullet points. No filler. ``` ## Reports **11. Turn notes into an executive summary** Use this when you have rough notes or a long document and need a short version for leadership. ``` Turn the following notes into a 150-word executive summary. Audience: senior leadership who were not involved in this project. Focus on outcomes and decisions, not process. Notes: [paste your notes here] ``` **12. Summarize a long document** Use this for long reports, policy documents, or articles you do not have time to read in full. ``` Summarize this document in plain language. Use bullet points. Highlight: the main point, any action required from me, and any deadlines mentioned. Document: [paste text here] ``` **13. Turn data points into a written narrative** Use this when you have numbers or results and need to write about them for a non-technical audience. ``` I have the following data. Write two short paragraphs explaining what these numbers mean in plain language — no jargon. Write for someone who is not a numbers person. Data: [paste your figures or bullet list here] ``` **14. Create a presentation outline** Use this when you need to build slides and are not sure where to start. ``` Create a slide-by-slide outline for a [10/15/20]-minute presentation on [topic]. Audience: [describe who they are]. Goal: [what should they understand or do after the presentation?] Include a title slide, 5–7 content slides, and a closing slide. ``` **15. Write an end-of-week report** Use this every Friday — it takes about two minutes and keeps your manager in the loop without a standing meeting. ``` Write a brief end-of-week report. Format: one short paragraph overview, then a bullet list of accomplishments. This week: [paste 4–6 bullet points of what you worked on] Wins worth highlighting: [one or two if any] Next week's main focus: [one priority] ``` --- ## Make it a habit The people who save the most time with AI are not the ones who find the perfect prompt. They are the ones who reach for AI first whenever they sit down to write something routine. A two-minute habit, applied consistently, becomes several hours a week. Start with the one category that takes the most time in your current job — meetings, inbox, or reports — and try two or three of those prompts this week. **What to try next:** These prompts work just as well for job searching. Learn how to [write a professional email with AI](/guides/write-professional-email-with-ai) for the tricky messages, or see how to [write a cover letter with ChatGPT that sounds like you](/guides/chatgpt-cover-letter). --- # 20 Ways Parents Can Use AI to Save Time This School Year Source: https://aiforregulars.com/guides/ai-prompts-parents The school year brings a flood of small tasks that eat up more time than they should: writing emails to teachers, building after-school schedules, organizing field trip logistics, figuring out what goes in the backpack. AI tools like ChatGPT, Gemini, and Copilot can handle the first draft of almost all of them. Here are 20 ready-to-use prompts organized by task. Copy them, fill in the brackets with your details, and edit the output before using it. These work on any free AI chatbot. Open ChatGPT, Gemini, or Copilot and paste this, filling in the brackets: ``` Write a short, polite email to my child's teacher. My child is [name], in [grade]. I want to let her know [explain the situation, e.g., "he's been struggling with reading at home and ask if there are extra resources"]. Keep the tone warm and professional, under 150 words. ``` Read it before sending. AI emails tend to be slightly more formal than most parents write — adjust the tone if needed. Adding one specific detail from your situation makes it feel personal rather than templated. ``` Create a weekly after-school schedule for a [age]-year-old. Activities: [list activities, e.g., soccer Tuesday and Thursday, piano Wednesday]. Homework should take about [30/45/60] minutes. Add a 30-minute outdoor or free-play block each day. Format it as a simple table, Monday through Friday. ``` The table AI produces is easy to paste into a Google Doc or print out. Ask it to adjust if any day looks too packed. ``` Write a permission slip for a school field trip to [destination] on [date]. Include: purpose of the trip, departure and return time, cost if any [$X], what to bring, and a signature line at the bottom. Keep it to one short paragraph plus the signature block. ``` Always check that the details match what the school sent home before printing. AI might fill in plausible-sounding but incorrect specifics if you leave the brackets vague. ``` Create a packing list for the first week of school for a [grade] student. Include: backpack essentials, supplies, lunch items, and one section each for: [gym day, art class, outdoor recess]. Format as a checkbox list. ``` Add your child's specific needs: "no nut products" or "needs an extra change of clothes." AI will work around them and keep the list clean. ``` Plan 5 school lunches for a [age]-year-old who [likes/dislikes: e.g., "likes pasta and fruit but won't eat sandwiches"]. Each lunch should take under 10 minutes to prepare. No peanuts. Format as a simple list with one prep note per day. ``` State allergies and dislikes clearly upfront. You can also ask it to repeat the format next week with different options so you build a rotation over time. I started using AI for teacher emails last fall and the response rate went up noticeably. I think it is because I stopped putting off writing the email when I was tired. AI lowers the activation energy enough that I actually send it the same day. ``` Write a brief absence note for school. My child [name] was absent on [date] due to [reason, e.g., illness/family appointment]. Keep it to 2–3 sentences, polite and direct. ``` These take about 30 seconds with AI and look more composed than a handwritten sticky note written in a hurry. ``` I'm going to paste a school newsletter. Read it and give me: the 3 most important dates, any action items I need to take, and anything time-sensitive. Here's the newsletter: [paste text] ``` This works for any long document: curriculum nights, PTA announcements, handbook updates. Paste the full text and let AI pull out what matters. ``` Help me prepare for a parent-teacher conference for my [grade] child. I want to ask about: [list concerns, e.g., "reading progress, how homework is graded, whether there is any differentiation for kids ahead of grade level"]. Write 5 questions in a natural, friendly tone. ``` Print the list or keep it on your phone during the meeting. Having questions written down prevents the kind of blank-mind moment that happens when you suddenly have 15 minutes with the teacher. ``` Write a sincere thank-you note to my child's [subject] teacher. My child [name] has [specific improvement or experience, e.g., "gone from hating math to asking for extra problems"]. Keep it genuine and specific, under 100 words. ``` Add your own sentence or two before sending. The AI version handles structure; you handle the specific detail that makes it feel real. ## 11 More Quick-Start Prompts Here are more you can copy and adapt for common school-year situations: **10.** "Write a group text to the parents on my child's soccer team reminding them about [event] on [date]. Friendly, under 80 words." **11.** "Create a morning routine checklist for a [age]-year-old, from wake-up to out the door in [X] minutes." **12.** "Help me write a response to a school survey about [topic] in 3–4 sentences." **13.** "Draft an email to the school librarian asking for book recommendations for a [grade] student who loves [genre]." **14.** "Write a short speech for a class parent volunteer introduction — 60 seconds, warm and brief." **15.** "Create a study schedule for the week before [subject] finals for a [grade] student." **16.** "Write a kind but firm email to another parent about [situation, e.g., a conflict between our kids at recess]. Keep it constructive." **17.** "List 10 quick dinner ideas for school nights — under 20 minutes, kid-friendly, no [allergy]." **18.** "Draft a note for my child's coach explaining why she will miss practice on [date]." **19.** "Create a birthday party invitation for a [age]-year-old's [theme] party on [date] at [location]." **20.** "Write a short 'about me' bio for the school directory — 2–3 sentences about our family, warm and brief." ## What to Try Next If you want to get better at writing your own prompts beyond these templates — so you can adapt them quickly when the situation is different — [the guide to writing good prompts](/guides/how-to-write-prompts) explains the key habits in plain language. And if you are curious what a monthly AI subscription actually costs versus using the free tools, [this breakdown of ChatGPT's pricing](/guides/chatgpt-monthly-budget) covers it honestly. --- # The New QR Code Scam Powered by AI — and How to Avoid It Source: https://aiforregulars.com/guides/ai-qr-code-scam You've probably scanned dozens of QR codes in the past year — to look at a restaurant menu, pay for parking, or check in somewhere. That little square is now one of scammers' favorite tools, and AI has made it dramatically easier for them to create convincing fake ones. The trick is called quishing, and it's spreading fast enough that the FBI and FTC have both issued warnings about it. ## What Is Quishing? The word combines "QR" and "phishing" — the broader category of scams that impersonate a trusted organization to steal your login or payment details. Here's how it works: a scammer creates a QR code that, when scanned, takes you to a fake website that looks exactly like your bank, your city's parking payment portal, or a shipping company's tracking page. The page asks you to log in, confirm a payment, or verify your identity. If you do, your credentials or card number go straight to the scammer. **The reason QR codes are particularly effective for scammers** is that email and text security filters are built to scan for suspicious text links — they're not built to read images. A fake QR code embedded in a PDF or an image attachment passes right through most filters undetected. ## Where Fake QR Codes Show Up Fake codes are turning up in some very ordinary places: **Parking meters.** A sticker with a fake QR code gets placed over the legitimate payment code. You scan it, enter your credit card on a convincing fake "parking payment" page, and the meter never registers your payment. **Restaurant menus.** Most QR menus are safe, but scammers have been caught placing stickers over the original codes. The fake version redirects to a page that might ask for your credit card "to reserve your spot" or offer a discount sign-up. **Package delivery texts.** You get a text saying your delivery is delayed and asking you to scan a code or follow a link to reschedule. The URL leads to a fake page asking for your address and payment details. **Flyers and posters.** Fake codes have appeared on community flyers — a "free gift card" or "exclusive offer" — placed in laundromats, libraries, and apartment building lobbies. **Emails and PDF attachments.** A message from a supposed bank, utility, or delivery company contains a QR code asking you to "verify your account." This is the most common digital version, and the one most likely to slip past email security software. ## Why AI Made This Much Worse Before AI writing tools became widely available, a scam email was often easy to spot: clunky writing, grammatical errors, an odd request that didn't quite sound like how a real bank would communicate. That's no longer a reliable tell. According to the Phishing Trends Report from security firm Hoxhunt, the share of phishing attacks that were AI-generated jumped roughly 14-fold in a single month at the turn of 2025 and 2026 — from under 5% to 56% of detected attacks — before settling back to around 40% in the months since. Scammers are using AI to write perfectly professional emails, generate realistic-looking fake login pages, and create QR codes at scale — hundreds of slightly different versions targeting different people or regions. The combination is especially effective: the email reads like it came from a real company, the QR code bypasses text-based filters, and the fake website looks identical to the real one. The only moment you have to catch it is when you check the URL the code actually points to. I now make a habit of reading the URL preview on my phone before tapping any QR code — not just ones from strangers, but from any source. The preview appears for about two seconds before the page loads, and it takes less time to read than this sentence. If the domain looks odd or unfamiliar, I back out and go to the official website directly. ## How to Check a QR Code Before You Tap You don't have to give up using QR codes entirely. You just need one quick habit: read the URL preview before you tap. **On iPhone:** When you point your camera at a QR code, a small notification appears at the top of the screen showing the destination URL. Don't tap that notification immediately — read the address first. **On Android:** Most Android camera apps show a small URL bar below the QR code frame. Read it before tapping. **What to look for in the URL:** - Does the domain match the organization you'd expect? A city parking app might be "cityname.gov" or "parkXYZ.com" — it shouldn't be "parking-secure-pay.net" with a random domain you don't recognize. - Is there a long string of random characters in the URL? Legitimate QR codes for a menu or payment page usually have clean, short addresses. - Does the domain use subtle misspellings — "rnybank.com" instead of "mybank.com"? That extra letter is easy to miss at a glance. ## Safe Habits Going Forward **Type addresses manually when you're uncertain.** If a QR code in an email asks you to "verify your account" with your bank, don't scan it — open your bank's app or type the address you know into your browser. This lines up with the FTC's core guidance on QR codes: don't scan a code from an unexpected text or email, and go to the company directly using a number or address you already trust rather than anything in the message. **Never enter payment or login details after scanning a public code.** If a parking meter, poster, or any public QR code asks for your credit card or password, stop. Legitimate parking apps and public payment systems let you pay through an established app you already have, or accept your card directly at the meter. They don't need you to enter your full card number on a website you reached via a QR sticker. **Check for stickers.** Before scanning a QR code on a physical surface — a meter, a table card, a poster — look at whether the code is printed directly or has a sticker edge on top. A sticker placed over a legitimate code is the most common physical form of this scam. **Be extra skeptical of QR codes in emails.** Real banks, utilities, and delivery companies rarely need you to scan a QR code. They have apps and websites you can reach directly. An unexpected email containing a QR code is a red flag even if the branding looks right. ## What to Watch Out For These habits will catch most quishing attempts, but no single check is foolproof. Scammers keep refining their technique, and the fake pages they build are sometimes visually indistinguishable from the real ones. If you've entered information after scanning a code and then felt uncertain, act quickly: change your password, call your bank, and report the incident at reportfraud.ftc.gov. The sooner you act, the better your chances of limiting any damage. One thing this guide doesn't cover: AI-powered follow-up phone calls that sometimes accompany QR scams. A scammer may call shortly after you've scanned a code, claiming to "verify your payment" and asking for card details you didn't enter on the fake site. If you get an unexpected call right after scanning a QR code, hang up and call the organization directly using a number you look up yourself. ## What to Try Next Quishing is part of a broader wave of AI-powered scams hitting inboxes right now — [How to Spot AI Phishing Emails](/guides/spot-ai-phishing-emails) covers the signs that a message wasn't written by a human. And if you're getting suspicious calls alongside suspicious messages, [The Best AI Scam Call Blockers](/guides/ai-scam-call-blocker) walks through the tools that can help filter them out. ## Sources - [Scammers hide harmful links in QR codes to steal your information](https://consumer.ftc.gov/consumer-alerts/2023/12/scammers-hide-harmful-links-qr-codes-steal-your-information) — Federal Trade Commission - [Scam alert: QR code on an unexpected package](https://consumer.ftc.gov/consumer-alerts/2025/01/scam-alert-qr-code-unexpected-package) — Federal Trade Commission - [Cybercriminals Tampering with QR Codes to Steal Victim Funds](https://www.ic3.gov/PSA/2022/PSA220118) — FBI Internet Crime Complaint Center (IC3) - [Phishing Trends Report](https://hoxhunt.com/guide/phishing-trends-report) — Hoxhunt --- # How to Get AI to Read Handwritten Letters and Family Recipes Source: https://aiforregulars.com/guides/ai-read-handwritten-letters-recipes A box of recipe cards in your grandmother's writing. A bundle of letters in cursive you can barely read. An AI chatbot on your phone can look at a photo of the page and type out what it says. It won't always be right, though, so this guide shows both how to do it and how to catch the mistakes. You can use ChatGPT or Gemini. Both are free to start with, and both let you attach a photo to a chat. If you haven't used one yet, see [how to use ChatGPT](/guides/how-to-use-chatgpt) first. Most bad readings start with a bad photo. Before you press the button: - Use **even light**, near a window or under a lamp, and avoid shadows from your hand or phone. - Hold the phone **straight above the page**, not at an angle, so the lines stay straight. - Keep it **steady** until the writing is sharp, then take the photo. - For a long letter or a page with writing in the margins, take **one photo per section**. Photograph hard-to-read passages separately. - Crop out what you don't need, such as a table or other papers, using your phone's photo editor. Open ChatGPT or Gemini, tap the button for adding a photo or file (look for a **+** or a paperclip), and choose your picture. Then type a request. Tell the AI what the page is, and ask it to be honest about what it can't read. You can copy this: > This is a photo of a handwritten family recipe. Please transcribe it exactly as written. If you can't read a word or a number, don't guess: write [?] and give me your best alternative readings. Don't add anything that isn't on the page. For a letter, change the first sentence: "This is a handwritten letter, probably from the 1950s, written in English." Giving the type of page, the rough period and the language helps the AI read unusual letter shapes and old phrases. Read the AI's text next to the photo, line by line. Pay special attention to: - **Names and places.** In one test, ChatGPT read a messy handwritten name as a completely different person's name. - **Dates and numbers.** A 1 and a 7, or a 3 and an 8, are easy to confuse. - **Recipe amounts and units.** Teaspoon or tablespoon? Half or one and a half? Oven temperature? Check each one against the card. - **Words the AI "fixed".** Chatbots like to make a sentence sound natural, which may mean smoothing over what was really written. Anything marked [?] is where the AI is telling you it's unsure. Look at those words yourself, or ask a relative who knew the writer. For a recipe, ask the AI to retype it as a clean list of ingredients and steps *after* you have checked the transcription, not before. Fixing a misread amount in the plain transcription is much easier than finding it later inside a tidied-up recipe. If the first reading has gaps, you can improve it: - Tell the AI what you've worked out. For example: "This writer makes the letter 'u' with a long tail, and 'ss' looks like 'fs'." Then ask it to read the page again. - Start small. Crop one tricky line, paste it in, and ask for just that line. - Give the same photo to a **second AI assistant** and compare. Where the two readings differ, the page is probably hard to read and deserves a closer look from you. ## What AI does well, and where it struggles Clear, fairly modern handwriting, such as a recipe card, a postcard or a shopping list, usually goes well. Take the photo the right way up, though: OpenAI warns that its image feature may misread rotated or upside-down text. Old or faded writing, unusual spelling, old abbreviations and local place names are much harder. People who use these tools for family history report that the AI can guess, skip words or produce something that sounds convincing and is wrong. OpenAI also says its image feature may be less accurate when an image is unclear. So treat the AI's text as a **starting point**, never as the final word. For an old document that matters, such as a will or a record you want to cite, the original page and a person who knows the period come first. ## Privacy: a letter is not just text Old letters often contain full names, addresses, health details and family secrets, some about people who can't agree to you sharing them. Before you upload: - Crop out addresses and anything you don't need read. - Check whether your chatbot is set to use your chats to improve the AI. In ChatGPT this is the setting called "Improve the model for everyone" under Settings, then Data controls. In Gemini, look at your Gemini activity settings. - Don't upload documents with ID numbers, bank details or passwords. See [what not to tell an AI chatbot](/guides/what-not-to-tell-ai-chatbot). ## What to try next If your handwritten pages are in another language, [translating a photo or document with your phone](/guides/translate-photo-document-ai) is the next step. For tracing a family tree from old records, read [AI for family history](/guides/ai-family-history-genealogy). And if you're wondering how far to trust any of this, see [can you trust ChatGPT](/guides/can-you-trust-chatgpt). ## Sources - [Document Deciphering: Using AI to Read Historical Handwriting](https://nextgengenealogy.substack.com/p/document-deciphering-using-ai-to) — Next Gen Genealogy - [Can ChatGPT Read Handwriting?](https://agptaweek.substack.com/p/can-chatgpt-read-handwriting) — A GPT a Week - [AI as "magnifying glass" to translate selective old hand written text](https://community.familysearch.org/en/discussion/comment/620808/) — FamilySearch Community - [Image inputs for ChatGPT FAQ](https://help.openai.com/en/articles/8400551-image-inputs-for-chatgpt-faq) — OpenAI Help Center - [Data controls FAQ](https://help.openai.com/en/articles/7730893-data-controls-faq) — OpenAI Help Center --- # How to Turn Fridge Leftovers into Recipes with AI Source: https://aiforregulars.com/guides/ai-recipe-from-leftovers Opening the fridge to a random collection of leftovers and having no idea what to cook is one of those small frustrations of everyday life. AI has a genuinely useful answer: just tell it what you have. Here's how to do it in a few minutes, starting with the exact prompt to use. Go to [chat.openai.com](https://chat.openai.com) in your browser, or open the ChatGPT app on your phone. You don't need a paid account for this — the free version works fine. Gemini (Google's AI) and Claude also work well for recipe generation if you prefer those. Look in your fridge and write down what you have. Don't worry about being precise — "some chicken thighs," "half a bell pepper," "a few eggs" is perfectly fine. Also note: - Any dietary needs (vegetarian, no gluten, no dairy) - How much time you have ("I want something ready in 30 minutes") - How many people you're cooking for Paste this into ChatGPT, filling in your actual ingredients: ``` I have these ingredients in my fridge: [list them here]. I also have basic pantry staples like olive oil, salt, pepper, garlic, and onions. Can you suggest a recipe that uses most of what I have? I'm cooking for [number] people and have about [X] minutes. [Any dietary notes, e.g. "no dairy" or "I don't eat pork"] ``` For example: *"I have chicken thighs, half a zucchini, some feta cheese, a lemon, and cherry tomatoes. Basic pantry staples on hand. Cooking for 2, have about 45 minutes."* ChatGPT will return a full recipe with ingredients, quantities, and step-by-step instructions. Read it through before you start cooking. If something doesn't work for you, just ask: - "I don't have feta — what else could I use?" - "Can I make this in an air fryer instead of the oven?" - "How do I know when the chicken is fully cooked?" You can have a full cooking conversation. It's like texting a knowledgeable friend who happens to know a lot about food. AI can suggest recipes, but it can't see your food. Apply normal food safety rules: - Don't use meat or fish past its use-by date. - Cook poultry to an internal temperature of 165°F (74°C). - If something smells off, trust your nose — no recipe is worth getting sick. - When in doubt about a leftover's age, throw it out. AI is great at creative combinations, but you are the last line of defense on whether the food is actually safe to eat. I use this almost every Sunday night when the fridge has random stuff from the week. The results have genuinely surprised me — some of my favorite meals have come from "use what's about to go bad" prompts. The key is being specific about what you have and not hiding the weird combinations. AI doesn't judge a fridge situation. ## What to Try Next If you want to take this further and plan meals for the whole week, read [AI Meal Planning: How to Plan a Week of Meals with ChatGPT](/guides/ai-meal-planning). And if you're new to writing AI prompts and want to get better results faster, [How to Write Good AI Prompts](/guides/how-to-write-prompts) covers the basics in plain English. --- # How to Remove Objects or People from Photos with AI Source: https://aiforregulars.com/guides/ai-remove-object-photo You took the perfect photo — but there's a stranger walking through the background, a trash can in the corner, or a power line cutting across the sky. AI object removal tools can fix all of those in about 30 seconds, right on your phone. Here's how to do it, starting with the tools already on your phone. The easiest options are already built into your phone's photo app — no download needed. - **iPhone (iOS 18+):** Open the photo in Apple Photos, tap **Edit**, then tap the **Clean Up** brush icon at the bottom. - **Android (Google Photos):** Open the photo, tap **Edit**, scroll to find **Tools**, then tap **Magic Eraser**. - **Samsung Galaxy:** Open the photo in Galaxy Gallery, tap the pencil/edit icon, then tap **Object Eraser** in the edit tools. If your phone doesn't have a built-in tool, download **Snapseed** (free, iOS and Android) — it works on any phone. Open your photo in whichever app you chose. Look for the object removal or healing tool — it usually looks like a brush or magic wand icon. In Snapseed: tap **Tools** at the bottom, then choose **Healing**. In Google Photos: tap **Edit → Tools → Magic Eraser**. In Apple Photos: tap **Edit**, then the brush icon (Clean Up). You'll see your photo with a brush you can paint over things. Use your finger to paint over the object or person you want to erase. Go slightly beyond the edges of what you're removing — a little overlap helps the AI understand what it needs to cover. - For a person: paint over their whole body, including their shadow if there is one. - For a small object like a sign or trash can: a rough paint stroke is usually enough. - For a power line or fence: paint along the full length of it in one stroke. Don't worry about being perfectly precise — the AI figures out the edges. Once you lift your finger (or tap the process button), the AI analyzes the pixels around the painted area and generates what the background should look like. This takes one to three seconds on most phones. On simple backgrounds — blue sky, green grass, a plain wall — the result is usually seamless. On complex backgrounds — a crowd, a patterned floor, overlapping branches — it may take a couple of tries. I've found the results are best when the object you're removing takes up less than about a quarter of the photo. Removing a whole person who's standing close to the camera is harder than removing someone in the distance. If the first try looks off, undo and paint a slightly different area — sometimes a small adjustment makes a big difference. Zoom in on the area where the object was. Look for: - Smearing or blurring where the fill happened - Repeated patterns that look unnatural (the AI sometimes copy-pastes nearby texture) - Edges that look cut-and-pasted If something looks wrong, tap **Undo** and try again. In Snapseed you can also zoom in and use a smaller brush for precision work on details. For tricky spots — complex backgrounds, large objects, overlapping shadows — the browser tool **Adobe Firefly** (firefly.adobe.com) gives you more control with its Generative Fill feature. Once you're happy, tap **Done** or **Save**. Most apps save a copy and keep the original intact — but check your app's settings if you're not sure, especially before removing someone from an irreplaceable photo. In Google Photos and Apple Photos, the edit is non-destructive: you can revert to the original at any time. In Snapseed, tap **Export → Save** to save a copy (the original stays untouched in your camera roll). ## When the AI Struggles Object removal works best on backgrounds that are relatively uniform — sky, grass, water, plain walls. It struggles when: - The object you're removing overlaps with another person or important detail - The background behind it is very complex or has strong patterns - You're trying to remove something very large that fills most of the frame In those cases, a more powerful tool like Adobe Photoshop or Canva Magic Edit (in a browser) gives you more control — you can manually paint in what should go there. ## What to try next Now that you can clean up your photos, you might want to explore the [full range of AI photo editors for non-designers](/guides/best-ai-photo-editors) to find the right tool for different editing jobs. Or if you have old family photos that need more than just object removal, [AI photo restoration](/guides/restore-old-photos-ai) can bring faded and damaged prints back to life. --- # AI and Your Pension: What AI Can Realistically Help You Understand Source: https://aiforregulars.com/guides/ai-retirement-pension-guide Pension statements can feel like they're written in a foreign language. Phrases like "projected annuity value," "defined contribution scheme," and "eligible rollover distribution" aren't designed for everyday reading. And retirement planning involves numbers that feel simultaneously important and impenetrable. AI turns out to be quite good at translating this kind of complexity into plain language. It's less good — and genuinely risky — when people treat it as a substitute for professional financial guidance. ## What AI Is Actually Useful For **Reading your pension statement.** If you receive an annual pension statement and aren't sure what the numbers mean, AI can walk you through it. Paste the text of the statement (omitting sensitive identifiers — see the caution below) and ask "Can you explain what each of these numbers means and what I should pay attention to?" **Learning the vocabulary.** Terms like "vesting schedule," "defined benefit vs. defined contribution," "required minimum distribution," and "catch-up contribution" are worth understanding before you make any decisions. AI explains these clearly and patiently, and will answer follow-up questions. **Rough scenario modeling.** You can ask an AI to help you think through general scenarios: "If I have approximately $200,000 saved and I plan to retire in 10 years, what are the general factors I should consider?" This is educational exploration, not personalized advice, and it can help you arrive at a real adviser conversation knowing what questions to ask. **Preparing for a meeting with an adviser.** Many people feel intimidated in financial appointments and leave without asking the questions that actually matter to them. AI can help you draft a list of specific questions based on your situation and vocabulary level. ``` Prompt example: "I have a 403(b) from my old employer and a small IRA from a different job. I'm 58 years old and plan to retire around 67. What questions should I ask a financial adviser about combining or managing these accounts?" ``` **Understanding Social Security or government benefit statements** ("Social Security" is the US example; outside the US, apply this to your own country's equivalent state pension or retirement benefit statement). These statements often include projected benefit estimates at different retirement ages. AI can help you understand what the numbers mean and what the tradeoffs of different claiming ages generally look like. I've used AI to prep for financial appointments and it genuinely changes the quality of the meeting. Instead of spending half the time on "what does that term mean?", I arrive with specific questions about my actual situation. The adviser gives me more useful answers because I'm asking better questions. ## Where AI Falls Short **It doesn't know your full financial picture.** A good financial adviser reviews your complete situation: income, debts, tax status, health, family situation, risk tolerance, other savings, Social Security timing, and goals. AI knows only what you tell it in the current conversation — and even then, it can't account for everything the way a trained adviser can. **It doesn't know your jurisdiction's rules.** Retirement and pension rules vary significantly by country, and within the US they vary by state, account type, and employer plan. Tax treatment of retirement withdrawals, required minimum distributions, survivor benefits — these are areas where jurisdiction-specific rules matter enormously. AI often gives accurate general information but misses the nuance that applies to your situation. **It has no legal accountability.** If a licensed financial adviser gives you bad advice that costs you money, you have legal recourse. If an AI chatbot gives you bad advice, you have no recourse. This is not a minor distinction — it reflects the fact that the adviser has skin in the game and the AI does not. **It can be confidently wrong.** AI chatbots sometimes state incorrect information in a tone of confidence. In a domain like retirement planning, where "I thought this was true" can have lasting financial consequences, that risk is significant. ## The Regulated-Advice Boundary: A Clear Line There is a clear line between using AI as an educational tool and using it as a financial adviser. On one side: understanding what things mean, learning vocabulary, modeling rough scenarios, preparing questions. On the other side: deciding how much to contribute, when to take Social Security, how to allocate investments, when to take withdrawals, and how to structure your estate. The second category requires a registered, licensed professional — ideally a fiduciary, meaning someone legally required to act in your best interest rather than their own. If you're not sure which side of the line a question falls on, apply this test: "Would the answer to this question require knowing my full financial situation and local tax laws to be actually useful?" If yes, that's an adviser question. ## Protecting Your Information When using AI to help understand pension or retirement documents, be careful about what you share: - Do not share full account numbers, Social Security numbers, or login credentials - Avoid uploading full statements with personal identifiers; paste specific sections instead - Be aware that conversations with AI services are typically stored on company servers - Do not share information about assets you don't want widely known The AI doesn't need your full identifying information to explain what a term means or help you understand a general scenario. ## Where to Find Real Financial Help If you need regulated advice and cost is a concern, there are lower-cost options: - **Fee-only financial planners** charge by the hour or flat fee rather than earning commissions. A single session focused on retirement questions can cost much less than ongoing advisory services. (In the US, directories like NAPFA.org or CFP.net can help you find one.) - **Nonprofit credit counseling agencies** sometimes offer retirement planning help for people within certain income limits. - **Your employer's HR department** may offer access to a retirement plan adviser as part of your benefits. - **State insurance and financial regulators** can tell you if a specific adviser is licensed and in good standing. ## What to try next For a related question in the legal domain, [Can AI Write Legal Documents for You?](/guides/ai-legal-documents-safe) covers the same "where does AI help, where does it stop" question for wills, contracts, and legal letters. If you're working through taxes, [AI for Taxes](/guides/ai-for-taxes) explains how AI can help you understand your tax situation before involving a professional. --- # AI Romance Scams: How Fake Chatbots Trick Online Daters Source: https://aiforregulars.com/guides/ai-romance-scam-dating-app Romance scammers used to spend months crafting messages by hand. Today, AI chatbots can carry on dozens of convincing conversations at once, around the clock, without ever getting tired or making a typo. And despite the name, these scams no longer start only on dating apps — the FTC reported that nearly 60% of people who lost money to a romance scam in 2025 said it began on a social media platform instead. If you use any dating app or social platform, understanding how these scams work could save you from real heartbreak — financial and emotional. ## How the Scam Unfolds Modern romance scams don't start with a desperate plea for money. They start with what feels like a genuine connection. **Week one — the match.** A profile appears with attractive photos (often stolen from a real person or AI-generated) and a compelling backstory: an engineer working overseas, a doctor with an international aid organization, a widowed entrepreneur. The first messages are warm and curious — questions about your life, your interests, your family. **Weeks two through four — the bond.** Messages arrive consistently, sometimes all day. They remember what you told them. They share personal-sounding stories. An AI model can be instructed to reference your previous messages and mirror your emotional tone, which makes the conversation feel surprisingly real. **The pivot — manufacturing a crisis.** Just as the bond feels solid, an obstacle appears. A sick child abroad. A business deal gone sideways. A customs fee to release equipment. The ask isn't always for money directly; sometimes it's a gift card "so my daughter can buy medicine," or an investment in their business so you can share in the profits. **The escalation.** If you pay once, the asks multiply. Each one has a reason why your money is temporarily stuck and just needs one more infusion to unlock. ## Tell-Tale Patterns to Watch For These signals don't guarantee a scam, but two or more together should make you slow down: - **They are always available.** Real people sleep, work, get busy. Bots respond at 3 a.m. with full, polished paragraphs. - **The profile is too perfect.** Reverse-image-search their photos using Google Images or TinEye. If the photos appear on other profiles or stock sites, that is a major warning sign. - **They avoid live video.** They will suggest audio calls, pre-recorded clips, or have technical problems every time you ask for a live video chat. Scammers know a live call would expose them immediately. - **The backstory keeps evolving.** Small inconsistencies pile up — the city they are from changes, their child's age shifts, details from last week don't match today. - **Emotional intimacy moves very fast.** Declarations of love within a week of matching, talk of a shared future before you have ever met — this is a tactic called love-bombing, designed to cloud your judgment. - **Any money request, however framed.** Gift cards, wire transfers, cryptocurrency, temporary loans — legitimate romantic partners don't ask people they have never met in person for financial help. ## Verification Tactics That Actually Work The single most effective thing you can do is insist on a live video call and pay close attention: **Ask them to do something spontaneous.** Wave at you, hold up three fingers, say a specific word you choose in the moment. Scripted deepfakes can't handle unpredictable prompts reliably. **Reverse-image-search their profile photos.** Right-click any photo and choose "Search image." A match on a different name or location is a strong red flag. **Search their name plus the word "scam."** Many scammers reuse the same stolen identities. Previous victims often post warnings online. **Never move to a private channel too quickly.** Scammers push to WhatsApp, Telegram, or email fast, partly because those platforms have weaker scam-reporting tools than dating apps. **Never send money or gift cards to someone you have not met in person.** This is the one rule with no real exceptions. It doesn't matter how long you have talked, how convincing the crisis sounds, or how much you have already sent — a legitimate partner does not need your money before you have met face to face. **Treat a successful video call as a good sign, not final proof.** A live call that goes well is the single strongest test in this guide, but real-time deepfake tools keep improving, and independent research on ordinary people's ability to spot manipulated video and audio has found accuracy far lower than most people expect — even when they know to look for it. If a video call goes fine but the conversation is still pushing toward money, verify through a separate channel (a call-back to a number you find yourself, not one they gave you) before you send anything. I have talked with people who felt embarrassed for nearly falling for one of these. There is nothing to be ashamed of — the technology behind these scams has genuinely improved. The people running them are professionals who do this full-time. Being skeptical early is not cynicism; it is self-care. ## Where to Report - **The dating platform itself** — every major app has a "report" button on profiles - **FTC (US):** reportfraud.ftc.gov - **FBI Internet Crime Complaint Center:** ic3.gov - **Your bank or card issuer** — immediately if money moved, even a small amount Reporting helps investigators identify patterns across thousands of victims and sometimes leads to takedowns of the networks running these operations. ## What to Try Next If you want to sharpen your ability to spot fake identities, [how to spot AI-generated photos](/guides/spot-ai-generated-photos) covers the visual tells in profile pictures that are easy to miss. And if a scammer has already tried a voice or video call, [AI voice cloning scams](/guides/ai-voice-cloning-scams) explains how that technology works and what you can do to stay ahead of it. ## Sources - [What To Know About Romance Scams](https://consumer.ftc.gov/articles/what-know-about-romance-scams) — Federal Trade Commission - [New FTC Data Show People Have Lost Billions to Social Media Scams](https://www.ftc.gov/news-events/news/press-releases/2026/04/new-ftc-data-show-people-have-lost-billions-social-media-scams) — Federal Trade Commission, April 2026 - [Report Fraud](https://reportfraud.ftc.gov) — Federal Trade Commission - [Internet Crime Complaint Center (IC3)](https://www.ic3.gov/) — FBI - [Search by image or video in Google Search](https://support.google.com/websearch/answer/1325808?hl=en) — Google - [TinEye — Reverse Image Search](https://tineye.com) — TinEye --- # New AI Tools That Block Fake Phone Calls Before They Start Source: https://aiforregulars.com/guides/ai-scam-call-blocker Scam calls have gotten much harder to ignore. The voices sound natural, the caller IDs look local, and some calls now use AI to clone a voice you recognize. The good news is that your phone — whatever kind you have — probably already includes tools to stop many of these calls before they even ring. This tutorial walks you through turning on every layer of protection available to you, in order from simplest to most thorough. You have up to three layers of scam call protection available: 1. **Built into your phone's operating system** — iOS and Android both have native call screening tools you can turn on right now 2. **From your phone carrier** — most major carriers offer free spam-call databases and blocking at the network level 3. **Third-party apps** — optional add-ons if you want extra control or a log of blocked calls You don't have to use all three, but each layer catches calls the others miss. Starting with what's already on your phone costs nothing and takes under two minutes. iPhone has two main tools for reducing scam calls. **Silence Unknown Callers** sends calls from numbers not in your contacts, email, or recent calls directly to voicemail — without ringing your phone at all. To enable it: 1. Open **Settings** 2. Tap **Apps** 3. Tap **Phone** 4. Tap **Screen Unknown Callers** (iOS 26) or **Silence Unknown Callers** (iOS 18) and select **Silence** or toggle it on If silencing all unknown callers feels too aggressive, a middle option is to let those calls ring but commit to only checking them after they go to voicemail. Real callers leave messages; scammers almost never do. Most Android phones include a **Caller ID and Spam** feature that automatically flags likely spam and scam numbers, based on reports crowdsourced from other users. On most Android phones: 1. Open the **Phone** app 2. Tap the three-dot menu and go to **Settings** 3. Look for **Caller ID and Spam** 4. Enable **Filter spam calls** — recognized spam is silenced automatically and moved to a separate list On Samsung phones, look for the same **Caller ID and spam protection** option in the same three-dot menu → **Settings**, and turn it on. There is also a **Call Screen** feature, where an AI assistant answers a call first and shows you a live transcript of what the caller says before you decide whether to pick up. This one is a Google Pixel feature, not a general Android one: the fully automatic version currently works only on Pixel phones in the US, and even the manual version is limited to a short list of countries — the US, Canada, UK, France, Germany, Italy, Spain, Ireland, India, and Japan — and needs a local SIM card. If you don't have a Pixel, or you're outside those countries, you won't find it in your settings; that's expected, not a mistake in the steps above. Your phone carrier maintains large databases of known scam numbers and can block or flag them before a call even reaches your device. Here's how to access this for major US carriers: - **AT&T** — ActiveArmor is available as a free app and blocks known fraud calls automatically - **Verizon** — Call Filter is free for basic protection; a paid tier adds a personal spam lookup feature - **T-Mobile** — Scam Shield includes free caller ID and scam-blocking features built into the account - **Other carriers** — Search your carrier's website for "call protection" or "spam blocking" These carrier tools work at the network level, so they can catch calls before they even touch your phone's operating system — a different layer from the on-device tools in steps 2 and 3. I set up my carrier's spam protection after a stretch of daily robocalls. Within about a week I noticed a clear drop in unwanted calls getting through. It took roughly three minutes to download the app and enable it. The built-in phone setting alone hadn't been enough — the carrier layer caught a different set of numbers. If you want more control — like a visual log of all blocked calls, or the ability to report new scam numbers yourself — free apps like **Hiya** or **Should I Answer** add a layer beyond what carriers provide. These apps work by crowdsourcing reports from millions of users, so they often catch newly active scam numbers faster than carrier databases get updated. One privacy note: these apps may upload your contact list or call log to their servers to provide caller ID. Review the app's privacy policy before installing if that matters to you. For numbers that keep getting through despite everything, you can block them one at a time directly on your phone: - **iPhone:** Open the Phone app, tap **Recents**, tap the info icon next to the number, scroll down and tap **Block this Caller** - **Android:** Open the Phone app, tap **Recents**, long-press the number, and tap **Block / report spam** This won't stop scammers who rotate through new numbers constantly, but it's effective against numbers that call repeatedly from the same source. ## What to Try Next Blocking calls before they reach you is the first line of defense. But knowing what to do when an AI-generated call does get through is just as important — [How to Tell If a Phone Call Voice Is AI](/guides/detect-ai-voice-phone-call) gives you six specific things to listen for. If you have an older family member who gets frequent calls, [The Grandparent Scam 2.0](/guides/grandparent-scam-ai) explains the specific tactics being used and includes a family safety plan you can set up together. ## Sources - [Manage unknown and spam callers on iPhone](https://support.apple.com/en-us/111106) — Apple - [Screen your calls before you answer them](https://support.google.com/phoneapp/answer/9118387?hl=en) — Google (Phone app) - [Use Caller ID & spam protection](https://support.google.com/phoneapp/answer/3459196?hl=en) — Google (Phone app) - [Hiya Call Blocker App | Advanced Caller ID & Spam Shield](https://www.hiya.com/products/apps/hiya-spam-blocker) — Hiya - [Should I Answer – App Store](https://apps.apple.com/us/app/should-i-answer/id1199812713) — Mister Group s.r.o. --- # AI Search vs Google: Which One to Use and When Source: https://aiforregulars.com/guides/ai-search-vs-google Search used to be simple: type words, get links, click through. Now every major search tool has an AI layer that writes answers for you. That's faster — but it also means you have to know which tool to trust for which job. Here's an honest look at three tools you're most likely to encounter: Google AI Overviews (part of regular Google Search), Perplexity AI, and ChatGPT Search. ## The three tools at a glance **Google AI Overviews** appear automatically at the top of many Google searches. You didn't sign up for them — they're just there. They draw on Google's massive index and are best for everyday quick questions you'd normally Google without a second thought. **Perplexity AI** (perplexity.ai) is a standalone search tool built specifically around AI answers. Its defining feature is numbered citations on nearly every factual claim, so you can verify what you're reading before you rely on it. **ChatGPT Search** is the web-search feature inside ChatGPT. You can trigger it by asking ChatGPT a question that requires current information, and it fetches real-time results. It's strongest as a conversational research partner that builds on itself. ## Side-by-side comparison The ratings below reflect each tool's publicly documented capabilities and widely-reported patterns from independent user reviews — not proprietary testing by this site. | Question type | Google AI Overviews | Perplexity AI | ChatGPT Search | |---|---|---|---| | Quick factual ("capital of Peru") | Excellent — fast, accurate | Good | Good | | Recent news (last 48 hours) | Good — Google News integration | Good | Good | | Research with citations | Fair — sources visible but not per-claim | Excellent — numbered inline citations | Good — links shown, less per-claim | | Medical / health questions | Fair — verify carefully | Fair — use Academic focus, still verify | Fair — verify carefully | | Shopping and prices | Good — Google Shopping integration | Fair | Fair | | Follow-up conversation | Limited | Good | Excellent | | Long research deep-dive | Fair | Good | Excellent | | Local businesses and maps | Excellent — Google Maps integration | Poor | Poor | ## When to use each tool **Use Google for:** - Finding a specific website (bank login, airline check-in, government forms) - Local business hours, addresses, and maps - Shopping comparisons — Google pulls real prices from retailers - News from the last 24 hours — Google's news index is hard to beat - Quick trivia and definitions when you don't need citations **Use Perplexity for:** - Questions where you want to see exactly where each fact came from - Health and science questions (use the Academic focus mode for peer-reviewed sources) - Comparing options on a topic — Perplexity gives clean, referenced summaries - Any time you want to trust but verify an AI answer without extra steps **Use ChatGPT Search for:** - Open-ended questions where you want to think through options - Research that requires follow-up questions and course corrections mid-conversation - Tasks that blend search with writing — "Find the current requirements and help me draft an application" - When you already have ChatGPT open and need a quick web check mid-conversation ## The honesty section: what all three get wrong No matter which tool you use, all three share the same fundamental limitation: they can state wrong information confidently. This happens most often with: - **Specific numbers** — prices, statistics, distances, dosages - **Very recent events** — anything from the last few days - **Niche topics** — anything outside mainstream coverage - **People's quotes** — AI tools sometimes generate plausible-sounding quotes that were never actually said The best defense is the same for all three: if the answer matters, click a source and read it yourself. I ran informal comparisons over several weeks asking the same questions across all three tools. Google AI Overviews won on speed and local queries. Perplexity felt most trustworthy for health questions because I could see a source for every claim. ChatGPT Search was where I ended up for anything I needed to think through — it's the best at follow-ups that build on each other. I now switch between them depending on the task rather than picking one as "the best." ## A simple decision rule Before you open any search tool, ask yourself one of these questions — whichever matches your need: - Do I need to find a specific website, business, or map? → **Google** - Do I want to verify every key fact with a source link? → **Perplexity** - Do I need to ask five follow-up questions to reach my real answer? → **ChatGPT Search** When none of those fit, start with whichever you already have open. ## What to try next If you want to go deeper on either tool: [How to Use Perplexity AI](/guides/how-to-use-perplexity) walks you through Perplexity step by step, and [Google AI Mode Explained](/guides/google-ai-mode-explained) covers everything you need to know about what Google changed in Search. --- # Can an AI Agent Really Shop and Pay for You? What's Real Right Now Source: https://aiforregulars.com/guides/ai-shopping-agent-safety The idea of telling an AI assistant "find me a good pair of walking shoes under $60 and just order them" sounds like science fiction — but small pieces of it are already real. The bigger truth, though, is that the flashiest version of this already came and went: a major AI company built an in-chat "buy it now" button, launched it widely, and quietly shut it down about six months later. That's worth knowing before you hand any AI agent your payment details. This guide separates what AI shopping agents can actually do today from what's still marketing, and gives you a short, practical checklist for staying safe if you decide to try one. Almost every AI tool that talks about "shopping" is doing one of two very different things, and it's worth knowing which one you're using: - **Product discovery** — the AI searches, compares prices, and suggests options, then hands you off to the retailer's own website or app to actually pay. This is the overwhelming majority of what "AI shopping" means today, including ChatGPT's current shopping experience. - **Completed payment inside the AI tool** — the AI actually finalizes the purchase using a payment method you've linked. This is rare, and every real example of it that we could verify still asks you to confirm the final order before money moves. If a company advertises an AI "shopping assistant," assume it means discovery unless it specifically says it can complete a purchase — and even then, expect a confirmation step. No mainstream tool available to ordinary consumers right now buys things fully on autopilot with zero review from you. In September 2025, OpenAI launched a feature called Instant Checkout directly inside ChatGPT, built on something called the Agentic Commerce Protocol, developed jointly with Stripe. It started with Etsy sellers, with Shopify merchants set to follow. The pitch was simple: ask ChatGPT a shopping question, see products, tap buy, and finish the purchase without ever leaving the chat window. By March 2026 — about six months later — OpenAI quietly retired it. Results were weak: Walmart reportedly saw roughly a third of the conversion rate it gets on its own website, and only around 30 Shopify merchants had bothered to sign up. ChatGPT now works purely as a "product discovery" tool — it helps you find and compare items, then sends you to the retailer's website to pay, just like a search engine would. The lesson isn't that AI shopping is a dead end. It's that even a well-funded, well-known company's first serious attempt at an AI "buy button" didn't survive contact with real shoppers. Don't assume any single AI checkout feature you encounter is mature or permanent — treat all of them, for now, as experiments. The clearest working example right now is Perplexity, the AI answer engine, which partnered with PayPal starting in the summer of 2025. Ahead of Black Friday in November 2025, the two companies launched a feature called Instant Buy: inside Perplexity — including its Comet web browser — you can ask it to find a product, and it can complete checkout using a PayPal or Venmo account you've already linked, at participating retailers that reportedly include brands like Abercrombie & Fitch, Ashley Furniture, Fabletics, Adorama, and Newegg. What matters most for your safety isn't the brand name — it's the design pattern. Perplexity's Comet browser can navigate a retailer's site and build a shopping cart on your behalf, but it's built to stop short of completing payment without showing you the final order and asking you to confirm it. That "pause and confirm" step is the single most important safety feature in any AI checkout tool. If you ever encounter one that skips it, treat that as a red flag rather than a convenience. Here's the part that matters more than any single company's feature: security researchers have started documenting what's being called "agentic commerce fraud" — fraudulent storefronts specifically engineered to look legitimate to an AI agent's scanning and product-comparison process, not necessarily to a human eye. If an AI agent completes a purchase at one of these fake stores, the scammers can harvest your payment details the same way a phishing site would. This matters because an AI agent evaluates a "good deal" differently than you do. It might weigh price, apparent reviews, and product-description keywords in ways a convincing fake storefront can game — sometimes more easily than it could fool a skeptical human shopper who notices a site "feels off." The other open question is liability. If an AI agent gets tricked into an unauthorized or fraudulent purchase, who's responsible — you, the AI company, the payment network, or the merchant? Industry surveys from 2026 found only about 16% of consumers say they're comfortable letting an AI agent spend money autonomously even with permission granted, and over half believe AI increases fraud risk overall. That skepticism is reasonable: the rules for exactly this scenario are still being written. **What this means for you in practice:** your normal card-issuer fraud protections and dispute rights generally still apply to a purchase made through an AI agent, because the transaction still runs through your regular card network like any other charge. But because AI-specific liability rules aren't fully settled industry-wide, don't assume a dispute involving an AI agent will move as smoothly or quickly as an ordinary one. Keep every confirmation and receipt. I tried Perplexity's shopping feature out of curiosity rather than need, and the thing that stood out was how normal the final step felt — it showed me the exact item, seller, and price and waited for me to tap confirm, the same as any checkout page. That's reassuring, but it also means the safety of the whole system depends on that confirmation screen always showing up. I'd treat any AI shopping feature that skips it as untrustworthy by default. If you decide to try any AI shopping or checkout feature, a few habits go a long way: 1. **Use a dedicated, low-limit payment method first.** A prepaid card, a virtual card number, or a card you don't use for anything else limits your exposure while you learn how the feature actually behaves. 2. **Always review the exact cart before confirming.** Check the item, price, seller name, and shipping cost yourself — don't tap "confirm" on autopilot just because the AI presented it as a good match. 3. **Be suspicious of unusually low prices from an unfamiliar seller**, even when the AI flagged it as a top result. AI agents can be fooled by fake storefronts the same way humans can be fooled by phishing sites — sometimes more easily. 4. **Treat "no final confirmation screen" as a warning sign.** The safest AI shopping tools available today, like Perplexity's Comet, are built to pause and show you the order before paying. A tool that skips that step deserves extra caution, not extra trust. (If you're building your own agent rather than using an off-the-shelf shopping tool, the same pause-and-approve pattern can be added with a service like [Impri](https://impri.dev) — disclosure: built by the author of this site.) 5. **Keep your receipts and confirmations.** Your card issuer's dispute process still applies, but a paper trail makes any dispute involving an AI-initiated purchase easier to resolve. ## What to Try Next If you want to explore AI shopping tools that stop at the discovery stage — no payment involved at all — [AI Shopping Tools That Find Better Prices While You Browse](/guides/ai-shopping-tools) covers the safer, more established options. And since any AI shopping conversation involves sharing some personal preferences and sometimes payment details, it's worth reading [What Not to Tell an AI Chatbot](/guides/what-not-to-tell-ai-chatbot) before your next session. ## Sources - [OpenAI revamps shopping experience in ChatGPT after struggling with Instant Checkout offering](https://www.cnbc.com/2026/03/24/openai-revamps-shopping-experience-in-chatgpt-after-instant-checkout.html) — CNBC - [PayPal and Perplexity Launch Instant Buy Ahead of Black Friday](https://newsroom.paypal-corp.com/2025-11-PayPal-and-Perplexity-Launch-Instant-Buy) — PayPal Newsroom - [Perplexity announces free product to streamline online shopping](https://www.cnbc.com/2025/11/19/perplexity-ai-online-shopping-paypal.html) — CNBC - [Mastercard Unveils Agent Pay, Pioneering Agentic Payments Technology to Power Commerce in the Age of AI](https://newsroom.mastercard.com/news/press/2025/april/mastercard-unveils-agent-pay-pioneering-agentic-payments-technology-to-power-commerce-in-the-age-of-ai) — Mastercard Newsroom - [Mastercard Debuts Agent Pay to Promote 'Agentic Commerce' Future](https://www.pymnts.com/mastercard/2025/mastercard-debuts-agent-pay-to-promote-agentic-commerce-future/) — PYMNTS - [Visa, Mastercard expand agentic AI deployments](https://www.americanbanker.com/payments/news/visa-mastercard-expand-agentic-ai-deployments) — American Banker - [Agentic Commerce: Threats and Risks](https://corporate.visa.com/en/sites/visa-perspectives/security-trust/the-threats-landscape-of-agentic-commerce.html) — Visa - [AI Shopping Agents Pose Novel Liability, Authorization Risks](https://news.bloomberglaw.com/banking-law/ai-shopping-agents-pose-novel-liability-authorization-risks) — Bloomberg Law --- # Amazon Rufus and AI Shopping: How to Use AI to Buy Smarter Source: https://aiforregulars.com/guides/ai-shopping-amazon-rufus Online shopping can feel overwhelming. Hundreds of products, mixed reviews, confusing specs — and you just need a good blender. Amazon's AI assistant Rufus lets you ask shopping questions the same way you'd ask a knowledgeable friend, right inside the app you probably already use every day. ## What is Amazon Rufus? Rufus is Amazon's built-in AI assistant, available in the Amazon Shopping app and on amazon.com for all US customers. Instead of typing keywords into a search bar, you can ask full questions: "What coffee maker is easiest to clean?" or "I need a birthday gift for a 7-year-old who loves dinosaurs." Rufus reads product details and customer reviews to give you a real answer. Open the Amazon shopping app on your phone or tablet. Look for a small chat bubble or sparkle icon near the search bar at the top of the screen. Tap it to open the Rufus chat panel. If you don't see it yet, make sure your app is updated to the latest version. Type your question as if you're texting a friend. You don't need product codes or category names. Good examples: ``` What's the best robot vacuum for a small apartment with a cat? ``` ``` I need noise-canceling headphones under $80 that work with Android. ``` ``` What's the difference between an air fryer and a convection oven? ``` Rufus will respond with product suggestions and a brief explanation of why they fit. You don't have to start over every time. Rufus remembers your conversation, so you can refine: ``` Which of those is easiest to set up for someone who isn't tech-savvy? ``` ``` Does the first one come in black? ``` ``` What do customers complain about most with that model? ``` That last question is especially useful. Asking about complaints often reveals real-world downsides that the product listing glosses over. Once you have a short list, ask Rufus to compare them directly. Tap a product to open it, then use the Rufus chat to ask: ``` Compare this to the [other product name]. Which one lasts longer? ``` ``` What's the main difference between this model and the one you suggested earlier? ``` Rufus will pull key specs and review highlights side by side, saving you the work of opening multiple tabs. Product pages can have thousands of reviews. Ask Rufus to do the reading for you: ``` What are people saying about the battery life? ``` ``` Are there complaints about it breaking after a few months? ``` ``` Is this good for older adults who don't like complicated gadgets? ``` Rufus summarizes patterns it finds across real reviews — not just the five-star ones. AI assistants can make mistakes. Before you add anything to your cart, scroll down the actual product listing to verify the key details Rufus mentioned — return policy, warranty, what's included in the box. If the numbers don't match, trust the product page. I've found Rufus most useful when I ask about downsides, not upsides. Sellers write glowing product descriptions, but Rufus will tell you "many reviewers mention the lid leaks" or "setup requires a smartphone app, which some older buyers found frustrating." That honest summary has saved me a few bad purchases. Gift shopping is where Rufus really shines. Instead of browsing aimlessly, ask: ``` What's a good gift for a 70-year-old woman who loves gardening? Budget around $50. ``` ``` My dad is learning to play guitar. What accessories would help a beginner? ``` ``` What are popular gifts for a new baby that parents actually find useful? ``` You'll get a focused list instead of thousands of unfiltered results. ## A few things Rufus can't do Rufus only searches Amazon's catalog — it won't compare prices with other stores. It also relies on the reviews Amazon has, which can include fake or incentivized reviews. For big-ticket purchases, consider cross-checking with an independent review site before deciding. ## What to try next Now that you know how to shop smarter with Rufus, you might want to explore the other AI tools built into your everyday apps. Check out [all the AI tools Amazon and other stores now offer](/guides/ai-shopping-tools), or learn [what else you can ask an AI chatbot to help with](/guides/things-to-ask-chatgpt) — the uses go well beyond shopping. --- # AI Shopping Tools That Find Better Prices While You Browse Source: https://aiforregulars.com/guides/ai-shopping-tools Shopping online can feel like a lottery — you never know if you're getting the best price. AI-powered browser extensions have changed that. They quietly work in the background while you browse, comparing prices across stores and applying coupon codes at checkout automatically. Here are the best free tools to try — each does something slightly different, so the right pick depends on how you shop. ## The Tools at a Glance | Tool | What It Does Best | Works On | Free? | |------|-------------------|----------|-------| | PayPal Honey | Finds and applies coupon codes at checkout | 30,000+ sites | Yes | | Capital One Shopping | Price comparison + price-drop alerts | Major retailers | Yes | | Google Shopping | Built into Google search results | Any product search | Yes | | Rakuten | Cashback on purchases | 3,500+ stores | Yes | | CamelCamelCamel | Amazon price history tracker | Amazon only | Yes | ## PayPal Honey — Best for Coupon Codes Honey is a browser extension that activates automatically on checkout pages. When you're about to pay, it tries dozens of coupon codes in seconds and applies the best one. You do nothing extra. It works on a wide range of sites beyond Amazon — clothing stores, electronics, travel booking, and food delivery. The success rate varies: on some sites it finds nothing, on others it can knock 10–20% off. **Get it at:** honey.com (free Chrome, Firefox, Safari, and Edge extension) I've had Honey apply a discount code I never would have found myself — it showed up right at the moment I was about to pay. Even when it finds nothing, the attempt takes two seconds and costs you nothing. ## Capital One Shopping — Best for Price Comparisons Capital One Shopping shows you the same item cheaper at other stores while you're viewing a product page. A small popup appears with competing prices so you can decide whether to switch before you buy. You don't need to be a Capital One customer — anyone can install and use it for free. It also offers price-drop alerts. Save an item and get a notification when its price falls, which is useful for big purchases you're not ready to make just yet. ## Google Shopping — No Extension Needed Google Shopping is built into regular Google searches. Search for any product and you'll see a "Shopping" tab with prices from multiple retailers. Click one and you go directly to the store. It's not as proactive as a browser extension — you have to search deliberately — but it's the easiest option for someone who doesn't want to install anything extra. ## Rakuten — Best for Cashback Rakuten works differently from the others. Instead of comparing prices or finding coupons, it earns you cashback on purchases you make at partner stores. The cashback pays out quarterly as a check or PayPal transfer. Activate Rakuten before shopping at a supported store, and the cashback is tracked automatically. Over a year of regular online shopping, this can add up to a meaningful amount. ## CamelCamelCamel — Amazon Price History This one is Amazon-specific, but it solves a real problem: knowing whether a sale price is actually a good deal. Amazon changes prices constantly, and a "30% off" badge doesn't always mean the price is at a genuine low. CamelCamelCamel shows you a price history chart so you can see what the item normally sells for. You can also set an alert for when the price drops below a target you choose. ## How to Choose - **Want the least effort?** Install Honey. It runs silently and only pops up at checkout. - **Buying something on Amazon?** Use CamelCamelCamel to check whether the sale is genuine. - **Regular online shopper?** Sign up for Rakuten to earn cashback alongside whatever coupon tools you use. - **Researching a big purchase?** Use Google Shopping first to scan prices across stores. ## What to Try Next Once you've got your shopping tools set up, [ChatGPT for budgeting](/guides/chatgpt-monthly-budget) can help you track whether you're actually spending less overall. If you're planning a trip and want to comparison-shop travel deals, [AI vacation planning](/guides/plan-vacation-with-chatgpt) shows how to research and compare options the same way. --- # AI Smart Glasses Explained: What They Actually Do, and the Privacy Risks to Know Source: https://aiforregulars.com/guides/ai-smart-glasses-explained AI smart glasses are no longer a novelty gadget. Meta's Ray-Ban line has sold well enough that you've probably already been near someone wearing a pair without realizing it, since they look almost exactly like ordinary sunglasses. This guide explains what they actually do, what they cost, and the privacy questions around recording that have been making headlines through 2026. ## What AI Smart Glasses Actually Do The best-known model, Ray-Ban Meta, is a normal-looking pair of glasses with a built-in camera, microphones, small speakers, and Meta's AI assistant. Say "Hey Meta" and you can: - Take a photo or record video hands-free - Ask what you're looking at ("what kind of plant is this?") and get a spoken answer - Get a menu, sign, or document translated out loud - Make calls, play music, or livestream to Instagram or Facebook None of this requires touching the glasses or pulling out your phone, which is the whole appeal: it's a camera and an assistant you're already wearing. ## What They Cost Ray-Ban Meta prices vary by model. The standard camera-and-AI glasses start around $299–$379 depending on the frame and lenses, with roughly 8 hours of battery life for general use and a charging case that stretches total use to a day or more between charges. A premium version called Ray-Ban Meta Display costs $799 and adds a small screen built into the right lens for reading messages and previewing photos, controlled by a wristband ("Neural Band") that senses hand and finger movement instead of a touchscreen or voice command. That model is only sold after an in-person fitting at a store, since it needs to sit correctly on your face to work. ## The Privacy Problem: A Blinking Light Is the Only Warning The glasses have a small LED that lights up while recording, meant to let people around you know they might be on camera. In practice, it has real limits: it's easy to miss, especially outdoors in bright light, and for a while it could be disabled with cheap third-party mods sold online. In July 2026, Meta updated the glasses so the camera automatically shuts off if that light is covered or tampered with, and said it is pulling marketplace listings for LED-disabling accessories and may ban accounts or take legal action against people who sell them. Privacy groups have pushed back that a blinking light is still a weak substitute for actually asking someone before you record them. ## The "Pervert Glasses" Trend That Made Headlines In late July 2026, reporting surfaced that some users, aspiring influencers and self-styled "pickup artists" among them, were using the glasses to secretly film unsolicited interactions with women in public and post the footage online, earning the glasses an ugly nickname on social media. Instagram head Adam Mosseri said the platform would remove that kind of content and confirmed some creator accounts had already been taken down for it. Separately, a UK convention organizer banned the glasses from its events entirely after attendees raised concerns about being filmed without consent. If you're weighing whether to get a pair for yourself, think of them less like a fun gadget and more like an always-with-you camera, because that's what they are. The genuinely useful part for most people is hands-free photos on a walk or a trip, and quick answers about what you're looking at. The part to be deliberate about is that anyone near you may not realize they can be recorded, so treat the same courtesy you'd want from someone else: ask before you record a person, and don't rely on the light to do that job for you. ## Is It Legal for Someone to Record You? It depends on what's being captured, not on the device. Silent video of you in a public place, like a street or a store, is generally legal in the US, the same as showing up in the background of a stranger's photo. Audio is treated differently, because it's covered by state wiretapping and eavesdropping laws: in most states, only one person in a conversation needs to consent to it being recorded, but twelve states, California, Connecticut, Delaware, Florida, Illinois, Maryland, Massachusetts, Montana, New Hampshire, Oregon, Pennsylvania, and Washington, require everyone's consent. Secretly recording a private conversation in one of those states can carry criminal and civil penalties, regardless of whether the recording device is a phone, a hidden recorder, or a pair of glasses. And no matter the state, recording someone in a place they'd reasonably expect privacy, a bathroom, a changing room, someone's home, is illegal everywhere under voyeurism laws. ## Where Does the Video Actually Go? When you use an AI feature on the glasses, the footage is sent to Meta's app and servers for processing, since the glasses don't run AI on the device itself. Reporting has found that some of that footage can be reviewed by human workers as part of training Meta's AI systems. If keeping your captured video off Meta's servers matters to you, check the privacy and cloud-processing settings inside the Meta AI companion app, and be selective about when you actually trigger an AI feature versus just taking a photo you plan to keep locally. ## What to Watch Out For **A recording light is not consent.** Even where the light works as intended, most people don't register what it means at a glance. If you're wearing the glasses, the safer habit is to tell people directly when you're about to record them, especially indoors or in a personal conversation. **"Everyone can see the light" is not a legal defense.** In states that require all-party consent for audio, a blinking LED doesn't substitute for actually asking. If you're recording a private conversation, get a clear yes first. **Never record in private spaces.** Bathrooms, changing rooms, and other places people expect privacy are off-limits everywhere, no exceptions for a "visible" recording light. **If you think you were secretly recorded and shared without consent**, most platforms, including Instagram, have policies against non-consensual filming and harassment content, and you can report it directly through the app. ## What to Try Next If your bigger concern is a smart speaker or phone listening in at home, [how always-listening voice assistants actually work](/guides/alexa-always-listening-privacy) breaks down what's really being recorded and when. And if you're curious what your existing phone's camera can already do with AI, [the AI camera features hiding in your phone](/guides/ai-camera-features-phone) covers the same hands-free tricks without buying anything new. --- # What Does the AI on Your New TV Actually Do? Source: https://aiforregulars.com/guides/ai-smart-tv-explained TV marketing has discovered the word "AI" in a big way. Walk through any electronics store and you will see it everywhere: AI Upscaling, AI Sound Pro, AI Processor, AI Remote. Most of these labels are real features, not just stickers — but they mean different things. Here is a plain-language breakdown. ## AI Upscaling: Making Older Content Look Sharper When you watch something filmed or recorded in lower resolution — an old movie, a show from the 1990s, or a YouTube video — your 4K TV has to stretch that image to fill millions of pixels. Without help, stretched images look soft or blurry. AI upscaling uses a neural network trained on millions of image pairs. It predicts what a high-resolution version of each frame *should* look like and fills in the extra detail on the fly. The result is a sharper, less grainy picture — especially visible on large screens from a normal viewing distance. Different brands call this the same thing with different names. Samsung calls theirs the Neural Quantum Processor. LG calls theirs the Alpha AI Processor. Sony calls theirs the Cognitive Processor XR. The underlying approach is similar across all of them. **Does it work?** Yes. The improvement is most obvious on standard-definition content on a big screen. On already-sharp 4K content, the difference is subtle. I tested this by watching the same episode of a 2003 TV show on an older set and a new one with AI upscaling side by side. The difference was immediate — faces and backgrounds had a crispness on the new TV that looked closer to modern production quality. It does not add detail that was never there, but it makes the best guess and usually gets it right. ## AI Sound: Audio That Adjusts Itself Most TV speakers struggle with dialogue — it gets buried under music and sound effects. AI sound processing analyzes the audio in real time and decides which parts are speech, music, and background noise. It then adjusts each independently, turning up dialogue and keeping the rest balanced. Some TVs also adapt audio based on what is on screen. If a quiet nighttime scene suddenly cuts to an action sequence, the TV can soften the sudden jump in volume before it wakes up the house. Look for names like AI Sound, Sound Optimizer, or Adaptive Volume in your TV's settings. These are usually on by default, but you can dial them back if you prefer the raw audio mix. ## AI Recommendations: What to Watch Next Every major streaming TV platform — Roku, Fire TV, Google TV, Apple TV, Samsung Tizen, LG webOS — uses a recommendation algorithm. This is sometimes called AI because it learns from your watch history over time. The quality varies a lot. Google TV's recommendations tend to be more accurate because Google has years of data about your interests from Search and YouTube. Roku's recommendations are partly driven by which apps pay for placement. You can usually reset or limit this data in Settings > Privacy. ## Conversational Remotes: Talking to Your TV High-end TVs now ship with remotes that have a microphone button. Press and hold it, ask a question, and the TV responds. Depending on your TV, that could be Google Assistant, Alexa, or the brand's own assistant. You can say things like "Show me action movies on Netflix," "Turn up the brightness," or "What time does this channel show the news?" The TV responds in text or voice and acts on the request. A useful detail: most AI remotes also work as a universal remote for your soundbar, cable box, or streaming stick, since the TV can learn those device's commands too. ## What "AI Processor" Actually Means You will see "AI Processor" listed as a spec. This refers to a dedicated chip in the TV that runs the upscaling and sound algorithms without slowing down the main processor. The practical effect is smoother, lag-free performance even when upscaling fast-moving content. Higher processor tiers (like Samsung's Neo Quantum Processor versus their base Quantum Processor) run more layers of the neural network and handle the upscaling more accurately. For most viewers in a living room at normal distance, mid-tier processors are already very good. ## What to try next If you want to go deeper on voice assistants beyond your TV, [Alexa vs. Google vs. Siri: Which Assistant Is Best?](/guides/alexa-google-siri-comparison) is a helpful comparison. And for a broader look at what AI actually means as a term, [What Does AI Mean? A Plain-English Explanation](/guides/what-does-ai-mean) is a short, jargon-free read. ## Sources - [How to turn off smart TV snooping features](https://www.consumerreports.org/electronics/privacy/how-to-turn-off-smart-tv-snooping-features-a4840102036/) — Consumer Reports - [VIZIO to pay $2.2 million to FTC, State of New Jersey to settle charges it collected viewing histories without consent](https://www.ftc.gov/news-events/news/press-releases/2017/02/vizio-pay-22-million-ftc-state-new-jersey-settle-charges-it) — Federal Trade Commission --- # AI Tools for Adults Going Back to School or Retraining Source: https://aiforregulars.com/guides/ai-tools-adult-learners Going back to school as an adult is already hard. You are balancing work hours, family time, and the feeling that everyone else already knows the material. AI will not fix the time problem — but it can cut down on how long it takes to do the work you have to do anyway. Here are the tools and techniques that make the biggest difference for adult learners specifically, organized by what they help you do. ## Note Summarization: NotebookLM Google NotebookLM is the best tool most adult learners have never heard of. You upload your syllabus, textbook chapters, or lecture notes, and it becomes a private AI assistant that only draws on your actual course material. What you can do with it: - Ask "What are the three most important ideas in this chapter?" and get a focused summary - Ask it to explain a concept using only what is in your uploaded notes - Generate a study guide from your own materials before an exam - Ask it questions the way you would ask a knowledgeable study partner The key advantage: it does not pull from the open internet. It works from your documents, so the answers stay grounded in what your course actually covers. For professional certifications or graduate work with specific required texts, this matters a lot. Read the [full NotebookLM guide](/guides/google-notebooklm-guide) for setup instructions — it takes about five minutes to upload your first document and start asking questions. ## Flashcards: AI-Generated Memorization Spaced repetition is still the most effective memorization method, and AI can now generate the flashcards for you from your own notes. **Option 1 — Generate in ChatGPT and add manually:** ``` Create 15 flashcard pairs from this text. Format each as: Front: [term or question] | Back: [definition or answer] [paste your notes or textbook section] ``` Then add them to Anki, Quizlet, or any flashcard app. Takes a few minutes to set up but saves hours of card-writing. **Option 2 — Use a dedicated tool:** Apps like Quizlet have added AI generation features that create cards from pasted text or uploaded documents directly. Adult learners often have limited study time and cannot afford inefficient memorization. Flashcard generation from your actual course notes is one of the fastest wins AI offers. ## Essay Feedback: Use AI as a Writing Coach Many adult learners feel rusty about academic writing. AI is an effective writing coach — as long as you ask it to comment on your draft rather than write it for you. A prompt that works well: ``` I'm an adult student returning to school. Here is a paragraph from an essay I'm working on. I want to know: Is my main point clear? Is there anything that sounds unclear or too informal? What would a college professor likely want me to add or change? Do not rewrite it for me — give me specific feedback only. [paste your paragraph] ``` If your instructor uses a grading rubric, paste it into the same message and ask the AI to evaluate your paragraph against each criterion. This is a much faster feedback loop than waiting for a graded paper to come back. Returning students often underestimate their own writing instincts. The AI feedback I found most useful was not "this is bad" — it was "here is a sentence that might confuse a reader who has not read what you have read." That is the exact thing a busy professor might not have time to explain in office hours. ## Concept Explanation on Demand Adult learners often cannot sit in office hours or wait for classmates to reply in a forum. AI is available at 11pm when you finally sit down to study. Use it like this: ``` I'm studying [subject] at the [community college / graduate] level. I don't understand [concept]. Can you explain it starting from basics and building up? Use a concrete analogy if it helps. ``` For STEM subjects in particular, being able to ask "but why does that formula work?" and get a patient, step-by-step answer is genuinely valuable. You can ask the same question five different ways without embarrassment until something clicks. ## Study Scheduling for Complicated Lives Adult learners have time constraints that traditional student schedules do not account for. A study schedule for someone with two kids and a part-time job looks nothing like a schedule built for someone living in a dorm. ``` I'm an adult student taking [number] credits while working [X hours] per week and managing family responsibilities. I have exams in [X weeks]. I typically have [30/60/90] minutes in the evening and [X hours] on weekends. Build me a realistic study schedule for the next two weeks. Include buffer time for things that come up. ``` The buffer time request matters. AI tends to build tight schedules if you do not specify otherwise, and adult learners know from experience that things come up. ## What to Try Next If you are using AI-assisted study to prepare for a career change, the [guide on using AI for job searching over 50](/guides/ai-job-search-over-50) covers how to use the same tools for the job search once you have your credentials. And if you want to sharpen how you phrase requests to any AI tool so you get better answers faster, [the guide to writing effective prompts](/guides/how-to-write-prompts) pays off quickly. --- # AI Tools for People Caring for an Aging Parent (That Actually Help) Source: https://aiforregulars.com/guides/ai-tools-for-caregivers If you're caring for an aging parent while also managing a job, your own family, and everything else — you already know the job is mostly paperwork and coordination. The hospital visit is exhausting; the discharge papers that follow can be just as hard to navigate. According to AARP, roughly 53 million Americans provide unpaid care to an older adult, and the administrative burden is one of the top reasons caregivers burn out. AI won't replace a good doctor or a real support network, but it can take a real bite out of the hours you spend deciphering, drafting, and organizing. This guide is for the adult child managing a parent's care — not for the senior themselves. For technology designed to help your parent directly, see [AI Companions for Seniors](/guides/ai-companions-seniors) and [AI Home Monitoring for Elderly Parents](/guides/ai-home-monitoring-elderly). ## 1. Translating Hospital Discharge Papers Into Plain English Discharge papers after a hospitalization are written for clinicians, not families. They're full of abbreviations, Latin drug names, and instructions like "ambulate with supervision as tolerated" that mean nothing if you're just trying to know whether your mom can walk to the bathroom on her own. This is one of the most genuinely useful things AI can do for caregivers right now. **What to do:** Open ChatGPT, Gemini, or Claude and type something like: > "My father was just discharged from the hospital after hip surgery. The discharge instructions say he should 'avoid weight-bearing on the left lower extremity for six weeks.' Can you explain what this means in plain English, and list the main things a family caregiver should watch for during recovery?" You can also retype specific sentences from the discharge papers that confused you and ask the AI to explain each one. You don't need to photograph or paste the entire document. **What to do with the result:** The AI's explanation gives you a starting point for questions — write them down and ask the nurse or doctor at the follow-up visit to confirm. Don't treat the AI's interpretation as the final word on medical instructions. ## 2. Drafting Sibling Update Emails and Care-Coordination Messages If you're the primary caregiver, you've probably found yourself on the phone three times in a row explaining the same situation to three different family members. AI can help you write a single, clear update email you can send to everyone at once. **What to do:** Ask AI to help you build a short template: > "Help me write a weekly update email I can send to my siblings about our mother's health. She has dementia and lives in assisted living. I want to cover: how she seemed this week, any medical updates, upcoming appointments, and one thing people can do to help. Keep it warm but practical, under 200 words." Once you have a template you like, you fill in the current week's details each time. AI can also help you draft harder messages — telling a sibling that a situation has gotten more serious, or writing formally to a nursing facility to request a care plan meeting. ## 3. Organizing Medication Schedules Across Multiple Providers When a parent sees three or four specialists, each may prescribe something, and the medication list can grow quickly. AI can help you turn a messy collection of pharmacy printouts and prescription bottles into a clean, organized reference. **What to do:** Describe the medications in plain text (or retype them from a pharmacy printout — just leave out your parent's name and date of birth): > "My mother takes the following medications. Can you organize these into a morning, evening, and as-needed table, and flag any combinations I should ask the pharmacist about? [list the medications and doses]" The AI can surface potential interaction questions worth raising — but always have a pharmacist or prescribing doctor review the final list. AI doesn't have your parent's full medical history. For appointments, AI can also help you create a simple tracking template or generate a list of questions to bring to each upcoming visit. What I do before any doctor's appointment for a family member: I spend five minutes describing the current situation to an AI and asking it to generate a list of questions I might not have thought to ask. Doctors see dozens of patients a day and the visit moves fast — having a written list means you don't walk out realizing you forgot to mention the most important thing. ## 4. Preparing Insurance Paperwork and Appeals Insurance paperwork is another place AI can genuinely help, because the core task is writing — explaining a situation clearly and formally. If a claim was denied, AI can help you draft an appeal letter. **What to do:** > "My father's insurance denied coverage for a skilled nursing facility stay after his hospital discharge. The reason given was 'not medically necessary.' Help me draft a formal appeal letter. The discharge papers say he needs daily wound care and physical therapy. Keep the tone professional." You fill in the specific claim number, dates, and your parent's name. The structure, formal language, and knowing which points to include are where AI saves you real time. AARP has published guidance on using AI tools as a caregiver resource and offers sample appeal letters that work well alongside whatever AI helps you draft. ## 5. Finding Respite Resources Near You Respite care means temporary coverage so the primary caregiver can rest — whether that's a few hours a week or a few days while you travel. Finding it can feel like a full-time job on its own. AI can help you understand the landscape — what types of respite exist (adult day programs, in-home aides, short-term nursing facility stays), what Medicaid waiver programs typically cover, and what questions to ask when you call a provider. But AI often has outdated information about specific local programs, names, phone numbers, and current availability. **Better sources for actual local services:** - **Eldercare Locator** (eldercare.acl.gov) — a free US government referral service - Your local **Area Agency on Aging** (find it through the Eldercare Locator) - Your state's **Medicaid HCBS waiver programs** if your parent qualifies Use AI to prepare for those calls: "What questions should I ask an adult day program to find out if it's right for someone with mild dementia?" is exactly the kind of prompt AI handles well. ## 6. Phone-Based AI Companionship for Your Parent This one is for your parent — but it addresses a caregiver concern: loneliness and isolation between visits. CareYaya's QuikTok is a phone-based AI companionship service designed for older adults who don't use smartphones. Your parent calls a regular phone number and has a conversation with an AI companion. No app to download, no screen to navigate, no account to set up. For parents who are more tech-comfortable, apps like Claude on a phone or tablet can hold a conversation, read things aloud, or help recall details. But the key advantage of phone-based services is that they require nothing on the parent's end — which matters when your parent lives alone and you can't be there to help troubleshoot. This is different from the companion apps covered in [AI Companions for Seniors](/guides/ai-companions-seniors), which tend to be app-based and more feature-rich. Phone-based options lower the barrier to entry considerably. ## What to Watch Out For **The privacy rule.** Free general-purpose AI tools — ChatGPT, Gemini, Claude — are not HIPAA-compliant. Don't paste full medical records, insurance ID numbers, Social Security numbers, or your parent's full name and date of birth into them. You don't need to. Describe the situation in general terms ("my 78-year-old father, recovering from hip surgery") and the AI can still help you draft, organize, and summarize effectively. **AI makes things up.** AI tools sometimes sound confident while being wrong — especially about specific programs, local services, phone numbers, and current eligibility rules. Always verify any specific claim, provider name, or benefit amount with an official source before acting on it. **AI is not a doctor, nurse, or social worker.** For clinical decisions — adjusting medications, deciding whether a symptom needs emergency attention, choosing a care facility — there is no substitute for a qualified professional. AI is a drafting and organizing assistant, not a care advisor. **Caregiver burnout is real.** AI can reduce time spent on paperwork. It can't reduce the emotional weight of caring for a parent. If you're overwhelmed, the Family Caregiver Alliance (caregiver.org) offers real human support and resources. ## What to Try Next If your parent's doctor visits feel rushed or confusing, [How to Use AI to Prepare for a Doctor's Visit](/guides/ai-doctor-visit-prep) covers exactly how to use AI before and after appointments to make the most of limited time with a clinician. And if prescriptions have become hard to track or understand, [AI Tools for Understanding Prescriptions](/guides/ai-prescription-explainer) walks through how to ask an AI to explain what a medication does and what side effects to watch for. --- # AI Tools for People with Low Vision: Free Apps That Help Every Day Source: https://aiforregulars.com/guides/ai-tools-low-vision Smartphones have quietly become one of the most powerful assistive technology devices ever made — and the AI features added in recent years have made them significantly more useful for people with low or no vision. The best part: most of these tools are free and run on a phone you may already own. This guide covers the most practical options, how to get started with each one, and which tasks each does best. ## At a glance | App / Feature | Platform | Cost | Best for | |---|---|---|---| | Seeing AI | iOS + Android | Free | Reading text, identifying products, describing scenes | | Be My Eyes | iOS + Android | Free | Real human or AI help for any visual task | | Google Lookout | Android | Free | Fast text reading, scene description, explore mode | | iPhone VoiceOver + Magnifier | iOS | Built in | Screen reading, magnification, live camera text | | Android TalkBack + Magnification | Android | Built in | Screen reading, magnification | | ChatGPT photo feature | iOS + Android | Free (limited) | Describing photos, reading documents | ## Seeing AI (iOS + Android) Seeing AI is a free app from Microsoft designed specifically for blind and low vision users. Point your phone's camera at something and the app describes it out loud. **What it does well:** - Reads printed text aloud — letters, menus, receipts, signs - Identifies people and describes their apparent expression and position - Scans barcodes and tells you what product it is - Describes scenes ("a kitchen with a table, two chairs, and a window") - Reads handwritten notes (with reasonable accuracy) - Recognizes currency **Setup steps:** 1. Open the App Store (iPhone) or Google Play (Android) and search "Seeing AI" 2. Download and open the app — no account needed 3. Point your camera at text and the app begins reading automatically 4. Use the Read, Describe, and More tabs to switch between feature groups — Read for text and documents, Describe for scenes and photos, More for barcodes, currency, and colour detection **Honest limitation:** The app's layout was updated in 2025 — features are now organised into Read, Describe, and More tabs rather than the older swipeable channel list. ## Be My Eyes Be My Eyes connects people who need visual help with either a human volunteer or an AI assistant. The human volunteer option connects you with a sighted person over video who can help with anything visual. The AI option (powered by a large language model) handles many tasks instantly, any time of day. **What it does well:** - Identifying what's in the fridge and suggesting what to cook - Reading medicine labels and instructions - Helping with tasks that are hard to describe ("I'm trying to set up my new router — can you help me read what's on the back?") - Any complex visual task where context matters **Setup steps:** 1. Download Be My Eyes from the App Store or Google Play 2. Create a free account (email or phone number) 3. Open the app and tap "Get Help" — choose AI assistant or a human volunteer 4. Hold up your phone camera to whatever you need help with and ask your question out loud I had a family member test Be My Eyes to read a prescription label that had very small print. The AI assistant read every line accurately and then explained what the abbreviations meant without being asked. That combination of reading and explaining is genuinely useful — and it took about 30 seconds. **Honest limitation:** The human volunteer service is impressive but wait times can vary. The AI assistant is instant but handles complex visual tasks less reliably than a human. ## Google Lookout (Android) Google Lookout is Android's equivalent to Seeing AI. It uses the phone's camera and AI to describe the world in real time via audio. **What it does well:** - Explore mode: describes your surroundings as you move - Documents: reads printed text on letters, forms, books - Images: describes a photo you take - Food labels: reads nutrition facts and ingredients **Setup steps:** 1. Open Google Play and search "Lookout" 2. Install the app (it's free — tap Install on the Google Play listing) 3. Open it and grant camera and microphone permissions 4. Select a mode from the main screen and point the camera ## Built-in phone features (often overlooked) Before downloading any app, it's worth knowing what your phone already does. Both iPhone and Android have substantial built-in accessibility tools that many people don't know about. **On iPhone:** - **Magnifier:** A built-in app that turns your camera into a powerful magnifying glass with adjustable zoom and lighting. Add it to your Control Center for quick access. - **VoiceOver:** A full screen reader that describes everything on screen and allows navigation by touch and gestures. - **Live Text:** Point your camera at any text (menus, signs, handwriting) and the phone reads it. Available directly in the Camera app. **On Android:** - **Magnification:** Triple-tap the screen to zoom in anywhere on the phone, or use a button shortcut. - **TalkBack:** A screen reader that works similarly to VoiceOver. - **Lookout:** Described above — available as a free download. To find these features on iPhone, go to Settings > Accessibility. On Android, go to Settings > Accessibility. ## ChatGPT's photo feature The free version of ChatGPT on mobile lets you take a photo and ask a question about it. This is useful for tasks where you need an explanation, not just a description: ``` [Photo of a document] What does this letter say, and is there anything I need to respond to? ``` ``` [Photo of a pill bottle] Read the label and explain the dosage instructions in simple terms. ``` The limitation is that it's not real-time — you take a photo, send it, and wait for a response. For quick tasks, Seeing AI or Lookout are faster. For tasks requiring understanding and explanation, ChatGPT is often better. ## What to try next If you're interested in other AI tools useful for daily life as an older adult, see [AI Health Gadgets for Seniors](/guides/ai-health-gadgets-seniors) for a roundup of monitoring and reminder tools. To explore using AI with your voice instead of touch, [ChatGPT Voice Mode](/guides/chatgpt-voice-mode) explains how to set it up on your phone. --- # AI Toys for Kids: What Every Parent Should Know Before Buying Source: https://aiforregulars.com/guides/ai-toys-kids-privacy AI-connected toys are exciting. They talk back, answer questions, tell stories, and seem to genuinely understand what your child wants. But unlike an app on a phone, they run in a bedroom or living room with a microphone active — often without a screen to show you what is happening. That makes oversight harder than most parents expect, and it is why a few extra questions before you buy are worth the effort. ## What AI Toys Actually Collect The most significant category is voice recordings. When a toy listens for a trigger word or waits for a spoken command, it captures audio. In many cases, those recordings are sent to and stored on company servers — not just the toy itself. That means a data breach at the company could expose your child's voice recordings. Beyond audio, connected toys often collect: - Usage patterns: how often the toy is used and what kinds of questions get asked - The child's name, age, or grade level if entered during setup - Location information tied to your home WiFi network - In some cases, the content of full conversations, not just commands The detail that matters most is where this data lives and for how long. Some companies store recordings indefinitely. Others delete them after a fixed period. The privacy policy will say — if you can find it and read it clearly. ## Why This Matters More Than Most Parents Realize Toy companies have experienced data breaches, and children's data is sensitive in ways that go beyond what most people expect. Voice recordings capture a child's voice, the way they speak, the topics they ask about, and sometimes things said in the background by other family members. In the United States, COPPA — the Children's Online Privacy Protection Act — requires companies to get verifiable parental consent before collecting personal information from children under 13. Similar laws exist elsewhere. But laws require enforcement, and not every product complies fully. In recent years, the FTC has brought enforcement actions against connected toy makers for collecting children's data — including location information — through third-party software built into companion apps, without notifying parents or obtaining their consent. The practical point is this: a toy that seems harmless at the store may operate quite differently once it is connected to your home network and running in your child's room every day. ## Questions to Ask Before You Buy These questions are worth answering before a toy comes home — not after. **Where are voice recordings stored, and for how long?** "On our secure servers" with no mention of a retention period or deletion process is not a complete answer. **Can I delete them?** There should be a clear way to do this — through a parent app, a web dashboard, or a physical control on the toy itself. **Is there a physical mute button or microphone disconnect?** A physical switch that cuts the microphone is more trustworthy than a software setting. Software can have bugs; a physical switch cannot be overridden remotely. **What is the company's history with children's privacy?** A quick web search for "[company name] privacy" or "[toy name] data" will often surface past issues or regulatory attention. **Does it require a parent account with verified consent?** This is a signal that the company takes its COPPA obligations seriously and has built a way for you to manage and delete data. Before I buy any AI toy as a gift, I look for two things: a physical off switch on the microphone and a privacy policy I can read in under five minutes. If I cannot find either, I move on. The toy does not have to be perfect — but I want to know it is honest about what it does with the recordings my child makes in their own bedroom. ## Red Flags in Product Listings A few things in a product listing or packaging should prompt a harder look before you buy. **"Always listening" without a clear mute option.** If the microphone is active all the time and there is no easy way to cut it completely, that is worth weighing carefully before the toy enters your home. **A privacy policy full of vague language.** Phrases like "we may share data with trusted partners" without naming who those partners are, or "data is retained as long as necessary" without a specific timeframe, suggest the company has not committed to clear protections. **No mention of COPPA or an equivalent law.** Products designed for children in the US should acknowledge their COPPA obligations. Its absence does not guarantee non-compliance, but it is worth investigating before you buy. **No parent app or oversight tools.** If there is no way for a parent to view, manage, or delete what the toy collects, that gap is meaningful — and it is likely to stay that way after purchase. **No English-language privacy contact.** If you cannot find a way to contact the company to ask questions or request data deletion, that is a practical problem regardless of where the company is based. Assess the actual policy and contact options, not just the country of origin. ## A Pre-Purchase Checklist Before any AI-connected toy comes home, go through these five steps: 1. Read the privacy policy — who receives the recordings, and is there a specific retention and deletion period? 2. Look for a physical mute or off switch, not just a software setting. 3. Confirm there is a parent account that allows you to view and delete stored data. 4. Search "[toy name] privacy" online to check for past complaints or regulatory attention. 5. After purchase, set a calendar reminder to delete stored recordings periodically — do not assume it happens automatically. The checklist takes about fifteen minutes. That is a small investment for a device that will sit in your child's room for years, listening. **What to try next:** If you want to extend safety controls beyond connected toys to the AI apps your child uses on a screen, [Parental Controls for AI Apps: ChatGPT, Gemini and Copilot in One Guide](/guides/ai-parental-controls-all-apps) walks through each platform side by side. And for a five-minute family safety project, [Safe Words for Families](/guides/family-safe-words) explains how to set up a code word that protects against voice scams — a natural next step once you start thinking seriously about what microphones can capture. --- # Turn Your Family Photos into a Slideshow Video with AI (Free) Source: https://aiforregulars.com/guides/ai-video-from-photos Old photos sitting in your phone's camera roll deserve better than just scrolling past them. With AI tools, you can turn a pile of family photos into a real video — with music, smooth transitions, and a professional look — in about five minutes, entirely on your phone. No editing skills. No computer. No cost. Start by collecting the photos you want in the video. You don't need to move them anywhere yet — just scroll through your phone's camera roll and note which ones you want to use. A few tips for choosing photos: - Aim for 10–30 photos for a 1–3 minute video. Fewer than 8 feels too short; more than 40 can feel long. - Mix close-ups and wider shots — it makes the video more dynamic. - Don't worry about perfect quality. AI tools handle slightly blurry or older photos well. If you want to restore faded or old photos first, that's worth doing — there's a guide to [restoring old photos with AI](/guides/restore-old-photos-ai) that takes about 10 minutes. Here are three free options, all phone-friendly: **Google Photos** (Android and iPhone) The simplest option. If you already use Google Photos to back up your camera roll, it's already there. It can auto-generate a video, or you can pick specific photos and create one yourself. **CapCut** (Android and iPhone) Free app with more creative control — better music selection, text overlays, and transition styles. The AI will auto-arrange photos and suggest a theme. **Canva** (Android and iPhone, or browser) If you already use Canva, it has a video creation feature that works well for slideshows. More design options than Google Photos, but a little more to navigate. For a first-timer, start with **Google Photos**. It requires the least work. **In Google Photos:** 1. Open the app and tap **Library** at the bottom. 2. Tap **Utilities**, then **Movie**. 3. Select the photos you want (tap each one to add it). 4. Tap **Create** and Google Photos will arrange them automatically with music and transitions. **In CapCut:** 1. Open the app and tap the **+** button. 2. Select your photos from your camera roll. 3. Tap **Auto** or choose a video template. The AI picks music and transitions to match your photos. **In Canva:** 1. Open Canva and tap **+** to create a new project. 2. Choose **Video** as the format. 3. Browse templates labeled "Slideshow" and tap one you like. 4. Replace the placeholder images with your own photos. I made a video from 22 vacation photos using Google Photos on an older Android phone — it took less than three minutes from start to finish. The AI picked upbeat background music and crossfade transitions automatically. I didn't change a thing and it looked genuinely good. If you want more control over the style, CapCut is worth the extra few steps. All three tools include background music automatically. If you don't love the AI's choice: - In **Google Photos**, tap the music note icon to browse other tracks - In **CapCut**, tap the music bar at the bottom of the editor and pick from their free library, or upload your own audio file - In **Canva**, tap the audio track in the timeline and replace it If you're sharing the video privately with family, using any music is fine. If you plan to post it publicly on YouTube or social media, stick to tracks labeled **royalty-free** to avoid copyright issues. Before you download, press **Play** and watch the whole thing through. Look for: - Photos that appear in a weird order (drag them to reorder) - Any photo that looks out of place or should be removed - Whether the video feels too long or too short Most apps let you drag photos in the timeline to rearrange them. In CapCut you can also adjust how long each photo stays on screen. When you're happy, tap **Download** or **Export**. The video saves to your camera roll as an .mp4 file. From there you can: - Share it directly to WhatsApp, iMessage, or Facebook - AirDrop it to family members - Save it as a memory for a birthday, anniversary, or holiday - Upload it to a private YouTube or Google Drive link to share with relatives The video exports in HD quality (up to 1080p), which looks sharp on a phone screen or TV. If you share directly from Google Photos via a link, note that there is no password option — anyone who has the link can view it. ## What to try next Now that you've got a photo slideshow, try [making a fully AI-generated video from a text prompt](/guides/make-ai-video-free) — no photos needed, just a description of a scene. Or if you want to design a shareable card or invitation around one of your favorite photos, [Canva's AI tools](/guides/canva-ai-beginners) make that easy too. --- # AI Voice Cloning Scams: How They Work and How to Protect Your Family Source: https://aiforregulars.com/guides/ai-voice-cloning-scams Voice scams are not new, but AI has made them far more convincing than anything that came before. A caller who sounds exactly like your son, your granddaughter, or your spouse can now be created in minutes using a clip pulled from a public video. Knowing how these calls are built — and having a simple plan already in place — makes a real difference. ## How the Call Unfolds The call usually starts with an emergency. The voice on the other end sounds panicked: a car accident, an arrest, a sudden medical situation. The goal is to keep you emotional and moving fast. Panic short-circuits our usual caution. A second caller often joins in — posing as a lawyer, police officer, or doctor. They explain why money needs to move right now and tell you not to call anyone else first. The urgency is the point. Scammers know that a calm person who simply hangs up and calls their child directly will not fall for the trick. The payment method is almost always untraceable: gift cards, wire transfers, or cryptocurrency. Real emergencies do not work this way. ## Why a Few Seconds of Audio Is Enough AI voice tools can learn the distinctive characteristics of a voice — its pitch, rhythm, accent, and small personal quirks — from a surprisingly short sample. Security researchers at McAfee found that as little as three seconds of audio was enough to produce a voice match convincing to casual listeners; longer samples only make the clone more accurate. Videos posted to social media, TikTok or Instagram clips, YouTube appearances, podcast recordings, or even voicemail greetings are all potential sources. The copy is not perfect, but it does not have to be. Phone calls already compress audio and add background noise. A voice that is close enough, combined with a stressful story and a short call, is often convincing in the moment. This is why telling yourself "I would recognize my own daughter's voice" is not a reliable defense. The cloned voice is not trying to fool an expert in a quiet room — it just needs to fool a worried parent on an imperfect phone connection for ninety seconds. ## Red Flags to Watch For During the Call Even when the voice sounds right, other parts of the call often signal something is wrong. Watch for these: - **The caller pushes you to act immediately.** Real emergencies allow for a moment to pause and verify. Scammers do not. - **You are told not to call anyone else.** No real lawyer, police officer, or hospital staff will ever say this. - **The solution requires gift cards, wire transfers, or cryptocurrency.** Bail bonds, hospital bills, and legal fees are not paid this way. - **The caller avoids specific personal details.** A real family member will know your shared history, the name of a mutual friend, or a family detail a stranger would not. - **The call quality seems oddly off.** A slightly robotic tone, an unnatural pause, or missing background noise can be a clue with cheaper cloning tools — but this signal is fading fast. On today's best tools, listening alone often can't tell a clone from the real thing, so don't treat a "normal-sounding" call as proof. ## The Five-Minute Family Plan The most effective protection is a shared family safe word — a short, unusual phrase that only your family knows. If someone calls claiming to be a family member in trouble, you ask for the safe word. A scammer will not know it. **Choose an odd, memorable phrase.** Pick something you would never use in everyday conversation. Avoid pet names or anything that might come up naturally. Something like "purple hammock" or "Grandma's recipe box" works well. The more specific and unusual, the better. **Share it privately.** Tell every family member who might receive or make an emergency call. Do this in person or in a direct private conversation — not in a group chat where screenshots can easily be taken or shared by accident. **Agree on one rule.** Anyone who calls claiming to be a family member in an emergency must say the safe word when asked, no exceptions. Agree in advance that even if the caller says they cannot say the word right now, you still hang up and call back directly. A real emergency will survive a two-minute delay. **Practice it once.** Make one brief call where someone asks for the word and someone says it correctly. This makes it feel natural and confirms that everyone has it right. It takes about thirty seconds. When I set up a safe word with my own family, my mother-in-law thought it was a bit silly at first. Then she got a scam call — not even a voice clone, just a regular impersonation — and she knew exactly what to do. She asked for the word, heard silence, and hung up. The whole setup had taken about four minutes. ## What to Do When a Call Comes In Even with a safe word in place, here is the right sequence of steps for any suspicious emergency call: 1. Stay calm. Urgency is a tool the caller is using against you. 2. Ask for the safe word. If they cannot provide it, hang up. 3. Call your family member back on a number you already have saved — never use any number the caller provides. 4. Do not send money, gift cards, or cryptocurrency based on a single unexpected call. Verify first, always. 5. Report the call to the FTC at reportfraud.ftc.gov if you believe it was a scam. The most important thing you can do is give yourself time. Scammers rely on panic. A pause of even two minutes to call back on a known number is usually enough to break the illusion. **What to try next:** Get the full step-by-step setup in [Safe Words for Families](/guides/family-safe-words), and learn how the same AI technology is used to create fake videos in [How to Spot a Deepfake Video](/guides/how-to-spot-deepfake-video). ## Sources - [Scammers use AI to enhance their family emergency schemes](https://consumer.ftc.gov/consumer-alerts/2023/03/scammers-use-ai-enhance-their-family-emergency-schemes) — Federal Trade Commission - [Family Emergency Scams](https://consumer.ftc.gov/all-scams/family-emergency-scams) — Federal Trade Commission - [Artificial Imposters: Cybercriminals Turn to AI Voice Cloning for a New Breed of Scam](https://www.mcafee.com/blogs/privacy-identity-protection/artificial-imposters-cybercriminals-turn-to-ai-voice-cloning-for-a-new-breed-of-scam/) — McAfee - [Scammers Target Older Adult Victims](https://www.fbi.gov/news/stories/scammers-target-older-adult-victims) — Federal Bureau of Investigation --- # Using AI to Plan a Wedding: Where It Actually Helps (and Where to Stop) Source: https://aiforregulars.com/guides/ai-wedding-planning Weddings generate an enormous amount of paperwork-adjacent work — a budget to track, a timeline to build, a guest list to manage, vendors to email, and a hundred small decisions in between. None of that is the romantic part of the day, and increasingly, couples are handing it to ChatGPT. If you've wondered whether that's a reasonable thing to do, or where it stops being a good idea, here's a practical walkthrough. ## How to Use AI for Wedding Planning Tell ChatGPT your total budget, your guest count, and roughly where you are (city or region matters for pricing), then ask it to break the budget into categories — venue, catering, attire, photography, flowers, and so on — with a starting percentage for each. Ask it to lay this out as a table you can keep updating as real quotes come in. The value isn't the first answer; it's having a structure you return to every time a vendor quote changes something else. Give it your wedding date and today's date, and ask for a month-by-month checklist working backward — when to book a venue, when invitations should go out, when final headcounts are due. Ask it to flag anything time-sensitive, like dress alterations or marriage license paperwork, that's easy to leave too late. You'll still need to check locally specific deadlines yourself, but this turns a vague sense of "there's a lot to do" into an actual list with dates on it. Instead of staring at a blank email to a caterer or photographer, describe the situation — your date, venue, guest count, what you're asking about — and ask ChatGPT to draft a clear, polite email with the right questions included. This is where it saves the most real time, because it's genuinely good at professional-sounding drafts. Just don't let a draft assert a price, date, or availability as fact; those come from the vendor, not the chatbot. How to word a no-kids request, what to do about a plus-one you're unsure about, who traditionally pays for what — these are exactly the kind of questions people feel awkward asking a person but will happily type into a chatbot. Treat the answer as a reasonable starting point rather than a rule, especially where family, religious, or cultural traditions are involved, since those vary a lot and ChatGPT can't know your specific family's expectations. If you've been collecting ideas — a color palette from Pinterest, a playlist of songs, a screenshot of a centerpiece you liked — ask ChatGPT to help you turn that pile into an organized shortlist or a shopping list you can actually act on. It's much better at organizing scattered inspiration into something usable than at generating original ideas from nothing. This is the line most couples draw for themselves, and it's a reasonable one: use AI to organize your thoughts, fix the pacing of a speech, or suggest a structure if you're stuck — but write the actual sentences yourself, the same way you would for a eulogy. The parts of the day meant to sound like you are exactly the parts where a chatbot's fluency works against you. The single best prompt I've seen work for the budget step wasn't "make me a wedding budget" — it was giving ChatGPT the actual total and asking it to show the breakdown as a running spreadsheet-style table with a column for "estimated" and a column for "actual," so every real quote could update one row without losing track of the rest. Couples who used it that way kept coming back to update it for months; couples who asked for a one-time answer got a nice list they never looked at again. ## What to Watch Out For **Treating a draft price or date as a fact.** ChatGPT doesn't know your venue's actual availability, your caterer's current pricing, or whether a vendor even exists anymore. It's a drafting tool for your questions and your paperwork, not a source of current, local information — confirm anything that matters with the vendor directly. **Pasting in contracts or full guest details.** Don't paste a signed vendor contract, payment information, or your full guest list with addresses into a chatbot. If you want help understanding a confusing clause in a contract, summarize the specific part you're unsure about instead — and for anything with real legal weight, a second read from a person you trust (or a lawyer, for a large vendor contract) matters more than an AI's read of it. **Letting it write the emotional parts.** A budget spreadsheet doesn't need to sound like you. Your vows do. The more personal a piece of writing is meant to be, the less useful it is to hand the actual wording to AI — use it to organize, not to speak for you. **Assuming its etiquette answer is universal.** Wedding etiquette differs a lot by culture, religion, and family, and a chatbot's answer reflects whatever was most common in its training data — often a fairly Western, secular default. Cross-check anything tied to a specific tradition with someone who actually knows it. ## What to Try Next If the budget is the part keeping you up at night, [Managing Your Monthly Budget with ChatGPT](/guides/chatgpt-monthly-budget) walks through the same spreadsheet-style approach for ongoing money, not just the wedding. And if a vendor contract has language you don't fully understand, [How AI Can Help You Understand a Contract](/guides/ai-explain-contracts) covers how far you can trust AI with that, and where you shouldn't. ## Sources - [The AI Wedding Planning Guide: Best Tools, Budget Prompts & More](https://www.zola.com/expert-advice/ai-wedding-planning-guide) — Zola - [30+ AI Wedding Statistics (2026)](https://weddingtoasts.co.uk/blog/ai-wedding-statistics/) — Wedding Toasts UK --- # AI Weight Loss Apps Explained: Do They Actually Work? Source: https://aiforregulars.com/guides/ai-weight-loss-coaching Weight loss apps have been promising results for years, but the latest generation adds an AI coach that adapts to you, learns your patterns, and sends encouragement between meals. Three apps — Noom, Simple, and Healthify — lead the current crop, each claiming to use AI in meaningfully different ways. Before you sign up for a free trial that silently converts to an annual charge, it's worth understanding what the AI in each actually does, what real evidence exists, and whether a free chatbot might do the job just as well. ## What "AI Coach" Actually Means Here The phrase "AI coach" covers genuinely different things across these apps. In some cases it means a chatbot that responds to your questions with advice pulled from a nutrition database. In others it means a personalization engine that looks at your food logs, sleep data, and activity patterns and adjusts your recommendations over time. The difference matters, because one is far more sophisticated than the other — and the pricing doesn't tell you which you're getting. All three apps in this guide use AI in distinct ways, which is why they suit different kinds of people. ## Noom: Behavioral Coaching, Daily Lessons, and the Welli AI Coach Noom's central claim is that weight loss is mostly a psychology problem. Instead of building around a calorie budget, it pairs meal tracking with short daily lessons on habits, emotional eating triggers, and stress. The app's AI coach, called Welli, is there to answer questions and help you work through what's going wrong — whether that's stress eating on weekday evenings or a weekend pattern that keeps undoing the week's progress. **What the company says:** Noom reports that 64% of users lose 5% or more of their body weight, and 60% maintain that loss for more than a year. **What it costs:** Around $70 per month, though Noom regularly offers promotional pricing and shorter trials. **Who it's best for:** People who have tried calorie tracking before and found it didn't stick long-term. If you know roughly what healthy eating looks like but keep sliding back into old patterns, the behavioral coaching layer is what makes Noom different from a basic tracker. **What it's not:** Noom is not a medical program. It doesn't know what medications you take, doesn't coordinate with your doctor, and isn't equipped for complex health situations. It is also one of the more expensive options in this category. ## Simple: Habit Tracking and Photo Food Logging Simple takes a lighter approach. Instead of daily psychology lessons, it focuses on a small set of habits — intermittent fasting windows, hydration goals, and meal logging — and makes logging easier by letting you photograph your food rather than searching a database and manually entering every item. In June 2025, the TODAY show featured a 43-year-old man who lost 100 pounds over a year using the Simple app. His account pointed to the accountability factor — opening the app daily, seeing a streak, logging consistently — rather than any specific dietary formula the app prescribed. Individual stories can be motivating, but they're not the same as controlled research. **What it costs:** Simple uses a freemium model: a permanent free tier with limited features and a Premium plan at around $59.99 per year. **Who it's best for:** People who find detailed meal logging too tedious to sustain. Photo-based logging removes most of the friction, and the habit approach is less demanding than working through daily lessons. If you try photo food logging in Simple or Healthify, take the photo before you eat, not after. It sounds minor, but logging a half-eaten plate takes longer to identify and is one of the main reasons people stop logging consistently after a few weeks. One second before you pick up the fork makes the habit much easier to keep. ## Healthify: AI Coach Ria, with a Free Tier Healthify (also known as HealthifyMe) is popular across South Asia and increasingly available in other markets. Its AI coach is called Ria. The clearest practical advantage Healthify has over Noom and Simple is price: there's a free tier for basic calorie tracking, but Ria AI coaching requires a paid subscription. Premium tiers add features including in-depth calorie analysis and access to human coaches. **What it costs:** Free tier available for basic calorie tracking. Ria AI coaching requires a paid plan; prices vary by region. **Who it's best for:** People who want to try AI coaching before committing money, or users in regions where Noom and Simple have less localized content or support. ## The Free Option: ChatGPT for Meal Ideas and Habit Troubleshooting Before spending money on any subscription, spend ten minutes with ChatGPT and see what it gives you. It won't track your meals automatically, remind you to drink water, or protect a streak. What it will do — for free — is give you a thoughtful response to almost any nutrition question, help you plan a week of meals around what you already have at home, and explain in plain English why a habit you're trying to build keeps not sticking. > "I'm trying to eat more protein without adding much more meat. Can you suggest five dinners I could make this week with ingredients from a regular grocery store? I'm cooking for one." That kind of back-and-forth is genuinely useful and costs nothing. If you find yourself wanting more structure — tracked data over time, daily nudges, a streak to maintain — then a paid app starts to make sense. But the free option is worth trying first, especially if you're not yet sure what's blocking you. GoodRx and Fortune both published roundups of AI diet apps in 2026 noting that for people who primarily want meal inspiration and quick nutrition answers, a free chatbot often does the job. The apps earn their subscription cost for users who want ongoing tracking and behavioral accountability. ## What the Evidence Actually Says Independent research on food-tracking apps as a category is moderately encouraging: people who log their meals and receive feedback tend to lose more weight than those using no tools. The harder question — whether AI coaching specifically outperforms a plain food diary — has fewer independent studies behind it. Noom has the most published data, including a large-scale 2016 study and subsequent company-reported figures that suggest real outcomes. Critics point out that most of the supporting research is funded or co-authored by Noom itself. Simple's independent research base is thin. Healthify has published internal data, but peer-reviewed third-party studies are harder to find. **Realistic expectations:** Most nutrition experts say 1–2 pounds of weight loss per week is the sustainable rate for most people. Any app promising faster results deserves skepticism about what it's measuring and how. ## The Upsell Trap: Free Trials That Turn Into Annual Charges Free trials in this category are set up to convert, and they're designed to make cancellation easy to forget: **Trials are usually short.** Seven days is common. If you sign up on a Tuesday and forget, you can be charged by the following Wednesday. **Annual billing means a large upfront charge.** Plans advertised at a low monthly price often require a full year paid at once — meaning you're actually paying $60–$120 at sign-up, and refunds after day one of the paid period aren't guaranteed by the app. **You often need to cancel in the app store, not the app.** Many subscriptions are managed through the Apple App Store or Google Play, and cancelling inside the app itself does nothing. Always check your phone's subscription settings directly. Set a calendar reminder for 48 hours before any trial ends. That gives you time to decide, and enough time to cancel with a day to spare. ## When to Talk to a Doctor Instead AI weight loss apps are designed for generally healthy adults who want to lose a moderate amount of weight. They're not appropriate for everyone: **If you take medication**, diet can interact with what you're taking — diuretics, diabetes drugs, thyroid medications, and blood pressure medications all have dietary considerations an app won't know about. **If you want to lose more than 10–15% of your body weight**, a doctor or registered dietitian should be involved. Rapid or large-scale weight loss carries real health risks that an app cannot monitor. **If you have a history of disordered eating** — anorexia, bulimia, binge-eating disorder, or orthorexia — apps centered on food logging, calorie targets, and daily weight tracking can reinforce harmful patterns even in people who feel recovered. These apps are not eating-disorder-aware tools. **If you have diabetes, heart disease, or a thyroid condition**, weight management involves medical considerations that need professional oversight. If any of these apply to you, the right first step is your GP or a registered dietitian, not an app store. ## What to Try Next The workout side of AI health apps is a different world from nutrition coaching — [AI Fitness Apps for Beginners](/guides/ai-fitness-apps) covers what's actually worth trying for exercise. If you want to put AI to work on meal ideas without paying anything, [How to Use AI for Meal Planning](/guides/ai-meal-planning) walks you through exactly how to do it with a free chatbot. --- # Writing a Eulogy with ChatGPT: How to Get Help Without Losing Your Own Voice Source: https://aiforregulars.com/guides/ai-write-a-eulogy Being asked to speak at a funeral is an honor, and it usually arrives at the worst possible time to write anything — a few days' notice, a head full of grief, and a blank page. If you've opened ChatGPT wondering whether it's okay to ask for help with something this personal, you're not alone, and it's a reasonable thing to use it for. It won't know your loved one. But it can help you turn scattered memories into something you can actually stand up and read. ## How to Write a Eulogy with ChatGPT's Help Spend ten minutes with a pen and paper first, not a chatbot. Write down two or three specific memories — a moment, not a summary. A phrase they always said. What they were like on an ordinary day, not just at their best. What you think people got wrong about them, or never got to see. This raw material matters more than anything ChatGPT will suggest — it's the difference between a eulogy that sounds like a Wikipedia entry and one that sounds like the person actually lived. Skip "write a eulogy for my mother." Instead, paste in the memories and details you gathered, and ask ChatGPT to help you shape them into a eulogy — something like: "Here are some memories of my mother. Help me organize these into a eulogy of about 500 words, in a warm, plain-spoken tone. Don't add any details I haven't given you." That last line matters: it's the instruction that keeps ChatGPT from quietly inventing things that sound true but aren't. A eulogy that people connect with usually moves through a few clear beats: a short opening (who you are and your relationship), the heart of it — specific stories, told one at a time rather than as a list of traits, what they valued and how you saw it show up in daily life, and a closing that says what you'll miss or carry forward. If you're not sure how to order your memories, ask ChatGPT to suggest an order using this shape — but keep the stories themselves as yours. Read the draft out loud, sentence by sentence, and cross out anything you wouldn't actually say — a phrase that's too formal, a metaphor that isn't you, an ending that sounds like a greeting card. This step is the one that matters most. A eulogy that's technically well-organized but doesn't sound like the person delivering it will feel off to everyone listening, even if they can't say why. Read it aloud at least twice, at the pace you'd actually speak — slower than reading silently, with room for pauses. Time it. Most people speak more slowly than usual when they're managing emotion, so add a minute or so to whatever your read-through comes out to. If it runs long, this is the point to cut a story rather than trim sentences everywhere — a eulogy is stronger with two full stories than five rushed ones. Print the final version in large, clear type — reading a phone screen through tears in a dim room is harder than it sounds. If you think there's a real chance you won't be able to finish, ask someone close to you beforehand if they'd step in and read the rest, and let them know where a copy is. Having a plan for this isn't pessimistic; it takes one thing off your mind on a day that already has enough on it. The instruction that made the biggest difference when a friend of mine wrote her father's eulogy this way wasn't about tone or structure — it was telling ChatGPT explicitly not to add anything she hadn't said. Without it, the first draft added a warm-sounding line about him "always making time for family," which wasn't really true and would have bothered her brothers. With it, every draft stayed built only from what she'd actually told it. ## How Long Should It Be There's no single right length, but most funeral guides converge on a similar range: somewhere around **3 to 7 minutes spoken aloud**, which usually works out to **roughly 400 to 900 words**, depending on how quickly you naturally speak. If you're not a confident public speaker, or you suspect you'll be emotional, aim for the shorter end — a short eulogy delivered steadily leaves a stronger impression than a long one that trails off. When you ask ChatGPT to draft or shape your eulogy, tell it the target length up front ("about 500 words" or "around 4 minutes read aloud") so you're not cutting a finished draft down afterward. ## What to Watch Out For **Letting AI fill in gaps with things that sound true.** If you give ChatGPT sparse information, it may complete the picture with a plausible-sounding detail you never actually said — a generic "she touched everyone she met" where you gave it nothing specific about how. Read every line back and ask whether you know it to be true, not just whether it sounds right. Telling it explicitly not to invent details, as in Step 2, helps prevent this from the start. **Sensitive family details.** If a family member asked you to leave something out of the public story — an illness, an estrangement, the circumstances of the death — don't include it in what you tell ChatGPT at all. It's much easier to keep something out of a draft entirely than to catch and remove it later. **A eulogy that sounds like everyone else's.** AI tends to default to familiar phrases — "a light in every room," "words cannot capture" — because they show up often in its training data. If a line could describe almost anyone, cut it and replace it with something only true of the specific person you're talking about. **It's still supposed to be delivered by you, in your voice.** The goal isn't a polished piece of writing — it's something you can stand up and say out loud, in a room full of people who knew this person too, and have it sound like you meant every word. ## What to Try Next If you're also responsible for the obituary, [AI Wrote the Obituary — Is That Okay?](/guides/ai-obituary-condolence) covers the same balance of using AI as a starting point without losing the person in the process. And if writing the eulogy has left you wanting to preserve more of your loved one's story — not just for the funeral, but for the family afterward — [Let AI Help You Write a Life Story or Family Memoir](/guides/ai-write-life-story-memoir) walks through that longer project. ## Sources - [How to Write a Meaningful Eulogy (With a Simple Outline + Examples)](https://funeral.com/blogs/the-journal/how-to-write-a-meaningful-eulogy-with-a-simple-outline-examples) — Funeral.com - [Writing Funeral Eulogies That Will Be Remembered: Tips and Advice](https://www.funeralwise.com/celebration-of-life/eulogy/) — Funeralwise - [How to Use ChatGPT to Write a Eulogy](https://www.griproom.com/fun/how-to-use-chatgpt-to-write-a-eulogy-mastering-the-art-of-compassionate-ai-writing) — GripRoom --- # How to Write Your Life Story with AI: A Guide for Memoir Beginners Source: https://aiforregulars.com/guides/ai-write-life-story-memoir Every family has stories that exist only in one person's memory. The grandmother who came to this country with nothing. The father who never talked about the war. The years when everything was hard and then somehow wasn't. Those stories disappear when the person who lived them is gone. AI doesn't replace the work of remembering — but it does make the work of writing much easier. You don't need writing experience or a plan. You just need to start talking. The most common mistake new memoir writers make is trying to outline the whole story before writing a word. This leads to paralysis. AI gives you a better starting point: one memory, told in your own words, as rough as you like. Open ChatGPT or Claude and try this prompt: ``` I want to write my life story as a memoir. I'm going to start by telling you one memory, and I'd like your help expanding it into a readable scene. Here's the memory: [Write 3–5 sentences about something you remember clearly — a place, a moment, a conversation. Don't worry about grammar or order. Just describe what you remember.] Please ask me questions to help me add more detail: what it looked, sounded, and felt like; who was there; why this moment stayed with me. ``` Let the AI interview you. Answer each question in plain language. After a few exchanges, ask it to write the scene as a memoir paragraph using your answers. Read it, correct anything wrong, and save what works. Don't try to write your life in order. Memory doesn't work that way, and forcing it into chronological order early makes the project feel overwhelming. Instead, capture memories as they come — fragments, scenes, conversations — and organize them later. Use this prompt any time a memory surfaces: ``` I want to capture a memory for my memoir. Here's what I remember: [Write the memory in whatever form comes naturally — a few sentences, a rough description, even a list of details.] Help me turn this into a 2–3 paragraph memoir scene. Ask me any questions you need first. Keep my voice — plain, honest, and personal. ``` Aim for five to ten scenes before you start thinking about structure. Each session can be short — twenty minutes is enough to capture one memory fully. If you think of writing as sitting down at a keyboard, this process can feel harder than it needs to be. Speaking is often much easier. Most of us tell stories out loud every day; we just don't call it writing. Here's the simplest dictation workflow: - On an iPhone, open the Notes app and tap the microphone icon. - On Android, open Google Docs and tap the microphone in the keyboard. - Speak your memory out loud — tell it the way you'd tell it to a friend. - Copy the transcribed text and paste it into ChatGPT or Claude. - Use this cleanup prompt: ``` I recorded a memory by speaking it out loud. Here's the rough transcription: [paste text] Clean up the transcription while keeping my natural speaking style. Fix the grammar but don't make it formal. Then help me expand the most interesting parts with follow-up questions. ``` Many people find that once they stop "writing" and start talking, the memories come much more freely. Once you have ten or more captured scenes, you're ready to look for structure. This is where AI is genuinely useful as an organizer. Paste in your scene summaries (just a sentence or two describing each one) and ask for help finding the shape of the story. ``` Here are summaries of memoir scenes I've written so far: [List each scene with a one-sentence description] What themes do you see running through these? Suggest 3–5 possible chapter groupings that would give the memoir a clear arc — not necessarily in chronological order. Explain the logic behind each grouping. ``` You don't have to follow the AI's suggestions. Use them as a mirror. Sometimes seeing your own material organized by a neutral outside perspective helps you see what the story is really about. The best memoir I've read was written by a 74-year-old who told me she "couldn't write a sentence." She dictated her memories into her phone during morning walks, then spent Sunday afternoons cleaning them up with AI help. What made it extraordinary wasn't the writing — it was the specificity. She remembered the smell of her grandmother's kitchen, the exact words her father said when she graduated high school, the color of the car she drove on the worst day of her life. AI can't invent that detail. Only she had it. The biggest risk in AI-assisted memoir writing is that the final product sounds polished but impersonal — like a magazine article about your life rather than your life. The solution is to continually pull the writing back toward your natural voice. When you read an AI-generated paragraph and something feels off, use this correction approach: ``` This paragraph from my memoir doesn't quite sound like me. I tend to [describe your speaking style — for example: "use short sentences, say things directly, tell a joke sometimes, start stories in the middle"]. Here's what you wrote: [paste paragraph] Rewrite it to sound more like me. Keep all the facts; just change the tone and rhythm. ``` Do this for any passage that feels stiff or generic. Your voice is the thing that makes a memoir worth reading — AI can support it but not replace it. Most memoirs include difficult material — loss, conflict, regret, relationships that ended badly. AI can help you write about hard things without getting stuck in the emotion of them. When you need to write something painful: ``` I need to write about a difficult memory for my memoir. I want to include it honestly without dwelling on it or making readers uncomfortable. Here's what happened: [describe the memory] Help me write a brief, honest account — 1–2 paragraphs — that acknowledges what happened without dramatizing it or making it heavier than it needs to be. ``` You can always go back and add more depth later once you've broken through the difficulty of writing about it at all. ## What to try next Writing a memoir is a long project, but the first scene is just one conversation. If you found prompts useful here, the guide to [writing better AI prompts](/guides/how-to-write-prompts) will help you get more out of every session. And if you want to use AI for shorter writing projects while you build confidence, [things to ask ChatGPT](/guides/things-to-ask-chatgpt) has ideas for everyday uses that build your comfort with the tools. --- # Is Alexa Always Listening? The Truth About Smart Speakers Source: https://aiforregulars.com/guides/alexa-always-listening-privacy The question comes up constantly: "Is Alexa listening to everything we say?" It is a fair thing to wonder when you have a small device sitting in your kitchen that responds to your voice. The honest answer is more nuanced than a simple yes or no — and once you understand how it actually works, you will be in a better position to decide what privacy settings make sense for you. ## How the wake word really works Your Echo device runs a small program on-device that listens for the word "Alexa" (or whatever wake word you have set). This part happens entirely on the device itself, without sending anything to Amazon. The microphone is active, but it is comparing sounds against a narrow pattern, not recording or uploading a continuous stream of your conversations. When the device detects the wake word, the light ring turns on — that visual cue tells you recording has started. From that moment, audio is streamed to Amazon's servers, processed, and a response is sent back. The recording typically lasts just a few seconds: from the wake word through the end of your request. Think of it like a smoke detector. The sensor is always on, but it only triggers when it detects something specific. ## False activations: when it wakes up by mistake Alexa can mishear words that sound similar to "Alexa" and activate unintentionally. You might be watching TV, having a conversation, or listening to a podcast, and the light ring turns on. When this happens, the device does briefly send audio to Amazon's servers — Amazon receives a short clip that did not include an intentional command. This is what people mean when they worry about Alexa listening in. It does happen, but it is not the same as continuous recording. Amazon says it reviews a small percentage of these clips to improve accuracy. You can opt out of that review. I noticed my Echo in the kitchen activating during a cooking show I was streaming — the host apparently said something that sounded close enough to "Alexa" to trigger it. Once I knew to watch for the light ring, I became much more aware of when it was happening. Moving the device away from the TV reduced false activations significantly in my experience. ## What actually gets stored Amazon stores your voice request recordings in your account history. By default, these are kept indefinitely unless you delete them or set up auto-deletion. Amazon uses this data to improve Alexa and to personalize responses. You can review everything Amazon has stored by going into the Alexa app. ## Privacy settings you should actually review **Delete your voice history** Open the Alexa app, go to **More → Alexa Privacy → Review Voice History** (in some versions labelled **Review Alexa History**). You can listen to individual clips, delete specific ones, or delete all recordings at once. **Set up automatic deletion** In the same Alexa Privacy section, look for **Manage Your Alexa Data**. You can set recordings to auto-delete after 3 months or 18 months. This runs automatically so you do not have to remember to clean it up. **Opt out of human review** Under **Manage Your Alexa Data**, go to the **Help Improve Alexa** section and turn off the **'Use of voice recordings'** toggle. This prevents Amazon from using your recordings for service improvement and human review. **Use the physical mute button** All Echo devices have a physical button that turns off the microphone entirely. The light ring turns red to show the mic is off. When the mic is muted, Alexa cannot hear anything — including the wake word. This is the most reliable option if you are having a private conversation. **Review connected skills** Go to **Skills & Games** in the Alexa app and look at what third-party apps you have connected. Each skill may have its own privacy policy and data practices. Remove any you no longer use. ## What about kids in the house? If children use your Alexa, their voice recordings are stored in the same account unless you set up a separate child profile with Amazon Kids. Children's voices are smaller in volume and more likely to trigger false activations. It is worth setting up auto-deletion and reviewing what skills are enabled, especially interactive ones. For a broader look at AI tools and children's privacy, see our guide on [AI toys and kids' privacy](/guides/ai-toys-kids-privacy). ## The bigger picture Smart speakers are a trade-off like many connected devices. They offer real convenience in exchange for having a microphone in your home. The key is understanding what that microphone does and does not do, and adjusting the settings so the trade-off feels right for you. ## What to try next If you want to understand more broadly what tech companies know about you, [what does AI know about me?](/guides/what-does-ai-know-about-me) is a good next read. And if you want to take control of your data across services, [how to opt out of AI training and delete your data](/guides/opt-out-ai-training-delete-data) walks you through the steps. --- # Alexa vs Google Home vs Siri: Which Smart Speaker AI Is Best? Source: https://aiforregulars.com/guides/alexa-google-siri-comparison You can already talk to one of these three assistants without buying anything new — Siri is on your iPhone, Google Assistant is on most Android phones, and Alexa lives on Echo speakers and Fire devices. But which one should you actually rely on? Tech reviewers consistently put each assistant through the same categories of everyday tasks to see where they shine and where they fall short — here is how they compare based on published testing and vendor documentation. ## The 10-task methodology Same question or command, spoken out loud, to each assistant in turn. Tasks cover the most common reasons people use voice AI: 1. Setting a timer 2. Playing a specific song 3. Answering a factual question ("What is the capital of Australia?") 4. Giving today's weather 5. Adding an item to a shopping list 6. Turning off a smart light 7. Calling a contact 8. Sending a text message 9. Telling a follow-up question ("And what is its population?") 10. Setting a reminder for next Tuesday ## Head-to-head results | Task | Alexa | Google Home | Siri | |---|---|---|---| | Setting a timer | Excellent | Excellent | Excellent | | Playing music | Excellent (Amazon Music/Spotify) | Excellent | Good (Apple Music focus) | | Factual questions | Good | Excellent | Good | | Weather | Excellent | Excellent | Excellent | | Shopping list | Excellent (Amazon integration) | Good | Good (Reminders app) | | Smart home control | Excellent | Very good | Good (HomeKit only) | | Phone calls | Good (with Alexa Calling) | Excellent (Android/Google) | Excellent (iPhone) | | Sending texts | Limited | Good (Android) | Excellent (iPhone) | | Follow-up questions | Good | Excellent | Good | | Reminders | Good | Excellent | Excellent | *Ratings reflect consensus from published expert reviews (CNET, Tom's Guide, Wirecutter) and official vendor documentation. Individual results may vary by device model and software version.* ## Amazon Alexa — the smart home champion Alexa's biggest strength is the sheer number of smart home devices it supports. If you have smart lights, a thermostat, a doorbell camera, a robot vacuum, or a TV, there is a very good chance they all work with Alexa. The Skills system lets you add hundreds of third-party integrations — everything from ordering a pizza to checking a flight. **Where it wins:** - Smart home control across the widest range of brands - Amazon shopping (add to cart, reorder household items by voice) - Drop-in calling between Echo devices in different rooms **Where it falls short:** - Factual answers sometimes feel thinner than Google's - Sending regular SMS texts is cumbersome without an Android phone nearby - The experience outside the Amazon ecosystem is narrower **Best for:** Households with a mix of smart home devices, Amazon Prime members, families who want room-to-room intercom. ## Google Home — the knowledge leader Google Assistant is powered by the same search engine that handles billions of questions every day. That shows up most clearly when you ask it something specific: follow-up questions, multi-part queries, and obscure facts all get handled more smoothly than the competition. **Where it wins:** - Answering follow-up questions naturally ("What year was that?" after "Who wrote Romeo and Juliet?") - Connecting with Google Calendar, Gmail, and Google Maps - Works seamlessly on Android phones **Where it falls short:** - Smart home device support is broad but not as wide as Alexa's - Shopping integration is Google Shopping, not Amazon - Less useful if you live in the Apple ecosystem **Best for:** Android users, people who ask a lot of general knowledge questions, heavy Google Workspace users. ## Siri — the Apple ecosystem winner Siri's advantage is integration depth within Apple devices. If you have an iPhone, an Apple Watch, a Mac, and AirPods, Siri can handle tasks across all of them seamlessly — reading messages, navigating in Apple Maps, controlling Music, and using app shortcuts from your phone. **Where it wins:** - Sending iMessages and making FaceTime calls - Controlling Apple devices and apps hands-free - Privacy focus: many requests process on-device - Apple CarPlay in the car **Where it falls short:** - HomeKit smart home support is narrower than Alexa - Factual answers are sometimes less complete than Google's - Much less useful on non-Apple devices **Best for:** iPhone/iPad/Mac users who want seamless cross-device voice control. In my own home I have both an Echo and an Android phone. I use Alexa for lights and timers because it is always listening and the smart home response feels instant. I use Google on my phone when I am asking something factual and need a reliable answer. If I owned only Apple devices I would probably never need either — Siri handles the day-to-day perfectly within that world. ## Which one should you choose? The honest answer is: **whichever ecosystem you are already in**. If you use iPhone, iPad, or Mac as your main devices, Siri is already built in and handles most of what you need. If you have an Android phone and use Gmail and Google Calendar, Google Home is a natural extension of your existing setup. If you have a mixed-brand smart home or shop on Amazon regularly, an Echo with Alexa is hard to beat for the price. None of them are so much better that it is worth switching your whole device setup to access them. But if you are buying your first smart speaker and have no strong preference yet, Google Home gives the strongest general-purpose performance out of the box. ## What to try next Curious whether the newer AI chatbots do a better job at answering questions than these voice assistants? [Best AI Chatbot for Beginners Compared](/guides/best-ai-chatbot-beginners) covers ChatGPT, Gemini, and others side by side. If you want to try talking to an AI chatbot on your phone directly, [ChatGPT Voice Mode](/guides/chatgpt-voice-mode) explains how to set it up in a few minutes. --- # Alexa+ Guide: What's New Now That It Runs on AI Source: https://aiforregulars.com/guides/alexa-plus-new-features If you have had an Echo on your kitchen counter for years, you might have noticed that Alexa has gotten a lot smarter lately. That is not your imagination. Amazon rebuilt Alexa around a large language model — the same kind of AI that powers ChatGPT — and called the result Alexa+. Here is what that means for you and how to actually use the new features. ## Step 1: Check Whether Alexa+ Is Available on Your Device Go to the Alexa app on your phone. Tap **More** (bottom right), then **Settings**, then **Alexa+**. If your Echo is compatible, you will see an option to enable or subscribe to Alexa+. If you do not see it, your device may not be supported yet, or the rollout may not have reached your region. If Alexa+ is available, follow the on-screen steps to turn it on. Alexa+ is included at no extra cost with an Amazon Prime membership. If you are not a Prime member, it costs $19.99 per month — confirm current pricing in the app before signing up. Once enabled, your Echo does not need any updates — the AI runs in Amazon's cloud. The clearest way to feel the difference is to give Alexa a task that has two parts. Say something like: *"Alexa, add oat milk to my shopping list and remind me to check the expiration date on the current one when I get home."* Old Alexa would handle only the first part or get confused. Alexa+ connects both steps and confirms each one. You can also follow up naturally: *"Actually, add almond milk instead."* Alexa+ remembers what you said earlier in the same conversation. Ask: *"What's a good recipe for chicken soup?"* Then follow up with *"How long does it take?"* and then *"Can I make it in a slow cooker?"* — without repeating "chicken soup" each time. This works best on Echo Show devices where Alexa can display text, but it also works on standard Echo speakers. Say *"Alexa, help me write a birthday message for my sister — she's turning 60 and loves gardening."* Alexa+ will draft something and read it to you. You can say *"Make it shorter"* or *"Add a joke"* and it will revise. On an Echo Show, you can see the text on screen. You used to need the Alexa app to build a routine step by step. Now you can just say: *"Alexa, every weekday at 7 a.m., turn on the kitchen lights, tell me the weather, and play the news."* Alexa+ will create that routine for you automatically. You can still edit it later in the app if you want to fine-tune it. I was skeptical at first — my old Echo had let me down so many times mid-sentence that I had stopped trusting it. But when I tried Alexa+ with a genuinely messy request ("add bananas to my list, then cancel my dentist reminder and move it to Thursday at 2"), it handled all three steps without breaking a sweat. That single test sold me. Alexa+ is impressive but has real limits. It cannot browse the web freely, access your email without explicit setup, or make phone calls on your behalf. It also occasionally gets facts wrong — if Alexa+ tells you something important (a medication name, a business address), double-check it. Think of it as a very helpful assistant who sometimes guesses rather than checks. ## What to try next If you want to compare Alexa+ to Google and Siri side by side, read [Alexa vs. Google vs. Siri: Which Smart Speaker Assistant Is Best?](/guides/alexa-google-siri-comparison). And if you have privacy questions about how much Alexa hears, [Is Alexa Always Listening? What Actually Happens](/guides/alexa-always-listening-privacy) walks through exactly what gets recorded and how to delete it. --- # Am I Using AI Too Much? A Simple Self-Check Source: https://aiforregulars.com/guides/am-i-using-ai-too-much If you've noticed yourself opening ChatGPT, Gemini, or a similar app before you've even tried to think a problem through — or if a family member has said something like "you ask that thing about everything now" — it's a completely reasonable question to sit with. There's no shame in asking it, and no single right answer. Here's a practical way to check in with yourself, based on what researchers and clinicians who study this are actually finding. ## Signs Worth Paying Attention To None of these on their own means anything is wrong. But if a few of them feel familiar, it's worth a closer look. **You open it for a quick question and lose an hour.** You meant to ask one thing and look up later, well past when you meant to stop, having followed the conversation somewhere you didn't plan to go. **You think about it when you're not using it.** You catch yourself planning what you'll ask next, or feeling a pull to check in with it, in moments that have nothing to do with an actual task in front of you. **You ask it to decide things you used to decide yourself.** Not "what's a good recipe idea" but "what should I do about this argument with my sister" or "should I take this job" — handing over judgment calls that are really yours to make, on a habitual basis. **You reach for it before you reach for a person.** When something upsets you, the chatbot becomes the first stop, ahead of a friend, a partner, or just sitting with the feeling for a minute — not occasionally, but as the default. **You feel a flicker of unease when you can't get to it.** The app is down, you're somewhere without service, or you've decided to try going a day without it — and it doesn't feel neutral, it feels like something's missing. **You keep asking for reassurance on the same thing.** You've gotten an answer, but you ask again in a slightly different way, and again, looking for a response that finally makes the uncertainty go away. ## What This Isn't It's worth being clear about what the research actually supports, because a lot of the language around this online is more alarming than the evidence justifies. There is no officially recognized "AI addiction" or "chatbot dependence" diagnosis in either the DSM-5, the manual used in the U.S., or the ICD-11, the international one. Researchers who study behavioral addiction — the same field that studies things like gambling and gaming disorder — have specifically cautioned against treating heavy AI use as a clinical disorder before there's solid long-term evidence for it, warning that both "moral apprehension" and jumping straight to overpathologizing are real risks in either direction. That doesn't mean the feeling behind the question is imaginary. It means the honest, current answer is "pay attention to how it's affecting you," not "you have a diagnosable condition." If you're noticing real distress, or that it's getting in the way of work, sleep, or relationships, that's worth taking seriously regardless of what label does or doesn't exist for it yet. ## Small Changes That Tend to Help If a few of the signs above sounded familiar, the fix usually isn't quitting AI tools altogether — it's adding a small amount of friction back into how you use them. **Try answering yourself first, even badly.** Before you open the app, take thirty seconds to guess at your own answer. You can still ask afterward — but you'll notice how often your own first instinct was closer to fine than you assumed. **Give it a job, not a standing invitation.** Use it for a specific task — draft this, summarize that, explain this term — rather than leaving a chat open as a running companion for whatever crosses your mind during the day. **Notice what you're actually asking for.** If you catch yourself typing out a feeling rather than a question — venting about a fight, describing a worry — that's a moment where a person, even briefly, will generally serve you better than a chatbot that can't actually know you. **Set one specific boundary and see how it feels.** Not "use AI less" in the abstract, which is hard to act on, but something concrete: no chatbot for the first decision of the morning, or not while you're actually talking with someone else in the room. The check I find most useful isn't "how often am I using this" — it's "could I stop right now, mid-thought, without it bothering me." If the honest answer is yes, that's a good sign, whatever the frequency looks like. If it's no, that discomfort is the actual signal worth paying attention to, more than any number of chats. ## What to Try Next If the concern isn't really about your own use but about a teenager in your life, [warning signs of unhealthy AI companion use in teens](/guides/teen-ai-companion-warning-signs) covers a related but different pattern — AI apps built to feel like a relationship. And if you want a grounded sense of what these tools are actually built to do (and not do), [What Does 'AI' Actually Mean?](/guides/what-does-ai-mean) is a useful place to reset your expectations. ## Sources - [Can AI Dependence Develop Into AI Addiction?](https://www.forbes.com/sites/robertglatter/2026/07/19/can-ai-dependence-develop-into-ai-addiction/) — Forbes - [8 Tips for Managing AI Dependence](https://www.psychologytoday.com/us/blog/living-your-best-life/202601/8-tips-for-managing-ai-dependence) — Psychology Today - [Chatbot Addiction and Its Impact on Psychiatric Diagnosis](https://www.psychiatrictimes.com/view/chatbot-addiction-and-its-impact-on-psychiatric-diagnosis) — Psychiatric Times --- # Android Can Now Translate Phone Calls in Real Time — Here's How to Use It Source: https://aiforregulars.com/guides/android-real-time-call-translation Calling a family member who speaks a different language used to mean juggling a translation app on speakerphone, or having a third person relay messages back and forth. Android can now handle the translation automatically, in both directions, during the call itself — each person hears the conversation in their own language, in real time. ## What Real-Time Call Translation Actually Does When you turn on call translation before a call, your Pixel acts as an interpreter between you and the person you're calling. You speak in English; the other person hears it in their language. They reply in their language; you hear it in yours. Neither of you has to switch languages or stop the conversation to type into a separate app. The AI listens to both sides of the call, translates each sentence, and plays back the translated audio while the call is live. There's a brief gap — usually one to two seconds — while the phone processes each sentence. Longer or more complex sentences may come through sounding slightly more synthesized than natural speech. This guide covers live voice calls. If you need to translate a photo, a document, or text on your screen instead, [How to Translate a Photo or Document with AI](/guides/translate-photo-document-ai) covers that workflow. ## What You Need Before You Start Real-time call translation expanded significantly with the **June 2026 Pixel Drop**. It's available on **Pixel 10 and newer** models — Pixel 10, Pixel 10 Pro, Pixel 10 Pro XL, Pixel 10 Pro Fold, and Pixel 10a — running **Android 17**, which launched as a stable release on June 16, 2026. The feature works through the standard Phone app on your Pixel and runs entirely on-device — no internet connection is needed for translation. ## How to Set It Up Go to **Settings → About phone → Android version**. You need Android 17 or later. If you see an earlier version, go to **Settings → System → System update** and install any available updates before continuing. Open the **Phone** app — the default dialer on your Pixel. Tap the **three-dot menu** in the top-right corner, then tap **Settings**. In Phone settings, look for **Voice translate**. Tap it, then enable **Use voice translate** to turn on the feature. Choose your language (the one you'll be speaking) and the language the other person speaks. The phone will translate both directions automatically during the call. You can save this pairing for future calls or change it each time. Call the person you want to speak with. Once the call connects, tap the **Translate** button on the call screen. Both parties will hear an automated announcement confirming that translation is active. Talk at a natural pace. After you finish a sentence, a short pause lets the phone translate and play your words for the other person. Then the other person speaks, and you hear their side translated into your language. The alternating rhythm takes a few exchanges to feel natural, but most people find it settles into a comfortable flow within a minute or two. ## What a Translated Call Sounds Like The most noticeable quality of a translated call is the pace. After each person finishes speaking, there's a pause of roughly one to two seconds while the translation plays. It's consistent and predictable — nothing like a dropping connection — but it does slow the natural back-and-forth of conversation. The translated voice is AI-generated: cleaner and more even than the original speaker's voice, but also noticeably synthetic. Short, direct sentences come through clearly and naturally. Longer sentences with multiple clauses may sound slightly uneven or lose a nuance. Speaking in shorter, complete thoughts helps both the accuracy and the naturalness of what the other person hears. > "I'm going to use a translation feature during our call — there'll be a short pause after each of us speaks while it translates. That's completely normal, so don't hang up." Sending a message like that before calling someone who isn't expecting it prevents a lot of confusion about whether the line is working. When I test voice translation features, I start with simple, short exchanges before moving to anything complicated — "How are you?" and "I'll call you back tomorrow" before a detailed conversation about plans or problems. The feature works better when both people speak in complete, shorter sentences. Once you find that rhythm, it flows much better than you'd expect for a first try. ## What to Watch Out For **Quality varies by topic.** Everyday conversation — greetings, family news, travel plans, simple logistics — translates well. Technical vocabulary, medical or legal language, proper names, and idioms are more likely to come through inaccurately. For anything important, use the translation to understand the gist, then confirm specific details separately. **The other person is always told about the translation.** When you activate call translation, an announcement plays for both parties. There is no way to translate a call without the other person knowing. This is a deliberate privacy protection built into the feature. **Translation is fully on-device and private.** Voice Translate processes everything locally using the Tensor G5 chip — no audio is sent to Google's servers, no internet connection is required, and nothing is stored. Your conversation stays on your device. **Speaking clearly helps more than speaking slowly.** Many people instinctively slow down a lot when they know someone is being translated, which can make the rhythm feel stilted. A normal pace with clear, complete sentences works better than an exaggerated slow pace. ## If You Don't Have a Pixel Phone Real-time call translation in Android 17 is currently specific to Pixel hardware. If you're on a different Android phone or an iPhone, **Google Translate's Conversation mode** is the most practical alternative: 1. Open **Google Translate**. 2. Tap the **Conversation** button (the two speech bubble icon at the bottom). 3. Select the two languages. 4. Put the phone on speakerphone between you and the other person, or hold it near the speaker during a phone call. Each person speaks in turn; the app translates and can read the translation aloud. It's more hands-on than built-in call translation, but it works on any Android phone or iPhone with the app installed. For a full comparison of translation tools across different situations, [Which Translation Tools Actually Work — Tested and Compared](/guides/translation-tools-compared) has detailed coverage. ## What to Try Next Real-time call translation is one of several AI features quietly built into Pixel phones that most people haven't found yet — [Hidden AI Features on Your Smartphone You Probably Don't Know About](/guides/hidden-ai-features-smartphone) covers more of them. If you use Gemini on Android and want to get more out of it, [A Beginner's Guide to Gemini on Android](/guides/gemini-android-guide) is the natural next step. --- # Apple Intelligence Setup Guide: What It Does Source: https://aiforregulars.com/guides/apple-intelligence-guide Apple released its own AI system — called Apple Intelligence — built directly into the operating system. It is not a separate app you download; it is woven into the keyboard, notifications, photos, and Siri across iPhone, iPad, and Mac. If you have a supported device, it is already there waiting to be turned on. Here is how to get it running and what it is actually useful for day to day. Apple Intelligence requires specific hardware. On iPhone, you need an iPhone 15 Pro or Pro Max, or any iPhone 16 model. On iPad, you need an M1 chip or newer, or the iPad mini (7th generation or later) — it uses an A17 Pro chip rather than an M chip, but it is fully supported. On Mac, any Apple Silicon Mac (M1 or later) works. To check what chip your device has, go to **Settings → General → About** on iPhone or iPad, or **Apple menu → About This Mac** on a Mac. If you see an M1, M2, M3, or M4 chip listed, your device is supported. On iPhone or iPad, go to **Settings → Apple Intelligence & Siri**. If your device is supported and your language/region is available, you will see an option to turn on Apple Intelligence. Tap it and follow the on-screen steps. On Mac, go to **System Settings → Apple Intelligence & Siri**. The toggle is in the same location. After enabling it, some features may take a few minutes to become fully active as the system finishes setting up. Writing Tools is one of the most immediately useful Apple Intelligence features. Anywhere you can type — in Mail, Messages, Notes, or a third-party app — select some text, then tap or right-click and look for **Writing Tools**. You will see options like **Proofread** (fixes grammar and spelling), **Rewrite** (rephrases your text), **Make Shorter**, **Make Longer**, and **Change Tone**. Pick one and Apple Intelligence rewrites the selected text. You can accept the change or undo it. This works entirely on-device for most requests. If you get a lot of notifications, Apple Intelligence can summarize groups of them. Go to **Settings → Notifications** and look for **Summarize Notifications** or the Apple Intelligence option. Once on, your iPhone groups notifications from the same app and shows a short summary instead of a pile of individual alerts. This is especially helpful for group chats and email apps, where you might get a dozen messages that boil down to one piece of information. Open the Photos app and select a photo. Tap **Edit**, then look for the **Clean Up** button in the editing toolbar. Use your finger or the pencil to paint over an object you want to remove — a photobomber in the background, a piece of trash on the ground, a sign that distracts from your subject. Apple Intelligence fills in the area with what it thinks the background should look like. Results are best on simple backgrounds like sky, grass, or plain walls. I used the Clean Up tool on a family photo where a car was visible in the background. It took about three seconds to paint over it and the result looked natural — the grass filled in cleanly. It is not perfect for every situation, but for simple backgrounds it is genuinely impressive and much faster than doing it manually in another app. Apple Intelligence makes Siri more capable in two main ways. First, Siri can now understand what is on your screen. While looking at a restaurant website, you can say "Add this to my calendar for Saturday" and Siri finds the address and details automatically. Second, Siri is gaining the ability to take actions across apps — for example, asking it to find a photo you shared with someone recently and send it again, with Siri searching your library and starting the message. Reviewers find these cross-app personal-context actions work well in simple cases but can be hit-or-miss for more specific multi-step requests; simpler, direct commands are more reliable than chained ones. When you ask Siri something it cannot answer on its own, it may offer to send your question to ChatGPT. Apple has built in an optional integration where ChatGPT handles more complex requests. You will always see a prompt before anything is sent to OpenAI — Apple does not forward your questions silently. You can tap to allow it for that question, or you can say no. If you use ChatGPT regularly, you can also sign in with your OpenAI account in **Settings → Apple Intelligence & Siri → ChatGPT** to access any subscription features you already have. ## What Apple Intelligence does not do Apple Intelligence does not have access to all your files, browsing history, or app data unless you specifically use it within those apps. It does not learn from your activity in the way some other AI systems do. Each request is processed with privacy as a priority — for on-device tasks, your data never leaves your phone at all. It also cannot do everything a standalone AI assistant like ChatGPT can. For complex research, long document analysis, or extended conversations, a dedicated AI app will serve you better. ## What to try next Curious how Apple's approach compares to Google's on Android? Our [Gemini AI on Android guide](/guides/gemini-android-guide) covers the same kind of setup and everyday features from Google's side. And if you are thinking about which AI tools are worth your time overall, [ChatGPT vs. Gemini](/guides/chatgpt-vs-gemini) gives you a clear comparison. --- # How to Make Your Resume ATS-Friendly with AI Source: https://aiforregulars.com/guides/ats-friendly-resume-with-ai You spend hours on your resume. You apply for a job. And nothing happens. Sometimes it is not the resume itself — it is that an automated system filtered it out before anyone ever read it. Understanding why that happens is the first step to fixing it. ## What is an ATS, exactly? An ATS — short for Applicant Tracking System — is software that companies use to manage job applications. When you hit "Submit," your resume often goes through this system first. It scans for specific words and phrases from the job posting. If your resume does not include enough of them, it can be ranked low or filtered out automatically — before a single human being ever sees your name. It is not a person making that call. It is software. That is why the same great resume can get ignored at one company and land an interview at another — it depends on how well your words match their keywords. The good news is that AI can compare your resume to any job posting and tell you exactly what is missing. Open the job you want to apply for. Select all the text and copy it. Include the full description — the required skills section is especially important, but the whole thing helps AI spot patterns in what the employer is looking for. Open your resume in Word, Google Docs, or whatever you use. Select all the text and copy it. Plain text works better than a PDF — if you only have a PDF, paste what you can and note that some formatting may not transfer cleanly. Open ChatGPT (or any AI tool). Paste this prompt, then paste the job description and your resume below it: ``` I'm applying for a job and want to make my resume more ATS-friendly. Below is the job description, followed by my current resume. Please: 1. List keywords and phrases from the job description that are missing from my resume 2. Suggest where I could add them naturally 3. Flag any formatting choices that could confuse an ATS Job description: [paste here] My resume: [paste here] ``` AI will give you a specific list of gaps and suggestions — not vague advice, but the actual words that appear in the job posting and are absent from your resume. Most resumes have bullet points that are too vague — things like "Helped with projects" or "Worked on customer issues." These do not tell the ATS (or a recruiter) much. Pick your three weakest bullets and ask AI to improve them: ``` Rewrite these resume bullet points to be more specific and impactful. Use strong action verbs. Where I've put [X], I'll fill in the real number. [paste your bullet points here] ``` Then fill in the actual numbers and details. "Managed a team" becomes "Managed a team of 6 customer service reps." "Improved sales" becomes "Helped increase monthly sales by 18% over six months." The specifics matter. When I helped my sister update her resume after a long break from work, the biggest problem was not what she had done — it was how she described it. She had real, solid experience, but she used the same language she had used 15 years ago. When we ran her resume against three different job postings in ChatGPT, the keyword gaps were obvious immediately. We fixed it in one afternoon, and she got a callback within the week. ATS systems look for standard headings. If yours are creative — "My Story," "Where I've Been," "Things I'm Good At" — the system may not recognize them and skip the content underneath. Use these standard headings instead: - Work Experience (or just Experience) - Education - Skills - Certifications (if you have them) - Professional Summary (optional, at the top) If you are not sure whether your current headings are standard, ask AI: "Do any of my resume section headings look unusual for an ATS?" A few common formatting choices can trip up ATS software even if your content is perfect. - **Tables and columns:** Many ATS systems read text left to right across the page, so a two-column layout can come out scrambled - **Headers and footers:** Your name and contact info in the document header may never get read — put it in the main body as well - **Graphics and logos:** Remove them entirely from an application resume - **Unusual fonts:** Stick to Arial, Calibri, Times New Roman, or Georgia If you are unsure about your resume's formatting, ask AI: "What resume formatting choices could cause problems for an ATS?" ## The goal is to get in front of a human An ATS-friendly resume is not about gaming the system — it is about making sure your actual qualifications get seen. Once a person is reading your resume, your experience and personality take over. AI just helps you get past the first door. **What to try next:** Once your resume is ready, you will need a cover letter. Learn [how to write a cover letter with ChatGPT that does not sound like AI](/guides/chatgpt-cover-letter), or jump to [AI interview prep](/guides/ai-interview-practice) to practice for the conversation that comes next. --- # AI Junk Is Flooding the Internet: How to Find Real Information Again Source: https://aiforregulars.com/guides/avoid-ai-slop-online If you've noticed that searching for a recipe, a health question, or a home repair tip keeps returning pages that all sound vaguely helpful but never quite answer what you actually came to find out — you're not imagining it. A 2025 study by the web analytics company Ahrefs found that roughly 74% of newly published web content is now AI-generated. Merriam-Webster even named "slop" its Word of the Year in 2025, partly because the word captured something real: a flood of low-effort, AI-generated content churned out to fill web pages rather than to help the people reading them. The good news is that the real information is still out there. You just need to know where to look, and what to skip. ## What "AI Slop" Actually Means (and Why It's Everywhere) "AI slop" is a nickname for content produced quickly and cheaply using AI writing tools — not to answer a question well, but to rank in search engines and attract clicks. It's not always wrong, exactly, but it tends to be vague, generic, and interchangeable. Read one article about "the 10 best ways to remove a stripped screw" and you've essentially read them all. Why is there so much of it? Because publishing AI content at scale is cheap. A website owner can generate hundreds of articles on different topics in a single afternoon, optimized to appear in Google results without requiring a human expert to actually research or write them. The business model rewards volume over quality, and search engines are only starting to catch up. This guide focuses on everyday informational content — recipes, health questions, DIY how-tos, general advice. It's not about spotting AI-generated fake news (that's covered in [How to Spot AI-Generated Fake News](/guides/ai-generated-fake-news)) or fake product reviews on shopping sites (covered in [How to Spot Fake AI-Generated Product Reviews](/guides/spot-fake-ai-product-reviews)). ## Four Signs You're Reading AI Slop You don't need tools or browser extensions to recognize most AI slop. These patterns show up reliably. **No named author, or an author with no verifiable history.** Real writers and experts put their names on their work. When an article lists "Staff Writer" or "Editorial Team" as the author, or links to a bio page that's a single generic paragraph with no social media, no outside work, and no real history — that's a signal the content wasn't written by someone who actually knows the topic. **Generic superlatives without real substance.** Phrases like "the ultimate guide," "everything you need to know," and "the most comprehensive overview" are AI writing tics. They take up space without telling you anything. Real experts tend to get specific quickly because they have something specific to say. **Stock photos or no original photos at all.** A DIY article about fixing a leaky faucet, written by someone who actually fixed a leaky faucet, usually includes photos of their actual faucet. A slop article has a nice royalty-free image of a shiny faucet in a perfect bathroom. This isn't conclusive on its own — plenty of legitimate sites use stock photography — but it's worth noting alongside other signals. **Repetitive structure that sounds complete but leaves you with questions.** AI-generated content often follows a predictable template: introduction, numbered list with short paragraphs under each point, brief conclusion. The structure looks organized, but if you read to the end and still don't know the specific thing you came to find out, the article was optimized for appearance, not usefulness. My personal shortcut: when I'm looking for practical advice — a recipe, a home repair technique, a health symptom — I add "reddit" or "forum" to the end of my search. Real people asking and answering real questions are still easy to find this way, and the messiness of genuine human experience (someone's sourdough that didn't rise, the plumber who showed up and found the problem was something else entirely) is exactly what makes those posts more useful than a polished article. ## Better Search Habits That Cut Through the Noise A few small changes to how you search make a big difference in what you find. **Add "reddit", "forum", or "quora" to your query.** This doesn't guarantee quality, but it steers you toward communities of real people rather than content farms. "Reddit best cast iron skillet care" surfaces threads where actual home cooks disagree, share what didn't work, and update each other — which is a lot more honest than a top-ten listicle. **Use Google's Discussions and Forums filter.** After you search on Google, click "More" under the search bar, then select "Discussions and forums." This filters results to show content from community platforms rather than article sites, and it's one of the fastest ways to find genuine human experience on a topic. **Check the author and the date.** Before you read an article all the way through, spend 30 seconds finding the author's name and looking them up. Do they have a real web presence, other published work, credentials for the topic they're writing about? Also check the publication date — for anything that changes (drug interactions, software instructions, tax rules, product compatibility), an article from three years ago may not just be useless but actively misleading. **Search for the specific, awkward version of your question.** Slop articles target polished, generic queries. The more specific and oddly phrased your question is, the more likely you are to surface a real forum thread or a specialist's answer. "Why does my sourdough smell like nail polish remover" will find you a real answer faster than "sourdough starter problems." ## Where AI Slop Shows Up Differently **Google search results.** This is where most people first notice the problem. The first page of results for popular how-to queries is increasingly dominated by long articles that answer the question in theory but not in practice. The Discussions filter described above helps a lot here. **Amazon and other shopping sites.** AI-generated product descriptions and listing copy are widespread. This is separate from the fake-review problem (covered in [How to Spot Fake AI-Generated Product Reviews](/guides/spot-fake-ai-product-reviews)) but compounds it: it becomes hard to evaluate whether a product is what you need when the description was generated rather than written by someone who knows it. Checking independent review sites and watching YouTube reviews of the specific product helps. Amazon limits Kindle book uploads to three per day per author — and even that cap still lets a small operation flood a category with AI-written titles. **YouTube.** AI-narrated videos with generic visuals — often called "faceless channels" — have multiplied across topics like investing, health symptoms, and DIY repairs. They use AI-generated scripts dressed up with stock footage or simple animations. The tell is usually a flat, neutral voice reading content that sounds plausible but never gets into genuine specifics, and no real engagement in the comments from the person behind the channel. **Facebook and other social feeds.** AI-generated posts, pages, and even full fake accounts have become a significant share of what circulates in social media groups, especially on topics like health, finance, and politics. A post that sounds authoritative and cites a surprising statistic without linking to a source deserves the same skepticism as an article with no named author. ## Safer Sources for Health, Recipes, and DIY For topics where getting it wrong actually matters, going to a known-quality source first is worth the extra step. **Health questions.** For anything medical — symptoms, drug interactions, treatment options — go directly to established medical publishers: Mayo Clinic, Cleveland Clinic, the NHS (for UK readers), or MedlinePlus (run by the US National Library of Medicine). These sites have named medical editors, update policies, and review processes. For personal medical situations, a doctor or pharmacist still beats any web source. **Recipes.** Large recipe databases that have been around for years and test their recipes with actual cooks — Serious Eats, Smitten Kitchen, NYT Cooking, BBC Good Food — are a lot more reliable than the recipe articles that appear in generic search results. Searching directly on one of these sites, or adding the site name to your Google query, skips the guessing game. **DIY and home repair.** YouTube is genuinely better than most articles for DIY, because it's harder (though not impossible) to fake a real repair on video. Look for channels where a real person is doing the work, explains why they're making each choice, and shows what happens when something goes wrong. Forum communities like Reddit's home improvement subreddits and specialty sites like This Old House are similarly more reliable than generic article sites. ## What to Watch Out For **This problem isn't going away soon.** Search engines are actively working on ways to surface genuinely useful content and down-rank low-quality AI content, but it's an ongoing arms race. The habits in this guide will remain useful regardless of how that plays out. **Not all AI-assisted content is slop.** Some legitimate publishers use AI tools for drafting or editing, then have human editors review and verify the result. The presence of AI in the production process isn't automatically a red flag — what matters is whether the content is accurate, specific, and genuinely useful. **Be careful with AI search tools themselves.** Services like Google's AI Overviews, Microsoft Copilot in Bing, and Perplexity can surface answers quickly — but they have their own accuracy issues and sometimes confidently present incorrect information, especially on niche topics or recent events. Treat their answers as a starting point to verify, not a final word. More on this in [AI Search vs. Google: Which One Should You Trust?](/guides/ai-search-vs-google). ## What to Try Next If you want to go deeper on telling AI-written content from human writing, [How to Tell If Something Was Written by AI](/guides/how-to-tell-if-text-is-ai) covers the specific language patterns to watch for. And if you're wondering whether AI tools can reliably find good information for you at all, [AI Search vs. Google: Which One Should You Trust?](/guides/ai-search-vs-google) gives an honest comparison of both approaches. --- # Best AI Study Tools for Students This Fall — and Which Ones Schools Actually Allow Source: https://aiforregulars.com/guides/back-to-school-ai-study-tools-2026 A new school year is the perfect time to get intentional about how AI fits into studying — before your student has figured it out on their own, possibly in ways the school doesn't allow. The good news: some AI tools are genuinely designed to help students understand material rather than just hand them finished answers. The less-good news: only 22% of schools actively provide or recommend approved AI tools to students, which leaves a lot of guesswork about what's actually fair game. Here's a clear-eyed look at five tools worth knowing, what each is good for, and how to navigate the school-policy question. ## The Ground Rule Most Schools Are Landing On Before diving into tools, the principle worth understanding is this: most school districts that have issued AI policies are drawing the same line — **AI may explain, AI may not produce.** Using AI to understand a concept, check your reasoning, or practice a skill is generally treated the same as asking a teacher for help. Using AI to write your essay or complete your homework, then submitting that as your own work, is cheating under most school policies. A 2026 survey found that 72% of parents believe AI should be part of education, and 58% of students say they already use AI as a tutor rather than a shortcut. Fifty-one percent also know that submitting AI-generated work is cheating. The gap is real: the tools are already in use, but the rules are still catching up. ## Khanmigo — The AI Tutor That Asks You Questions Back **What it is:** Khan Academy's AI tutor, built into the free Khan Academy learning platform. **Best for:** Math, science, and standardized test prep (SAT, LSAT); students who want to understand the steps, not just copy the answer. **The standout feature:** Khanmigo is built to teach rather than answer. When you ask "what's the answer to this algebra problem?", it responds with a guiding question that leads you toward figuring it out yourself. It's the closest thing to a private tutor available at 11 p.m. the night before a test. **Scale as a signal:** Khanmigo grew from 40,000 to 700,000 student users in a single school year — a sign that both students and schools are finding it useful enough to keep using. **Cost:** A small monthly or annual fee for students and families; Khan Academy's educational content remains free. Check khanacademy.org/khanmigo for current pricing. **Age fit:** Middle school and up; strongest for high school and college prep. **School policy:** Khanmigo is widely considered acceptable because it explains and guides rather than producing finished work. If your school has an approved tools list, it often appears there. ## NotebookLM — Research Grounded in Your Own Materials **What it is:** Google's AI research tool that reads only the documents you upload — no general internet browsing, just the sources you provide. **Best for:** Writing research papers, summarizing textbook chapters or lecture notes, preparing for exams from class materials. **The standout feature:** You upload your sources — a PDF of a textbook chapter, a Google Doc of class notes, a research article — and NotebookLM answers questions strictly from those materials, citing the specific passage it draws from. It can't invent facts from the internet because it's only reading what you gave it. **Cost:** Free with a Google account. Google has an education tier for schools. **Age fit:** High school and up; most useful for students who already have source material to work with. **School policy:** Because NotebookLM summarizes and explains sources rather than generating original text from scratch, many teachers are comfortable with students using it for research and note-taking. Check before incorporating its output directly into an assignment. > **Try this:** Upload a chapter from your textbook and ask "What are the three most important concepts in this chapter and how do they connect?" It builds study notes faster than reading and re-reading the same pages. ## MagicSchool — What Your Teachers Are Using **What it is:** An AI platform built for educators, with more than 80 tools for generating lesson plans, quizzes, rubrics, differentiated assignments, and parent communications. **Why students should know about it:** MagicSchool is what many teachers are using behind the scenes to build the assignments and assessments that students receive. Understanding that teachers are also using AI to create work — not just students using it to complete work — is part of the fuller picture. **Student-facing tools:** MagicSchool is expanding its tools for students, so check whether your school has adopted it and ask your teacher directly. **Cost:** Free tier for teachers; school and district licensing also available. **School policy:** MagicSchool is built for educators and designed to meet FERPA and COPPA privacy requirements, making it one of the safer choices when a school or district adopts it. When a student uses any AI study tool, I think the best check is one question afterward: "Can you explain what you just learned?" If they can walk through the idea in their own words, the tool helped them understand something. If they have a finished answer they can't explain, that's the real problem — not which app they used. ## Socratic by Google — Quick Explanations, Step by Step **What it is:** A free Google app (iOS and Android) where students can photograph a homework question and receive a step-by-step explanation. **Best for:** Getting unstuck on a specific problem in math, science, history, or literature — especially when it's late and there's no teacher to ask. **The standout feature:** Socratic shows its reasoning. Rather than handing you the answer, it walks through the logic step by step, links to relevant Khan Academy videos, and explains the underlying concept. **Cost:** Free. **Age fit:** Middle school and up; simple enough for younger students with parent guidance. **School policy:** Using Socratic to understand a concept is generally acceptable under most school AI policies. Using it to copy answers for submission is not. > **Try this:** Photograph a problem you're stuck on and ask "Can you explain the first step?" rather than letting it jump straight to the solution. ## Duolingo Max — Language Practice With an AI Conversation Partner **What it is:** Duolingo's premium tier that adds AI-powered conversation roleplay and explain-my-mistakes features to the standard Duolingo language lessons. **Best for:** Students taking a second language who want more speaking and listening practice than classroom time allows. **The standout feature:** The roleplay feature lets students practice realistic conversations — ordering at a café, asking for directions — with an AI that corrects mistakes gently and explains why. That kind of low-stakes practice is hard to get enough of in a typical 45-minute class. **Cost:** A monthly subscription above the standard Duolingo Super tier; check duolingo.com for current pricing. **Age fit:** 13 and up for independent use; younger students should use with a parent. **School policy:** Language practice tools are generally treated the same as using a flashcard app or a language lab — fine for supplementary practice, not a substitute for assigned classroom work. ## What Schools Actually Allow — and How to Find Out in Five Minutes School AI policies are being written and rewritten right now. A rule your school announced last year may have already been updated, and many schools still don't have a formal policy at all. **How to check:** 1. Search your school or district's website for "AI policy" or "artificial intelligence" in the student handbook section. 2. If nothing specific comes up, look for the acceptable use policy (AUP) — newer versions are adding AI clauses. 3. When in doubt, email your teacher directly: "Is it okay to use [tool name] to help me understand this assignment?" A specific question about a specific tool gets a clear answer. **What the policies almost always agree on:** - AI for tutoring, explaining, and brainstorming: generally allowed - AI for generating work you then submit as your own: not allowed - Using a personal account with a tool that should use a school account: check with IT first ## What to Watch Out For **AI makes mistakes.** Khanmigo and NotebookLM are more careful than general-purpose chatbots, but every AI tool can produce errors. Never take a calculation or a factual claim at face value without checking against a trusted source. **Privacy matters.** Be thoughtful about uploading assignments that contain other students' names or school-confidential information. Check whether a tool stores your data and for how long. **The shortcut trap.** The tools above are useful for learning. But any of them can slide into shortcuts if the goal shifts from understanding to getting something done fast. The test is simple: can you explain what you learned? **Age limits apply.** Most AI tools require account holders to be 13 or older. If your student is younger, [AI Learning Apps for Younger Kids](/guides/ai-learning-apps-kids) covers age-appropriate options. ## What to Try Next For a broader look at study strategies that work well alongside AI — not just which apps to use, but how to use them without letting AI do the thinking — [How Students Are Using AI to Actually Study Better](/guides/ai-for-studying) covers the approach. If you're a parent looking for a way to start the conversation about responsible AI use at home, [Talking With Your Kids About AI and Homework](/guides/kids-chatgpt-homework-parents-guide) has specific talking points and questions worth asking. --- # Best AI Budgeting Apps Compared: Which One Is Worth Linking Your Bank? Source: https://aiforregulars.com/guides/best-ai-budgeting-apps Every budgeting app promises to help you save money. The AI ones go further: they categorize your spending automatically, spot unusual charges, and tell you when you're about to blow your budget before you do. But handing over access to your bank account is a big decision. Here's what each app actually does with that access — and whether the trade-off is worth it. ## The Privacy Question First Before comparing features, let's be direct about something: these apps see your financial life in detail. Most use **Plaid** or a similar financial connector, which reads your transaction history with read-only access. That means the app can see your purchases — but cannot move, transfer, or withdraw money. What varies is what apps do with that data beyond the app itself. Some sell anonymized data to third parties. Others are explicit that they don't. The key things to check before signing up: - Does the app sell or share your transaction data? - Can you delete your account and data completely? - What happens to your data if the company is acquired? ## The Top AI Budgeting Apps ### 1. YNAB (You Need a Budget) — Best for Active Budgeters YNAB uses a "give every dollar a job" method, and its AI assists by auto-categorizing transactions and flagging when categories go over budget. It's more hands-on than other apps — a feature if you want to truly understand your money, and a downside if you want a set-and-forget tool. **Privacy:** YNAB states it does not sell user financial data. **Cost:** Paid subscription (~$15/month or ~$109/year); free trial available. **Best for:** People who want full control and don't mind spending 10 minutes a week reviewing. ### 2. Monarch Money — Best All-Around Monarch shows net worth, spending trends, investments, and a cash flow dashboard all in one place. Its AI surfaces insights like "your grocery spending went up 30% this month" and lets you set goals with automatic progress tracking. **Privacy:** Does not sell personal financial data; explicit data deletion available. **Cost:** Paid subscription (~$15/month monthly, or ~$100/year (about $8/month) annually); free trial available. **Best for:** People who want a full financial picture, not just spending. ### 3. Copilot — Best for iPhone Users Copilot (iOS only) has a clean design and an AI that learns your spending patterns over time. The "rules" feature lets you teach it how to categorize tricky transactions — like the gym membership that shows up as a vague code. **Privacy:** Does not sell data to third parties. **Cost:** Paid subscription (~$13/month); free trial available. **Best for:** Apple device users who want the most polished experience. ### 4. Cleo — Best Free Option (With Personality) Cleo has a chatbot interface that talks to you in a casual, sometimes sarcastic tone. Ask it "how much did I spend on food this week?" and it'll tell you — and possibly roast you a little. It's genuinely fun to use, which helps with consistency. **Privacy:** Cleo is upfront that the free tier is supported by financial product recommendations. **Cost:** Free tier available; Cleo Plus and higher subscription tiers for more features. **Best for:** Younger users or anyone who finds traditional budgeting apps boring. ## Side-by-Side Comparison | App | Free plan | Bank linking | Sells data? | Best feature | |---|---|---|---|---| | YNAB | No (trial) | Yes | No | Strict budget method | | Monarch Money | No (trial) | Yes | No | Full financial overview | | Copilot | No (trial) | Yes | No | Clean iOS design | | Cleo | Yes | Yes | Recommends products | Chatbot interface | I tried linking my bank to four different apps while writing this. The setup process takes about 5 minutes and usually involves choosing your bank, entering your online banking credentials, and completing a two-factor authentication step. None of them asked for more than that. I'd still recommend reading the privacy policy first — it takes 2 minutes and tells you a lot about how a company thinks about your data. ## The No-Bank-Link Alternative If you're not comfortable connecting your bank account to any app, you're not out of options. You can use [ChatGPT as a monthly budget tracker](/guides/chatgpt-monthly-budget) — just export a CSV from your bank and paste the transactions into ChatGPT with a prompt like "categorize these expenses and show me where I'm spending the most." It's more manual, but your data never leaves the chat. ## What to Try Next If you decide to go the app route, start with Cleo's free tier to see if you'll actually use a budgeting app before paying for one. If you're also dealing with tax prep, [AI Tools for Taxes](/guides/ai-for-taxes) covers how AI can help you prepare without overpaying for an accountant. --- # Best AI Chatbot for Beginners Compared Source: https://aiforregulars.com/guides/best-ai-chatbot-beginners Picking your first AI chatbot can feel overwhelming when every tech headline mentions a different one. The good news: most of them are free to try, and the differences matter more for power users than for beginners. Here is a plain-English rundown of the five most popular options, what each one is actually good at, and which one to start with depending on what you want to do. ## Quick comparison table | Chatbot | Made by | Free tier | Best for beginners who want... | |---|---|---|---| | ChatGPT | OpenAI | Yes | A little of everything | | Gemini | Google | Yes | Google Workspace integration | | Copilot | Microsoft | Yes | Microsoft Office integration | | Claude | Anthropic | Yes | Long documents, careful writing | | Perplexity | Perplexity AI | Yes | Fast answers with sources | ## 1. ChatGPT — the all-rounder ChatGPT is the one that started the current wave of AI chatbots, and it remains the most versatile option for everyday use. **What it does well:** Writing, editing, explaining concepts, answering questions, summarizing text, helping with emails, brainstorming. The free version handles all of these. **What it lacks:** Text conversations on the free tier are unlimited since August 2026, but image generation (up to two images per day) and file uploads still have daily limits. Extended reasoning modes and features such as Deep Research require a paid plan. **Who it suits:** Anyone trying AI for the first time who wants to explore without committing to a specific use case. **Try it at:** chatgpt.com ## 2. Gemini — best if you use Google Gemini is Google's AI chatbot. The reason beginners who use Google products tend to prefer it is simple: it connects to Gmail, Google Docs, Google Drive, and Google Calendar. You can ask it to summarize an email, help draft a document, or find something in your Drive. **What it does well:** Google Workspace tasks, answering questions with up-to-date Google Search results, image understanding if you take a photo and ask a question. **What it lacks:** Less polished for pure creative writing tasks compared to ChatGPT or Claude. **Who it suits:** Gmail and Google Docs users who want AI that connects to their existing files. **Try it at:** gemini.google.com ## 3. Microsoft Copilot — best for Office users If you spend your day in Word, Excel, or Outlook, Copilot works right inside those apps instead of making you switch to a separate window. **What it does well:** Drafting and editing documents in Word, writing formulas in Excel, summarizing email chains in Outlook, taking meeting notes in Teams. **What it lacks:** The version built into Office apps requires a paid Microsoft 365 Copilot subscription. The free web version at copilot.microsoft.com works like a general chatbot but does not connect to your files. **Who it suits:** Office workers already in the Microsoft 365 ecosystem. **Try it at:** copilot.microsoft.com ## 4. Claude — best for long documents and careful writing Claude is made by Anthropic and has a reputation for being particularly good at reading and writing long pieces of text. It can handle a very long document — like a contract or a research report — and pull out the key points without getting confused. **What it does well:** Summarizing long documents, nuanced writing tasks, thoughtful back-and-forth conversation, explaining complex topics patiently. **What it lacks:** No built-in image generation, and the free tier has daily limits. **Who it suits:** Anyone who works with long documents, needs patient step-by-step explanations, or finds other chatbots too brief or too casual. **Try it at:** claude.ai When I need to understand a long agreement or summarize a lengthy report, I reach for Claude first. It tends to stay accurate across the full length of a document without drifting off-topic partway through — something I have noticed other tools do on very long texts. ## 5. Perplexity — best for quick research Perplexity works differently from the others. Instead of just generating an answer from memory, it searches the web and then summarizes what it finds — with clickable sources so you can verify everything. **What it does well:** Research questions, fact-checking, getting current information with references, comparing products or options. **What it lacks:** Less good for creative writing, document editing, or long conversations. It is really a research tool more than a general assistant. **Who it suits:** Anyone who wants fast, sourced answers to questions and does not want to wade through a list of search results. **Try it at:** perplexity.ai ## Which one should you start with? If you are completely new and not sure what you want to use AI for: **start with ChatGPT**. It is the most versatile, has the largest community of people sharing tips, and the free tier is genuinely useful. If you already use Gmail or Google Docs every day: **try Gemini** — the integration saves real time. If your work revolves around Word and Outlook: **Copilot** is worth a look, but check whether the free version meets your needs before paying for the Office integration. If you want to read fewer sources and just get a quick, cited answer: **Perplexity** is the most efficient tool for that specific job. ## What to try next Once you pick one, the fastest way to get comfortable is to start small. Our guide on [how to use ChatGPT for the first time](/guides/how-to-use-chatgpt) walks you through your first conversation step by step. And if you are trying to decide whether the paid upgrade is worth it, [Is ChatGPT free?](/guides/is-chatgpt-free) breaks down exactly what you get without paying. --- # Best AI Photo Editors for Non-Designers Source: https://aiforregulars.com/guides/best-ai-photo-editors You don't need to be a graphic designer to make your photos look great. AI photo editors do the heavy lifting — fixing lighting, removing backgrounds, erasing unwanted objects — with just a tap or click. The tricky part is knowing which tool to reach for. Here's a plain-English breakdown organized by what you actually want to do. ## At a Glance: Which Tool for Which Job | Use case | Best free option | Best paid upgrade | |---|---|---| | Remove background | Adobe Firefly (free credits) | Adobe Firefly (premium credits) | | Portrait touch-up | Snapseed (Portrait mode) | Remini Pro | | Fix blurry / old photos | Remini (free tier) | Topaz Photo AI | | Change or swap background | Canva | Adobe Photoshop AI | | Remove an object or person | Snapseed (Healing tool) | Samsung Galaxy AI, Apple Clean Up | | Add AI-generated elements | Adobe Firefly | Adobe Firefly (premium credits) | | Quick one-tap enhance | Google Photos (Magic Eraser, Unblur) | Lightroom AI | | Creative filters and effects | PicsArt | PicsArt Gold | ## The Tools, One by One ### Canva Magic Edit — Best for Beginners Canva started as a graphic design tool, but its AI photo editing features have become genuinely useful for everyday photos. Magic Edit lets you brush over part of a photo and type what you want — "remove the car" or "add a sunset sky." The Background Remover works in one click. Note: the one-click Background Remover is a Canva Pro feature ($15/month) — it is not included in the free plan. It runs entirely in your browser, so nothing to install. The free plan covers most basic edits. **Best for:** People who want simple, visual editing without learning new terms. ### Snapseed — Best Free Mobile App Snapseed is a Google app that's been around for years, and it's still one of the best free photo editors on phones. The Healing tool removes objects (tap and paint over whatever you want gone). The Portrait and Face Enhance tools smooth skin, brighten eyes, and fix lighting — all on-device, so your photos don't go to any server. **Best for:** Phone editing, privacy-conscious users, anyone who wants a capable free tool. ### Remini — Best for Fixing Old or Blurry Photos Remini uses AI to sharpen and enhance photos that are blurry, low-resolution, or old. Upload a fuzzy photo and it comes back noticeably clearer. It also does portrait enhancement that can make faces in photos look more detailed. The free tier lets you do a limited number of enhancements per day. The paid subscription price varies by offer and billing cycle — check the current price inside the app before subscribing. The free version adds watermarks to enhanced photos. **Best for:** Restoring old family photos, fixing blurry snapshots. ### Adobe Firefly — Best for Creative Edits Adobe Firefly is Adobe's AI tool that's built into products like Photoshop and Express, and also available standalone at firefly.adobe.com. You can remove objects, extend the edges of a photo (called Generative Fill), or add something entirely new by typing a description. The standalone browser version has a free tier with a monthly credit allowance. **Best for:** Users who want creative, generative edits — adding or changing parts of a scene. ### Google Photos — Best If You Already Use It If you store photos on Google Photos, you already have access to several AI tools: Magic Eraser removes distracting objects, Unblur sharpens faces, and Photo Unblur works on older snapshots. These are built right into the app — no separate tool needed. Magic Eraser, Unblur, and Photo Unblur are free for all users. The more powerful Magic Editor is limited to 10 saves per month on the free plan — unlimited access requires a Pixel device or a Google One Premium (2TB+) subscription. **Best for:** People who already use Google Photos for storage. I use Snapseed for most day-to-day edits on my phone because it's fast, free, and keeps my photos on my device. For anything more creative — like extending a background or removing a large object — I switch to Canva in the browser. ### Lightroom AI — Best for Multiple Photos at Once Adobe Lightroom's AI tools (Denoise, Masking, Enhance) are powerful for editing a batch of photos at once — useful if you take a lot of pictures and want them all looking consistent. It's more of a "photographer's tool," but the AI features make it much more accessible than it used to be. **Best for:** Anyone with a larger photo library who wants AI-assisted batch editing. ### Fotor — Best Web Alternative to Canva Fotor is a browser-based editor similar to Canva with strong AI features: one-click background removal, AI portrait retouching, and an AI image enhancer. The interface is clean and beginner-friendly. The free tier has watermarks on some outputs; the Pro plan costs $8.99/month (less on an annual plan) and removes all watermarks. **Best for:** People who want an alternative to Canva with similar simplicity. ## What You Don't Need to Pay For Many of the most useful AI photo features are already free: - Snapseed (full app, no subscription) - Google Photos Magic Eraser (free for basic use) - Adobe Firefly browser tool (limited free credits) You only need a paid plan if you edit a large volume of photos or need advanced features like batch processing or higher export quality. ## A Note on Privacy When you upload a photo to any AI tool, it goes to that company's servers. For casual shots, this is usually fine. For photos of children, medical images, or anything sensitive, stick to on-device tools like Snapseed or Google Photos (which processes on your own device when possible). ## What to try next Once you know your editor, learning to [remove specific objects or people from photos](/guides/ai-remove-object-photo) takes just a few minutes. If you want to go further with creating new visuals, the [beginner's guide to creating AI images for free](/guides/create-ai-images-free) walks through the options. --- # Best AI Resume Builders Compared Source: https://aiforregulars.com/guides/best-ai-resume-builders Polishing a resume used to mean hiring a professional career coach or spending hours staring at a blank Word document. AI resume builders change that equation. They can draft bullet points, suggest keywords, and export a clean PDF — all in under an hour. The catch is that there are dozens of them, and marketing copy all sounds the same. This guide cuts through that noise with a straightforward look at what each tool actually does. ## What to look for in an AI resume builder Before diving into specific tools, here is the short checklist I use when evaluating any resume builder: | Criterion | Why it matters | |---|---| | ATS optimization | Many companies filter resumes automatically; a beautiful design means nothing if the software rejects it first | | Template quality | Clean, readable formatting still impresses human reviewers | | AI writing quality | Does it produce natural-sounding bullet points, or stiff filler? | | Export options | PDF is a must; Word export helps when an employer asks for it | | Free tier limits | Some tools tease features then lock the download behind a paywall | | Privacy policy | You're uploading work history and contact info | ## The tools, compared ### Teal — Best all-around free option Teal is built specifically for job seekers. Its free plan includes a resume builder, a job tracker to organize your applications, and a basic keyword match feature. The AI helps you rewrite bullet points by asking what you actually did — you type a rough description, and it returns polished language. What sets Teal apart is the job tracker. You can save a job listing, and Teal will highlight which keywords your resume is missing. This feedback loop — apply, compare, refine — is genuinely useful and mostly free. The limitation: the free plan includes a one-time AI writing credit allotment — 10 bullet rewrites, 2 summaries, and 2 cover letter drafts. Once spent, AI generation locks until you upgrade. ### Jobscan — Best for ATS optimization Jobscan does one thing better than anyone else: it tells you exactly how well your resume matches a specific job listing. Paste in the job description, and Jobscan scores your resume and lists the missing keywords. It reads like a gap analysis, not a vague suggestion. This is especially useful if you have been applying to jobs and hearing nothing back. An ATS mismatch is often the silent reason. Jobscan's free plan gives you a limited number of scans per month; the paid plan removes that cap. The trade-off is that Jobscan's resume templates are functional but not especially attractive. Think of it as a diagnostic tool, not a design tool. Many people use Jobscan alongside a better-looking builder like Kickresume. ### Kickresume — Best for polished design Kickresume's templates look genuinely professional. There are dozens of them, and the layouts are modern without being flashy. The AI writing assistant helps you fill in each section — you describe your role, and it suggests bullet points you can edit. The free plan lets you create resumes using 4 available templates, with 40+ premium templates locked behind a paid plan. The paid plan unlocks everything and adds a cover letter builder that pulls details from your resume automatically. If presentation matters to you — creative fields, client-facing roles, or roles where a hiring manager actually reads the resume before any ATS does — Kickresume is worth the upgrade. ### Resume.io — Best for beginners Resume.io is the most beginner-friendly option on this list. The setup wizard walks you through every section with prompts and examples. You never face a blank box without guidance. The templates are clean and the export is reliable. The AI writing help is simpler than Teal or Kickresume — more about suggesting phrasing than deep rewriting — but for someone building their first resume or returning to the job market after years away, that simplicity is an advantage. One note: the free plan lets you build a resume but requires a paid subscription to download. This is a common pattern in the industry, so check before you invest significant time. ### ChatGPT (DIY approach) — Best for specific rewrites ChatGPT is not a resume builder in the traditional sense — there are no templates, no PDF export, no ATS scoring. But for rewriting individual bullet points or drafting a summary section, it is hard to beat. A prompt like "Rewrite this bullet point to start with a strong action verb and include a measurable result" often produces better output than the AI suggestions in dedicated tools, because you control the conversation. The downside is that you then need to paste the result somewhere that formats it properly. My approach: I use ChatGPT to rewrite bullet points until they feel sharp and specific, then I paste the final text into Kickresume for formatting and export. The two tools together give better results than either alone. ## Quick comparison table | Tool | Free tier | ATS check | AI writing | Best for | |---|---|---|---|---| | Teal | Generous | Yes (limited scans) | Good | All-around use | | Jobscan | Limited scans | Best in class | Basic | ATS optimization | | Kickresume | Multiple resumes, 4 templates (40+ premium locked) | Basic | Good | Design quality | | Resume.io | Build only, pay to download | No | Basic | Beginners | | ChatGPT | Yes | No | Excellent | Rewriting specific sections | ## Which one should you use? Start with Teal if you are in a full job search and want one place to track applications and optimize your resume. Add Jobscan if you're applying to roles at companies that use automated screening (most mid-size and large companies do). Use Kickresume if design quality matters for your industry. Try ChatGPT for any section you want to rewrite in your own voice. What to try next: read [how ATS systems actually screen your resume](/guides/how-ai-screens-resumes) so you understand what these tools are optimizing for, then check the guide on [writing an ATS-friendly resume with AI](/guides/ats-friendly-resume-with-ai) for step-by-step prompts. --- # Best AI Travel Planners Tested: Which App Builds the Best Trip? Source: https://aiforregulars.com/guides/best-ai-travel-planners Planning a trip used to mean hours on comparison sites, scrolling through review pages, and juggling spreadsheets. AI travel planners promise to cut that down to minutes. But which ones actually deliver? We put the top options through the same scenario — a 7-day solo trip to Japan in October with a $3,000 budget — drawing on published reviewer comparisons and each tool's documented feature set. Here's how they stacked up. ## What to Look For in an AI Travel Planner Before diving in, here's what I used to judge each tool: | Criteria | What it means | |---|---| | Itinerary quality | Does it build a logical, day-by-day plan? | | Budget awareness | Does it respect your stated budget? | | Local detail | Does it suggest off-the-beaten-path spots, not just tourist traps? | | Booking links | Does it connect to hotels, flights, or activities? | | Ease of use | Can a non-tech person get started in under 2 minutes? | ## The Top AI Travel Planners ### 1. ChatGPT — Most Flexible Overall ChatGPT is still the best general-purpose travel planner. You can have a real back-and-forth conversation: "Add one more museum day," "make the budget $500 cheaper," or "what's the best neighborhood to stay in for nightlife?" It adapts on the fly better than any dedicated app. The downside: it won't automatically show booking links or maps. You need to copy ideas into another tool to book. **Best for:** People who like to customize every detail. ### 2. Gemini (with Google integration) — Best for Booking Connections Gemini connects to Google Maps, Google Flights, and Google Hotels in a way ChatGPT doesn't. Ask it to plan your trip and it pulls up real hotels on a map, current flight prices, and local reviews. The integration is genuinely useful. **Best for:** People who want everything in one place and already use Google. ### 3. Perplexity — Best for Research-Heavy Trips Perplexity cites its sources, which matters for travel. When it says "this temple closes on Tuesdays," you can click the source and verify. Great for unusual destinations or complex visa situations. **Best for:** Cautious travelers who want to verify everything. ### 4. Mindtrip — Best Dedicated Travel App Mindtrip is built specifically for travel. It generates an itinerary and displays it on a map with photos, booking links, and estimated costs alongside. The visual layout makes it easy to see where you're going each day. **Best for:** Visual thinkers who want a polished, map-based itinerary. ### 5. Layla — Best for Group Trips Layla lets you share a saved destination list or exported itinerary with travel companions so everyone can weigh in before you commit. It's not a live shared workspace, but the handoff works well for simple group decisions. **Best for:** Group travel where everyone has opinions. ## How They Compared on My Japan Test | Tool | Itinerary quality | Budget respect | Local gems | Booking links | Ease of use | |---|---|---|---|---|---| | ChatGPT | ⭐⭐⭐⭐⭐ | ⭐⭐⭐⭐ | ⭐⭐⭐⭐ | ❌ | ⭐⭐⭐⭐⭐ | | Gemini | ⭐⭐⭐⭐ | ⭐⭐⭐⭐ | ⭐⭐⭐ | ⭐⭐⭐⭐ | ⭐⭐⭐⭐ | | Perplexity | ⭐⭐⭐⭐ | ⭐⭐⭐ | ⭐⭐⭐⭐⭐ | ❌ | ⭐⭐⭐⭐ | | Mindtrip | ⭐⭐⭐⭐ | ⭐⭐⭐ | ⭐⭐⭐ | ⭐⭐⭐⭐⭐ | ⭐⭐⭐⭐ | | Layla | ⭐⭐⭐ | ⭐⭐⭐ | ⭐⭐⭐ | ⭐⭐⭐ | ⭐⭐⭐ | When using AI travel planners, always start with a very specific prompt: destination, dates, budget, and 2–3 things I care about (food, museums, hiking). Vague prompts get vague itineraries. The more detail you give upfront, the less back-and-forth you need. ## One Thing AI Can't Replace AI travel planners are excellent at structure and ideas. But they're not great at the messy real world: a restaurant that closed last month, a temple under renovation, or a local festival just announced. Always do a quick final check on the official tourism site for your destination before you book anything. ## What to Try Next Once you have your itinerary, make sure you're not overpaying on flights. Read [AI Price Alerts: Never Overpay for Flights Again](/guides/ai-flight-price-alerts) to set up automatic deal alerts. And if you want to manage your total trip budget with AI, [ChatGPT as Your Monthly Budget Planner](/guides/chatgpt-monthly-budget) walks you through that step by step. --- # My Boss Now Requires AI at Work — What Should I Do? Source: https://aiforregulars.com/guides/boss-requires-ai-at-work Getting a directive to "use AI" at work can feel like being handed a tool you've never seen before and told to build something with it — today. The good news: this is much less complicated than it sounds, and most people feel comfortable within a few weeks. Here's how to approach it calmly. ## First, find out what tools you're actually supposed to use "Use AI" means very different things in different workplaces. Before you do anything else, clarify with your manager: - **Which specific tool?** Is the company providing Microsoft Copilot, Google Gemini, a custom internal tool, or asking you to use your own judgment? - **Is there a license or account provided?** Or are you expected to use free tools on your own? - **Is there written guidance?** Many companies have a policy document — ask HR or your IT department if one exists. Knowing exactly what tool you're expected to use matters a lot, because different tools have very different rules about what happens to your data. ## What you should never paste into a free public AI tool This is the most important safety rule, and it applies even if you're asked to use AI and given no other instructions. **Never paste the following into a free AI tool** like the public version of ChatGPT or a free Google account: - Client names, addresses, or any identifying information - Internal financial data, sales figures, or unreleased product information - Employee personal information or HR records - Legal documents under confidentiality - Passwords, API keys, or any credentials Free AI tools may use what you type to improve their models. A company-approved tool — like Microsoft Copilot inside your organization's Microsoft 365 tenant, or Google Gemini for Workspace — has a contract that covers this differently. If you're unsure which category your tool falls into, ask your IT department before pasting anything sensitive. ## How to get up to speed quickly You don't need to become an expert. You need to find two or three tasks in your regular job where AI genuinely saves you time, and build from there. ### Start with low-stakes writing tasks AI tools are most reliable at drafting and editing text. Good places to start: - Summarizing a long email or document you received - Writing a first draft of a routine email or memo - Cleaning up meeting notes into bullet points - Generating a list of ideas you then review and trim These tasks are easy to verify — you can read what the AI produced and catch anything that's wrong before it leaves your desk. ### Use what's already built into your tools If your company uses Microsoft 365, Copilot is probably already in your Word, Excel, and Outlook. If you use Google Workspace, Gemini is built into Docs and Gmail. These are the safest starting points because they're already approved and connected to your existing files. ### Ask to watch someone else first If a colleague already uses AI regularly, ask if you can watch them work through a task for ten minutes. Seeing it in action answers more questions than reading about it. When I was first learning, I made a point of using AI only for things I would have done anyway — not to replace my judgment, but to speed up the parts I found tedious. Drafting a rough email outline took me from eight minutes to two. That small win made me want to find more places it could help. ## How to talk to your manager about concerns If you're worried about data privacy, accuracy, or being asked to do something that feels off, it's worth raising it directly. A simple way to frame it: *"I want to make sure I'm using this the right way — can you point me to our data policy for AI tools, or let me know what's off-limits to paste in?"* Most managers will appreciate the question. It shows you're taking it seriously. If you're asked to use AI to generate content that will be presented as original expert work without any human review, that's worth flagging. In most professional contexts, you are still responsible for what you submit, regardless of how it was drafted. ## What to try next Once you know your tool and your guardrails, the next step is learning to give it better instructions. Most of the frustration people feel with AI comes from prompts that are too vague. The guide on [AI prompts for office work](/guides/ai-prompts-office-work) has copy-paste examples you can use for common work tasks. If your company uses Microsoft tools, [Copilot in Word and Excel](/guides/copilot-word-excel-guide) walks you through the specific buttons and panels step by step. --- # Keep Your Brain Sharp with AI: Learning Tools for Seniors Source: https://aiforregulars.com/guides/brain-training-ai-seniors Learning something new is one of the most enjoyable things a human brain can do. And AI has quietly become a surprisingly good companion for that — patient, available at any hour, and endlessly knowledgeable about whatever topic you happen to be curious about. This guide looks at three practical ways older adults are using AI for mental engagement: as a quiz partner, as a language learning coach, and as a curiosity engine for exploring new subjects. It also gives you an honest look at what the evidence does and doesn't say about AI and brain health. ## The honest caveat first No app, puzzle, or AI tool has been proven to prevent dementia or cognitive decline. The science on "brain training" specifically — doing tasks that are cognitively demanding — is mixed. What researchers do generally agree on is that staying engaged, continuing to learn, staying socially connected, exercising, and sleeping well all support brain health. AI can support the learning and engagement part. It's not a cure or a treatment. If you or someone you know has concerns about memory or cognition, a conversation with a doctor is the right starting point — not an app. ## ChatGPT as a quiz partner One of the best uses of AI for older learners is as a patient, inexhaustible quiz partner. You can use it to review things you want to remember, test yourself on topics you're studying, or revisit things you once knew and want to refresh. Try prompts like: ``` Quiz me on the capitals of European countries. Ask one at a time and tell me if I'm right. ``` ``` I'm trying to remember the names of U.S. presidents in order. Start from the beginning and ask me each one. ``` ``` I used to know basic Spanish. Test me with simple vocabulary — start easy and get harder. ``` What makes AI especially useful here is that you can tell it exactly how you want to be challenged. If a quiz is too easy, say so. If you want an explanation when you get something wrong, ask. There's no judgment, no impatience, and no time limit. ## Language learning with AI Learning a new language — or refreshing one from decades ago — is one of the most cognitively demanding and rewarding activities an adult can undertake. AI has become a genuinely capable language practice partner. Here's how to use it: **For beginners:** Ask ChatGPT to teach you simple phrases for a specific situation. ``` I'm planning a trip to Italy and want to learn a few restaurant phrases. Teach me five and then quiz me. ``` **For returning learners:** Practice conversation at your own pace. ``` Let's have a simple conversation in French. Correct my grammar gently, and keep sentences short. ``` **For reading comprehension:** Ask it to translate and explain. ``` Here's a paragraph in Spanish. Translate it and explain any grammar structures I should know. ``` The voice mode available in ChatGPT's mobile app makes this especially effective — you can practice speaking and listening without needing to type. See our [ChatGPT Voice Mode guide](/guides/chatgpt-voice-mode) for how to set it up. I used ChatGPT to dust off my high school German before visiting Austria. What surprised me was how much I'd retained — I just needed something patient enough to help me find it. The fact that I could speak out loud and have it respond was the key. It felt like actually practicing, not studying. ## AI as a curiosity engine Some of the best learning has nothing to do with a curriculum. It starts with a question you happen to be curious about, and follows where that curiosity leads. AI is exceptionally good at supporting that kind of open-ended exploration. Examples of where this goes: ``` I watched a documentary about the ancient Romans. What would daily life have looked like for an ordinary person, not an emperor? ``` ``` I've always wondered how they built the pyramids. What do historians actually know for certain, and what is still debated? ``` ``` Explain quantum physics like I'm smart but have no science background. ``` The key is following your genuine curiosity. The more interested you are in a topic, the more you'll engage with what you're learning — and engagement is what makes learning stick. There's no wrong answer here, and AI won't run out of patience or topics. ## Getting started with voice (no typing required) If typing is a barrier — whether due to arthritis, low vision, or simple preference — ChatGPT's voice mode on a smartphone lets you have a full conversation by speaking. You ask questions out loud; it answers out loud. To get started: 1. Download the ChatGPT app on your phone (it's free) 2. Tap the voice icon in the message bar — it appears as a waveform symbol at the bottom of the screen 3. Speak your question naturally The experience feels more like talking to someone than using a computer, which many older adults find far less intimidating. ## A note on consistency The biggest factor in getting any benefit from mental activity isn't which tool you use — it's consistency. A few minutes several times a week is more valuable than one long session a month. Consider tying a short AI learning session to something you already do regularly, like your morning coffee, to make it a habit. ## What to try next For a practical guide to picking up a new language with AI, see [Learn a Language with ChatGPT](/guides/learn-language-with-chatgpt). If you'd like to explore what else you can use ChatGPT for on a daily basis, [Things to Ask ChatGPT](/guides/things-to-ask-chatgpt) is full of ideas organized by category. --- # How to Get a Real Human When Stuck in an AI Customer Service Loop Source: https://aiforregulars.com/guides/bypass-ai-customer-service You've been transferred three times, repeated your account number twice, and the chatbot just sent you back to the main menu. Again. Getting a human on the line feels harder than it used to — because companies have deliberately built systems to reduce how often it happens. These steps actually work. Some require patience; a few require knowing the right words. Before escalating, say or type the clearest possible request for a human. These phrases trigger human-routing rules in most systems: - "Agent" - "Representative" - "I want to speak with a person" - "Transfer to human" - "Operator" Say one clearly and wait. If the system ignores you and continues the automated flow, repeat the phrase after the next prompt. On phone systems, pressing **0** repeatedly still works on many older setups. If direct requests aren't working, certain words prompt automatic escalation in many call center systems. Try saying: - "Complaint" - "Supervisor" - "Cancel my account" - "Legal" - "Dispute" You don't have to be planning to cancel — these words route to humans because companies treat them as high-priority. Once you reach a real person, explain what you actually need. Hold times are longest on Monday mornings and the day after a holiday. The shortest wait times tend to fall Tuesday through Thursday, before 9 AM in the company's time zone. If you can choose when to call, choose those windows. Shorter queues mean agents who aren't already worn down by hours of back-to-back calls. Many major companies now offer a "call me back" feature in their mobile app. Instead of waiting on hold, you request a callback and a real agent calls you when it's your turn in the queue. Look for this in the app's Help or Contact section before you call. Banks, airlines, and major telecom providers are most likely to have it. Using it takes the same total time but lets you do something else while you wait. When phone and chat fail, a public social media post often works faster than you'd expect. Post on Twitter/X or Facebook tagging the company's official account and describing your issue briefly. Companies monitor their public mentions and often respond quickly because other customers can see the problem. Keep the post factual and polite — you want help, not a viral complaint. Include your case number if you have one so they can pull up your account right away. I once spent 45 minutes on hold with a telecom company, then posted a single tweet tagging their support account. Someone responded within 20 minutes and resolved my issue in the DMs. It's not guaranteed, but it's worth trying after two failed phone attempts. A written complaint sent to the right address often triggers a faster human response than a phone call because it creates a paper trail the company has to respond to. Ask ChatGPT to help you draft it: ``` Help me write a formal complaint email to [company name]. The issue is: [describe the problem]. I've already tried: [what you did]. I want: [specific resolution — refund, fix, explanation]. Keep it firm but polite. ``` Send to the company's official support email. For financial products, address it to the compliance department. In the US, you can also file a complaint with the Consumer Financial Protection Bureau (CFPB) — companies typically respond quickly to those filings. When you reach a frontline agent who can't help, ask calmly: "I understand this is outside what you can do. Can you transfer me to a supervisor or someone with the authority to resolve this?" Be specific about what you need resolved. "I need a full refund for a charge I didn't authorize" is clearer than "I need this fixed." A clear, specific request is easier for a supervisor to approve on the spot. ## What to Try Next If your issue involves understanding what a company's contract or bill actually says, [AI as a plain-English contract translator](/guides/ai-explain-contracts) has prompts that cut through confusing language fast. For writing any kind of professional message to a company, [AI message templates](/guides/ai-message-templates) has copy-paste examples ready to go. --- # Can You Trust ChatGPT? When It's Wrong and How to Check Source: https://aiforregulars.com/guides/can-you-trust-chatgpt ChatGPT is genuinely useful for a wide range of everyday tasks. It is also capable of stating wrong information with complete confidence, as if it were fact. Understanding when to trust it — and when to double-check — is the most important skill for using it well. This guide explains the patterns behind ChatGPT's mistakes and gives you a simple approach to fact-checking. ## What ChatGPT Is Good At ChatGPT is most reliable when you are asking it to do something with language rather than asking it to retrieve a specific fact. **Where it works well:** - Drafting, editing, and rewriting text (emails, letters, summaries) - Explaining concepts in plain language — "what does this term mean?" - Brainstorming ideas and making lists - Translating text between languages - Summarizing documents you paste in - Suggesting questions you should ask a doctor, lawyer, or professional For these tasks, even an imperfect response is often useful because you are editing and adapting it anyway. ## Where ChatGPT Makes Mistakes **Specific facts, statistics, and numbers.** ChatGPT sometimes gets precise figures wrong — dates, prices, population numbers, research statistics. It may produce a number that sounds plausible but is not accurate. Treat any specific number as unverified until you confirm it. **Events after its training cutoff.** ChatGPT's knowledge has a cutoff date. It does not know about things that happened after that date. As of mid-2026, the default version of ChatGPT has a training cutoff of around December 2025, so anything that happened after that date is outside its built-in knowledge. If you ask about a recent news event, recent prices, or a new law, it may not have the information — or it may guess. **Medical and legal specifics.** ChatGPT can explain what a medical condition or legal concept means in general terms, but it should not be your source for a specific diagnosis, treatment plan, or legal strategy. These require professional judgment and knowledge of your individual situation. **Names, quotes, and citations.** ChatGPT sometimes generates names, book titles, research papers, or quotes that do not exist or are misattributed. This is sometimes called "hallucination." If it cites a specific study or quotes a specific person, look it up before repeating it. **Calculations and math.** Basic arithmetic is usually fine, but complex calculations or multi-step math problems can go wrong. If numbers matter, check the math yourself or use a calculator. I once asked ChatGPT about the side effects of a common medication and got a mostly accurate answer — but one of the listed side effects was fabricated. It was stated with the same confidence as the real ones. I only caught it by checking against the official prescribing information. I now treat any medical detail from ChatGPT as "worth looking up," not "confirmed." ## How to Tell When to Verify Ask yourself: "What happens if this is wrong?" - If the answer is "not much" — I just want a rough idea, a creative suggestion, or a plain-language explanation — then relax. ChatGPT is useful here even if imperfect. - If the answer is "it matters a lot" — I am making a health decision, spending money, telling someone else this information, or submitting something official — then verify it before acting on it. The stakes tell you how careful to be. ## Simple Ways to Verify a ChatGPT Answer **Ask it to explain its confidence.** Type: "How confident are you in that? Is there anything I should double-check?" ChatGPT will often flag its own uncertainty when prompted directly. **Search for the specific claim.** Take the key fact or figure and search for it on a site you already trust — a government website, a major news outlet, the official website of an organization, or a medical resource like a hospital or national health authority. **Look at the primary source.** If ChatGPT references a law, a study, or an official policy, look up the original document rather than relying on ChatGPT's description of it. **Ask a professional for anything important.** For health, legal, and financial decisions, ChatGPT can help you understand the concepts and prepare questions — but a doctor, lawyer, or financial advisor should make the final call. ## The Right Mindset ChatGPT is most useful when you treat it like a well-read, hardworking assistant who sometimes gets confident about things they are actually fuzzy on. You would not sign a legal document just because a friendly assistant drafted it — you would read it yourself. The same instinct applies here. Use ChatGPT to do the heavy lifting, then spend a moment checking anything that actually matters. For most everyday tasks — writing, explaining, summarizing, brainstorming — that mindset will serve you well and let you get real value from the tool without getting burned by its occasional mistakes. ## What to Try Next Knowing ChatGPT's limits makes you a smarter user. If you want to get better answers in the first place — including knowing how to ask ChatGPT to flag its own uncertainty — see our guide on [how to write better prompts](/guides/how-to-write-prompts). And if you are wondering whether something was written by AI, [how to tell if text is AI-generated](/guides/how-to-tell-if-text-is-ai) walks through what to look for. --- # Charged After Canceling? How to Actually Stop an AI Subscription Source: https://aiforregulars.com/guides/cancel-ai-subscription You canceled your subscription — or thought you did. Then the charge appeared on your statement again. Before assuming something went wrong on the company's end, it's worth knowing that this situation is common enough that OpenAI's own support pages warn about it. The reason it happens almost always comes down to the same thing: there are two separate billing paths for most AI services, and canceling in one place does nothing about the other. ## Why Canceling in the Wrong Place Does Nothing AI services like ChatGPT, Gemini, and others can be paid for in two completely different ways: through the service's own website (billed directly to your card), or through an app you downloaded on your iPhone or Android phone (billed through Apple or Google). These are entirely separate billing relationships. OpenAI's help center explicitly warns that it's possible to end up with two active subscriptions at the same time — one through the web and one through the App Store — if you signed up through both. And here's the part that catches most people: if you subscribed through the iPhone app, Apple controls the billing. Canceling on ChatGPT.com doesn't cancel the Apple charge. Only Apple can do that. The same is true in reverse. Canceling through your iPhone settings only stops the App Store subscription. A separate web subscription keeps going. Before you can cancel anything, you need to know who's actually charging you. Look at the charge on your bank or credit card statement. - If it says **Apple** or **Apple.com/bill** — even if the description mentions ChatGPT or another AI service — the billing is through the App Store. - If it says **Google Play** or **Google\*[Service Name]**, it's through Google. - If it shows the service name directly (OpenAI, Google, Anthropic, etc.) without Apple or Google, it's a direct web subscription. You can also check from the service's account page. Log into the website and look for a billing or subscription section in your account settings. If it shows no active paid subscription, the payment is going through an app store, not the website. This step surprises people, but it's worth doing before you cancel anything. Check both places: the service's website billing page AND your phone's app store subscriptions. - **iPhone/iPad:** Go to **Settings > [your name] > Subscriptions**. - **Android:** Open the Play Store, tap your profile photo in the top right, then go to **Payments & subscriptions > Subscriptions**. If you see the same service listed as active in both places, you're being double-charged. This is a documented issue — OpenAI has acknowledged it in their help documentation. You'll need to cancel both subscriptions separately. Once you know where the billing lives, cancel there — not somewhere else. **If billed directly through the website:** Log into the service's website. For ChatGPT: go to chatgpt.com, click your profile photo, then **Manage my plan**, then **Cancel plan**. For Gemini: go to gemini.google.com, click your account icon, then **Manage subscription**. For other AI services, look for "Billing" or "Subscription" in your account settings. **If billed through Apple (iPhone/iPad):** Go to **Settings > [your name] > Subscriptions**, find the subscription, and tap **Cancel Subscription**. Do this from the iPhone or iPad you originally used to subscribe. **If billed through Google Play (Android):** Open the Play Store app, tap your profile photo, go to **Payments & subscriptions > Subscriptions**, find the subscription, and tap **Cancel subscription**. Cancel at least 24 hours before your next renewal date to make sure the cancellation processes in time. I always screenshot the final cancellation confirmation screen — the one that says "Your subscription has been canceled" with a date. That screenshot has saved me twice: once when a charge appeared anyway and I needed proof for the company's support team, and once for a bank dispute. It takes two seconds and you'll thank yourself later. Don't assume — confirm. After canceling, check the subscription page again (in the app store or on the service website) and look for a line that says something like "Your subscription ends on [date]" or "You will not be renewed." An active subscription showing a future end date with no renewal planned means it worked. If you're still seeing "Active" with no end date, the cancellation may not have gone through. Try again, or contact support directly. If you have confirmation that you canceled before a charge date but were billed anyway, request a refund from whoever charged you. **If the service charged you directly:** Contact their support and show when you canceled. Most services have a refund process for charges after confirmed cancellation. EU customers additionally have statutory withdrawal rights that may allow a refund within a set window after the billing date. **If Apple charged you:** Go to reportaproblem.apple.com, find the charge, and select "I didn't mean to subscribe" or "I want to request a refund." Apple reviews each case, but post-cancellation charges are generally approved. **If Google charged you:** Go to payments.google.com, find the transaction, and request a refund. You can also contact the app developer directly through the Play Store listing. There are documented cases of users being charged many times after attempting to cancel — one reported 53 unauthorized charges. If you're seeing multiple charges, document every one with screenshots before contacting support. If the service, Apple, or Google refuses to refund a charge you believe is unauthorized, your bank is the final option. This is called a chargeback, and it's a legitimate consumer protection built into credit and debit card networks. Call your bank or use their app to report the charge as unauthorized. Explain that you canceled the subscription and were billed anyway, and provide your cancellation confirmation as evidence. Banks generally side with customers in clear-cut cases — a charge dated after a documented cancellation is usually straightforward to reverse. One practical note: most banks require you to try resolving it with the merchant first, which is why going through the service's support and then Apple or Google comes before this step. Keep records of those attempts too. ## What to Watch Out For **Cancellation confirmation vs. immediate access loss.** When you cancel a subscription, you usually keep access until the end of the billing period you already paid for. That's normal and expected. What you're stopping is the *next* renewal charge — read the cancellation confirmation carefully to understand what it stops and when. **Pausing vs. canceling.** Some AI services offer a "pause" option. Pausing delays the next billing cycle; it doesn't stop future charges entirely. If you want to stop being billed, cancel — don't pause. **Free trials that convert automatically.** If you signed up for a free trial and forgot to cancel, the first charge isn't a post-cancellation issue — it's a trial-to-paid conversion. Contact the service within 14 days — OpenAI treats forgotten trial conversions as accidental charges eligible for a refund within that window (documented in their help center). Google may also refund within 14 days at its discretion, but it is not guaranteed. Anthropic's policy is generally non-refundable outside the EU, EEA, and UK. Customers in those regions have a statutory 14-day withdrawal right at all three major services regardless of company policy. **Different email addresses.** If you use a different email to sign into the service's website than you used when you first subscribed through an app, you may be checking the wrong account. Try both email addresses if you can't find the subscription. ## What to Try Next If you're unsure whether ChatGPT Plus is worth paying for in the first place, [Is ChatGPT Plus Worth It?](/guides/chatgpt-plus-worth-it) lays out who genuinely benefits from the paid plan and who's fine on the free tier. And if you're looking at other AI apps and want to check whether they're legitimate before handing over payment details, [Is That App Actually AI, or Just a Scam?](/guides/is-ai-app-scam) covers the red flags to look for first. --- # Canva AI: Make Invitations, Cards, and Posts Without Design Skills Source: https://aiforregulars.com/guides/canva-ai-beginners You don't need to hire a graphic designer — or even know what a font is — to make a birthday invitation that looks like it came from a real designer. Canva is a free design tool that thousands of people use every day, and its AI features have made it even easier for complete beginners. This walkthrough will take you through making a birthday invitation from scratch. Once you know the steps, making cards, social media posts, or anything else follows the exact same process. Open your browser and go to **canva.com**. Click **Sign up** and create a free account with your email address, Google account, or Facebook account. No credit card required. Once you're in, you'll see the Canva home screen with a search bar at the top and lots of template thumbnails below it. On a phone, download the free **Canva** app from the App Store or Google Play and sign in the same way. In the search bar at the top, type what you want to make. For a birthday invitation, type: *birthday invitation* You'll see hundreds of templates — colorful, elegant, fun, or simple. Scroll through and click on one that matches the vibe you want. Don't worry about the text inside it; you're going to change everything anyway. Click **Customize this template** to open it in the editor. Inside the editor, click on any text element to select it. The text becomes highlighted and editable — just like clicking into a word document. Change the placeholder text to the real details: - The person's name - The date and time - The location or address - Any RSVP information Click somewhere else on the canvas when you're done to deselect the text box. Repeat for each text element you want to change. The first time I used Canva, I spent 20 minutes trying to make the design look "perfect" by tweaking every little thing. My advice: change only the text and maybe one color. The templates are already well-designed. The more you leave alone, the better it usually looks. This step is optional, but it's easy if you want to match a color scheme. To change a color: click the element you want to recolor, then click the colored square that appears in the toolbar at the top. A color picker opens — you can pick from preset colors or type in a specific hex code if you know it (like `#FF5733` for orange). To change a font: click a text element, highlight the text, then click the font name shown in the top toolbar. A dropdown menu appears with font options you can scroll through and preview. If any of this feels overwhelming, skip it entirely. The default template colors and fonts are already chosen to look good together. Most templates include a placeholder photo or illustration. You have two options: **Use your own photo:** Click the placeholder image in the template, then click **Replace** in the toolbar. Select **Upload** and choose a photo from your device. The AI will automatically crop and fit it to the space. **Generate an AI image:** In the left sidebar, click **Apps**, then find **Dream Lab** — Canva's main AI image generator. (A secondary option called **Text to Image** also appears in Apps if you want preset style filters.) Type a description of what you want — for example, *"colorful birthday balloons and confetti, flat illustration style"* — and click Generate. Pick the result you like and drag it onto your canvas. When your invitation looks the way you want it, click the **Share** button in the top right corner. Choose **Download**, then pick a format: - **PNG** — best for sending digitally (email, WhatsApp, text message) - **PDF** — best for printing at home or at a print shop Click **Download** and the file saves to your device. You can now share it however works best for your group — text it directly, email it, post it in a group chat, or print copies to hand out. For physical invitations, most print shops can print from a PDF on standard cardstock. ## Other things you can make with the same steps Once you know this process, you can make almost anything in Canva: - Holiday cards and thank-you notes - Flyers for a garage sale or neighborhood event - Social media posts for a birthday announcement - Simple resumes or one-page documents - Memorial cards or tributes The search bar is your starting point every time — just type what you want to make and browse the templates. ## What to try next If you want to add a custom AI-generated image to your Canva design, [creating AI images for free](/guides/create-ai-images-free) explains the process in more detail. And if you want to share a short video alongside your invitation — like a fun AI-generated clip — check out [how to make an AI video for free](/guides/make-ai-video-free). --- # ChatGPT on Your Phone: Setup Guide (iPhone & Android) Source: https://aiforregulars.com/guides/chatgpt-app-setup Having ChatGPT on your phone means you can ask it questions, get help with writing, or look something up wherever you are — without sitting down at a computer. The setup takes less than 10 minutes from start to first conversation. This guide covers both iPhone and Android. The steps are nearly identical on both devices. Before you search for ChatGPT in the app store, know this: there are dozens of apps with names like "AI Chat," "ChatGPT Assistant," "AI Chatbot for GPT-4," and similar. Many of these are not made by OpenAI and may show you ads, charge hidden subscription fees, or collect your data in ways you would not expect. **The real app:** Made by **OpenAI**. The developer name in the store will say "OpenAI." The icon is a simple white shape on a dark background. Only install the one from OpenAI. **On iPhone:** Open the **App Store** (the blue icon with an "A" on your home screen). Tap the Search tab at the bottom. Type **ChatGPT** in the search bar. Look for the result that says "ChatGPT" and shows **OpenAI** as the developer. Tap **Get** to download it. **On Android:** Open the **Google Play Store** (the multicolored triangle icon). Tap the search bar at the top. Type **ChatGPT**. Find the result made by **OpenAI** and tap **Install**. The app is free to download on both platforms. Once the download finishes, tap the app to open it. You will see a welcome screen with options to **Log in** or **Sign up**. If you already have a ChatGPT account from the website, tap Log in. Otherwise, tap **Sign up**. To create an account, you can use your email address or sign in with an existing Google, Apple (iPhone only), or Microsoft account. The single-tap options are faster — just tap one and follow the prompts to connect your account. If you sign up with email, you will receive a confirmation email from OpenAI — open it and click the verification link to activate your account. After creating your account, the app may show you a few quick setup screens: - A prompt to enter your first name (optional but useful — ChatGPT can address you by name) - A screen offering you a free trial of ChatGPT Plus (the paid plan) - A short introduction to what ChatGPT can do You can skip the paid plan offer — tap **Start for free** or the equivalent button. The free version is everything you need to get started. When I first set up the app, I tapped "Continue with Google" and was done in about 45 seconds. I did not have to create a new password or verify an email. If you use Gmail, that is the fastest route. You will land on a clean screen with a text box at the bottom that says something like "Message" or "Ask anything." Tap on it — your phone's keyboard will appear. Type a question or request and tap the send button (usually an arrow or a paper plane icon to the right of the text box). A few good first messages to try: - *"Suggest three easy dinner ideas for tonight."* - *"Explain what a deductible is in simple terms."* - *"Help me write a short thank-you text to my neighbor."* ChatGPT will respond within a few seconds. The app is simple, but here are the parts worth knowing: - **New chat button:** Look for a compose or pencil icon near the top of the screen — its exact location varies by app version. Tap it to start a fresh conversation on a new topic. - **Chat history:** Swipe from the left edge of the screen, or tap the menu icon, to see your past conversations. You can pick up any previous chat where you left off. - **Voice input:** Most phones let you tap a microphone icon on the keyboard to dictate your message instead of typing. - **Voice mode:** A separate voice feature that lets you have a spoken back-and-forth conversation with ChatGPT. See our [ChatGPT Voice Mode guide](/guides/chatgpt-voice-mode) to set that up. Before you use the app regularly, it is worth a quick look at your privacy options. Tap your profile icon or initials, then go to **Settings**. Look for **Data controls** or a similar section. There you can turn off the option that allows OpenAI to use your conversations to train its AI — if that matters to you. For a full walkthrough of your options, see our [ChatGPT privacy settings guide](/guides/chatgpt-privacy-settings). ## What to Try Next You are set up and ready to go. Try [25 ready-to-use prompts](/guides/things-to-ask-chatgpt) if you want ideas for what to ask right now. And if you want to talk to ChatGPT out loud instead of typing — great for when your hands are busy — the [ChatGPT Voice Mode guide](/guides/chatgpt-voice-mode) shows you how to use that feature on your phone. --- # How to Write a Cover Letter with ChatGPT (That Doesn't Sound Like AI) Source: https://aiforregulars.com/guides/chatgpt-cover-letter A blank page is the hardest part of any cover letter. You know what you want to say — you just cannot find the words. ChatGPT gets you past that blank page in about 60 seconds. What you do with the draft after that is what makes it yours. ## Why most AI cover letters fall flat The problem is not that AI writes badly. It is that AI writes generically. Without specific details from you, it produces something like: "I am excited to apply for this position. My experience in [field] makes me an excellent candidate." Every hiring manager has read that sentence a hundred times. It says nothing about you specifically. The fix is to give AI better input — and then make three targeted edits to the result. Do not open ChatGPT yet. Collect these four things first — having them ready makes the difference between a generic draft and one you can actually use. 1. The full job posting (copy all the text) 2. Your resume, or a quick summary of your relevant experience in your own words 3. The company's name and one specific thing you know about them — from their website, a recent news story, or a friend who works there 4. One real accomplishment that is relevant to this role — a number, a result, a moment you are proud of Paste this prompt into ChatGPT, filling in the brackets with your actual details: ``` Write a professional cover letter for the following job. Keep it to three short paragraphs. Tone: confident and warm, not stiff or overly formal. Do NOT start with "I am excited to apply." About me: [paste 3–4 sentences about your experience and one key accomplishment] Company name: [company name] One thing I like about this company: [one specific detail] Job description: [paste the full job description] ``` Read the result before you do anything else. It will almost always be usable — just not quite finished. That is exactly what the next three steps are for. AI almost always writes a generic opener, even when you tell it not to. Replace the first sentence with something that sounds like you noticed this particular job at this particular company. Instead of: "I am writing to express my interest in the Marketing Coordinator position at Acme Corp." Try something like: "When I saw Acme Corp was hiring a Marketing Coordinator, I stopped scrolling — I have been following your product launches for two years." It does not have to be dramatic. It just needs to be real and specific enough that it could not have been written for any other company. Find a sentence where the AI described your skills in the abstract. Replace it with a one- or two-sentence story. Instead of: "I have strong experience managing client relationships and resolving conflicts." Try: "Last year, a long-term client almost left over a billing issue. I called them directly, worked out a payment plan, and they renewed their contract the following month." A specific story is memorable in a way that a list of skills is not. The story does not need to be impressive — it just needs to be true and relevant. I have helped a few friends write cover letters with ChatGPT. The ones that got callbacks always had one thing in common: a moment that felt real. It did not need to be a big achievement — just something specific. A difficult situation handled well, a project rescued at the last minute, a decision that turned out to be right. AI cannot invent those moments. You have to add them yourself, and when you do, the whole letter changes. Read the full letter out loud. Notice any sentence that sounds stiff, wordy, or unlike how you actually talk. Rewrite those sentences in your own words. Common AI phrases worth replacing: - "I am eager to leverage my skills" → "I am excited to use what I have learned" - "I would be a valuable asset to your team" → "I think I could contribute a lot" - "Please do not hesitate to contact me" → "Feel free to reach out any time" This takes about five minutes and makes a real difference in how the letter reads. Before you send: - Is the company name spelled correctly? (AI can get this wrong, especially for smaller companies) - Did you address it to a person if you know their name? - Is it actually three to four paragraphs — not six? - Does it end with a clear, confident closing rather than an apology for your qualifications? Save it as a PDF and submit. ## Customize for every job The same base prompt works for every application. Just change the job title, company name, accomplishment, and the one specific detail about the company. You can customize a cover letter in about 10 minutes this way — versus the hour it might take to write one from scratch on a bad day. **What to try next:** Now that your cover letter is done, practice what happens next. Learn how to [use AI to practice job interviews](/guides/ai-interview-practice) so you walk into the room prepared, or go back and make sure your [resume is ATS-friendly](/guides/ats-friendly-resume-with-ai) before you hit submit. --- # ChatGPT Deep Research for Everyday Decisions: Appliances, Insurance, and Big Purchases Source: https://aiforregulars.com/guides/chatgpt-deep-research-everyday You're about to spend real money on something that will live in your home for a decade, or sign an insurance policy you'll carry for a year. The research feels overwhelming — dozens of product reviews, comparison sites that may be paid placements, forum threads from people whose situation is nothing like yours. ChatGPT's Deep Research feature can do a multi-step web investigation in the time it takes to make coffee, and hand you back a cited report you can actually use. ## What Deep Research Actually Does Deep Research is a feature inside ChatGPT that runs an extended, multi-step web investigation on your behalf. You give it a question, and it visits dozens of web pages, reads them, and synthesizes what it found into a report with clickable citations. It is not the same as regular ChatGPT, which draws from training data: Deep Research looks things up in real time. ChatGPT also has a separate **Shopping Research** mode, which works differently. Instead of a written report, it asks you follow-up questions — budget, preferences, dealbreakers — and returns product cards with the key specs compared side by side. More on that in Step 6. Before you type anything, decide which ChatGPT mode fits your goal: - **Deep Research** — best for open-ended questions that need synthesized analysis: "What should I know before buying a front-load washing machine under $900?" or "What should I look for when comparing home insurance quotes?" - **Shopping Research** — best for specific product comparisons: typing "Help me find a washing machine" will prompt it to ask about your budget and setup before showing product cards. - **Regular ChatGPT** — fastest, but uses training data rather than live results. Avoid it for anything price-sensitive or time-sensitive. Deep Research is available on all ChatGPT plans — free users get 5 lightweight queries per month, with higher quotas on paid plans. A good Deep Research request is specific enough to narrow the scope but open enough to let it surface things you didn't know to ask about. Include: - **What you're deciding** — name the decision, not just the product or topic - **Your constraints** — budget range, location if it matters, timeline - **Your dealbreakers** — features you won't compromise on, concerns you want addressed - **What you already know** — so it doesn't repeat information you have For a washing machine purchase: > I'm shopping for a front-load washing machine for a family of four in the US. Budget is $700–$950 for the machine only. I prioritize reliability and ease of repair over features — I don't need smart home integration. We do a lot of towels and bedding. What brands and models have the best reliability track record in this price range, what are the most common failure points I should ask about, and what should I look for in the warranty? For comparing insurance quotes: > I'm comparing home insurance quotes in [your state]. What are the most important exclusions and coverage gaps to look for in a standard policy, and what questions should I ask each insurer before I choose? For vetting a contractor: > I'm getting quotes to replace an asphalt shingle roof on a 1,800 sq ft house in the midwest US. What is a typical price range for this job, what should a legitimate contractor's quote include, and what are the warning signs of a contractor I should avoid? Since a February 2026 update, ChatGPT shows you the research plan it intends to follow before it starts browsing — a list of the sub-questions and angles it plans to investigate. You can read through it and edit it before clicking Start. This step is worth taking. Check whether: - The plan is covering the right angle. If you asked about reliability but the plan is focused on features and specs, redirect it. - There are sub-questions that don't apply to your situation — remove them to keep the report focused. - You forgot a constraint in your original request — add it here rather than starting over. Even one or two edits can meaningfully improve the output. If the plan looks right, click through and let it run. Deep Research takes longer than a regular ChatGPT response because it is actually visiting web pages one at a time. A typical report takes two to ten minutes. ChatGPT shows you a live progress view while it works, so you can see which pages it's reading. Keep the tab open and don't start a new conversation mid-way — you may lose the session. One of my favorite uses for Deep Research is the "is this price normal?" question — when a contractor gives me a quote and I want to know whether it's in the right ballpark before I negotiate or look elsewhere. I describe the job, the scope, and my region, and the report usually comes back with a realistic price range plus a note on what legitimate quotes at that level typically include. It doesn't replace getting multiple quotes, but it tells me fast whether a number is reasonable or wildly inflated. Deep Research returns a multi-section written report with numbered citations you can click. Here's how to read it well: **Skim first, then read carefully.** Read the section headings and the summary first. If the report has drifted from your question — answering something adjacent but not quite what you asked — you'll catch it before you spend time in the details. **Click through to two or three sources.** On anything you're going to act on — a price range, a safety claim, a warranty condition — open the linked page and check what it actually says. Deep Research occasionally misreads or misquotes sources. Spot-checking takes three minutes and catches the most consequential errors. **Notice the hedge language.** Phrases like "some sources suggest," "this may vary," or "as of early 2025" are the report flagging its own uncertainty. They tell you where to dig deeper before making a decision. **Check source diversity.** If every citation comes from the same type of source — all retailer sites, all forum posts — the report may be missing an important perspective. A good washing machine research report should draw from reliability surveys, consumer review sources, and manufacturer documentation, not just one of those. If your question is specifically about choosing between products — washing machines, laptops, air purifiers, baby monitors — ChatGPT's **Shopping Research** mode is often faster and more visual than a full Deep Research report. To start it, just ask ChatGPT to help you shop: > Help me find a front-load washing machine for a family of four, budget around $800. ChatGPT will ask a few follow-up questions about your preferences before returning product cards with specs compared side by side. ChatGPT's shopping use has grown from 2% to 30% of users over two years, and Shopping Research is built for exactly this kind of product search. You can use both tools together: Deep Research first to understand what features matter and which brands have strong reliability records, then Shopping Research to find specific models that match those criteria. ## What to Watch Out For **Your monthly quota is finite.** Deep Research reports count against a monthly limit that varies by plan. Save them for decisions that genuinely need them — replacing an appliance, choosing an insurance policy, vetting a contractor — rather than curiosity questions you could answer with a regular search. **Citations still need spot-checking.** The report tells you where it found its information, which is more than regular ChatGPT ever provides — but the tool occasionally misrepresents what a source says. Never act on a price, a safety claim, or a legal fact from a Deep Research report without clicking through to verify it yourself. **Health questions have real limits.** Deep Research is useful for understanding what a procedure typically costs or what questions to ask a specialist. But it can't access paywalled medical journals, so peer-reviewed clinical evidence may be underrepresented. It is not a substitute for a diagnosis or professional medical advice. **Reports go stale.** Prices, product availability, and insurance terms change faster than a cited report. Treat it as a well-researched starting point, not a final answer. **Be thoughtful about personal details.** If you're researching a medical condition, a financial situation, or a legal matter, avoid including identifying personal information in your prompts — your inputs are sent to and processed by OpenAI. ## What to Try Next If you want to see how another AI research tool compares, [How to Use Perplexity AI](/guides/how-to-use-perplexity) covers a strong alternative with a different interface that's worth knowing. And if you're wondering how much to trust ChatGPT's answers in general — not just in research mode — [Can You Trust ChatGPT?](/guides/can-you-trust-chatgpt) gives you the honest picture. --- # ChatGPT for Seniors: Your First Hour, No Jargon Source: https://aiforregulars.com/guides/chatgpt-for-seniors-first-hour ChatGPT is a free website where you can have a real conversation with an AI — typing a question and getting a thoughtful, plain-English reply. Millions of people use it every day to write notes, get explanations, look things up, and brainstorm ideas. This tutorial takes you through your very first session, step by step, with no assumptions about what you already know. On your computer, phone, or tablet, open the internet browser you normally use — Chrome, Safari, Firefox, and Edge all work fine. In the address bar at the top of the screen, type **chatgpt.com** and press Enter or tap Go. You'll land on a page with a large text box near the bottom. That box is where you'll type your questions. You can try a few questions without an account, but creating one lets you save your conversations and use ChatGPT without interruptions. Click **Sign up** and enter your email address, then choose a password. You can also sign in with an existing Google, Microsoft, or Apple account if you have one — this skips the need to create a new password. ChatGPT will ask for your birthdate (for age verification). You may be asked a short question or two about how you plan to use it. Just answer honestly. Before you type your first question, make everything on the screen larger. This is a built-in browser trick that works on any website: - **Windows computer:** Hold the **Ctrl** key and press the **+** key. Repeat until the text feels comfortable. Press **Ctrl** and **–** to go back if you overshoot. - **Mac computer:** Hold the **Command** key and press the **+** key. - **Phone or tablet:** Pinch two fingers apart on the screen to zoom in, just like zooming into a photo. Do this now, before you start typing, so everything is easy to read throughout your session. Click inside the text box at the bottom of the screen and type a question. Anything you've been curious about works. Here are a few safe and easy ones to start with: ``` What's a good way to start walking more at age 70? ``` ``` Explain what high blood pressure means in simple words. ``` ``` Write a short thank-you note to my neighbor who brought me soup. ``` Press **Enter** or click the arrow button. ChatGPT will type out its answer on the screen, usually within a few seconds. You can then ask a follow-up question — just like a conversation. If an answer is confusing, type "Can you explain that more simply?" and it will try again. The first question I ever typed into ChatGPT was about whether a food I'd been reading about was actually as healthy as people claimed. The answer was calm and clear, and it suggested I bring up a specific point with my doctor. That felt exactly right — not replacing the doctor, just helping me show up to the appointment with a better question. If typing is difficult or tiring, ChatGPT has a voice mode — you speak your question out loud and it responds. Look for a waveform or headphone icon on the right side of the text box and tap or click it. Speak your question naturally, and ChatGPT will answer out loud in a spoken voice — the full conversation is audio in both directions. Voice mode is available on free accounts as well as paid ones, though paid plans get access to the higher-quality voice model. If you are using ChatGPT on a phone, you can also tap the microphone icon on your phone's keyboard at any time to dictate your message, even without using ChatGPT's built-in voice mode. Some topics are ideal for a first session — low stakes, interesting, and a good way to see what ChatGPT can do: - **Recipes:** "Give me a simple chicken soup recipe for two people." - **Word explanations:** "What does 'atrial fibrillation' mean in plain English?" - **Writing help:** "Help me write a birthday message for my grandson who just turned 16." - **History questions:** "Tell me about the history of Route 66 in two short paragraphs." - **Travel ideas:** "What are some good places to visit in Arizona in the winter?" These let you see how the tool responds without worrying about the stakes. ChatGPT is helpful, but there are a few things you should never type into it — or any website that wasn't specifically set up to receive them: - Social Security number - Bank account or credit card numbers - Passwords or PINs - Your full home address ChatGPT helps with words, ideas, and questions. It has no reason to ask for financial or identity information. If any tool claiming to be AI asks for those details, leave immediately. **What to try next:** Now that you've had your first session, you might enjoy exploring more of what AI can help with day to day. Our guide [AI for Seniors: Five Easy Entry Points](/guides/ai-for-seniors) covers voice assistants, photo tools, and scam protection in one place. When you're ready to make your account more private, [ChatGPT privacy settings explained](/guides/chatgpt-privacy-settings) walks through the main options in plain English. --- # ChatGPT Can Now See Your Bank Account — What It Does and Whether It's Safe Source: https://aiforregulars.com/guides/chatgpt-link-bank-account Until recently, getting ChatGPT to help with your finances meant doing the data entry yourself — copying a list of transactions, typing in your balance, or describing your spending from memory. That changed on May 15, 2026, when OpenAI announced a direct bank connection feature (covered by TechCrunch and OpenAI's own blog), which rolled out to Plus and Pro subscribers in the United States in late June 2026 and, from October 2, 2026, to free users there too. It sounds convenient. It also sounds like exactly the kind of thing you should understand before you try it. This guide explains what it actually shares, what ChatGPT can and cannot do with that information, and how to step back out if you decide it is not for you. ## What the Bank Connection Actually Does ChatGPT connects to your bank through Plaid, a financial data service that already powers thousands of familiar apps — Venmo, Robinhood, Credit Karma, and many others. Plaid works as a secure middleman: it handles the login to your bank, then passes a limited, read-only data feed to ChatGPT. **Read-only means what it says.** ChatGPT can see your account balances and recent transactions. It cannot see your full account number or routing number. It cannot move money, schedule payments, or make any changes to your account whatsoever. **Which banks are supported?** OpenAI says the connection works with over 12,000 financial institutions through Plaid. That covers most major US banks, credit unions, and credit card issuers. **Financial memories.** If you share a financial goal during a conversation — "I'm saving for a car" or "I want to pay off my credit card by December" — ChatGPT can remember that goal and refer back to it in later conversations. This uses ChatGPT's existing memory feature, which you can review and delete separately in Settings. ## Who Should Think Twice Before Connecting This feature is not for everyone, and that is completely fine. Consider skipping it if: - You share a ChatGPT account with a family member or partner and do not want your spending data visible to them. - You are not comfortable with any third-party service having read access to your bank data, regardless of what it is used for. - You are dealing with a financial situation — debt proceedings, legal matters, divorce — where you want to be especially careful about data trails. - Your bank has had security issues or does not fully support Plaid. If any of those apply, the no-account alternative at the end of this guide gets you the same analytical help without the connection. ## How to Connect Your Bank Account The feature is available to ChatGPT users in the US on any plan, including free, on the web at chat.openai.com and on the iOS and Android apps. On the web, click your profile icon in the bottom-left corner and select **Settings**. On iOS or Android, tap your profile icon and choose **Settings**. Navigate to **Apps** in Settings, then select **Finances**. The Finances page will already be open (from step 1). Tap or click **Get started**, then select **Connect with Plaid** to open Plaid's secure authentication window. A Plaid window will open — this screen is run by Plaid, not OpenAI. Type the name of your bank in the search box. Plaid lists over 12,000 institutions, so most US banks should appear. Select yours from the results. Plaid will prompt you to enter your bank's username and password. This login goes directly to Plaid and your bank — OpenAI does not receive or store your bank login credentials. Your bank may then ask you to confirm your identity with a text code or app notification. Plaid shows you a summary of what it will share with ChatGPT: typically your account balances and recent transaction history. Read it carefully. If you are connecting more than one account, you can usually choose which ones to include. Once connected, try these prompts to get started: > "What did I spend the most money on last month?" > "Do I have any recurring charges I might have forgotten about?" > "Based on my recent spending, can I afford a $400 expense this month?" > "I'm saving for a car — how does my current spending look compared to my goal?" The most useful thing I have found is asking ChatGPT to spot patterns I would not notice on my own — things like "are there any subscriptions that renewed without me realizing?" The answer is almost always yes, and finding even one forgotten subscription usually covers the cost of a Plus membership for the month. ## What ChatGPT Can Do Once Connected Having your bank data available opens up conversations that feel different from manual budgeting. Instead of typing your numbers in, you can ask: **Spending summaries by category.** ChatGPT can break down what you spent on groceries, dining, transport, or subscriptions in a given month, using the transaction categories your bank already assigns. **"Can I afford this?" questions.** Ask "I want to spend $200 on something — do I have room in my budget this month?" and get an answer based on your actual current balance and recent spending, not a rough estimate. **Subscription spotting.** Many people have recurring charges they have lost track of — streaming services, gym memberships, software trials that converted to paid plans. ChatGPT can scan your transactions and list what is charging you regularly. **Progress toward goals.** Tell ChatGPT a saving target and a deadline, and it can look at your recent income and spending patterns to tell you whether you are on track. **What it cannot do.** ChatGPT cannot file your taxes, make transfers, or pay bills through this feature. It does connect to investment and retirement accounts — including brokerage accounts, Traditional and Roth IRAs, 401(k) and 403(b) plans, HSAs, and 529 plans — through the Plaid integration. ## What to Watch Out For **Your conversations are not private from OpenAI.** When you discuss finances in a ChatGPT conversation, that conversation is processed and stored by OpenAI the same as any other chat. You can turn off chat history in Settings, which stops conversations from being used to train future models — but OpenAI still processes them to run the service. **Plaid has its own privacy policy.** The connection flows through Plaid, a separate company with its own data practices. Plaid may share anonymized transaction data with third parties for research or analysis. Worth reading before you connect if you want the full picture. **Financial memories need a separate step to delete.** If ChatGPT stores a financial goal in its memory, that memory persists even after you disconnect your bank. Go to Settings → Personalization → Memory to review and delete specific entries. **This is not financial advice.** ChatGPT can describe your spending patterns and help you think through decisions, but it does not know your complete financial picture — debts, investments, tax obligations — unless you tell it, and it can make mistakes. For decisions that really matter, a human financial advisor or certified planner is the right resource, not a chatbot. **Account sharing is a real privacy risk.** If another person can access your ChatGPT account, they can see your connected bank data and the conversation history about your finances. Be aware of this in any shared household or family account setup. ## How to Disconnect and Delete Your Data Disconnecting is straightforward, and OpenAI says your financial data is deleted within 30 days. 1. Go to **Settings** in ChatGPT (same path as when you connected: profile icon → Settings). 2. Find **Apps** → **Finances**. 3. Select your bank connection and click **Remove**. 4. You can also revoke the connection at the Plaid level by visiting my.plaid.com, signing in, finding ChatGPT/OpenAI in your connected apps, and clicking **Disconnect** there. 5. To clear any financial memories, go to **Settings → Personalization → Memory** and delete any relevant entries. ## The No-Account Alternative If you want budgeting help without the live connection, there is a practical workaround. Download a recent statement from your bank (available as a PDF or CSV from most bank websites), remove or cover your name, address, and full account number, and paste the transaction list directly into a ChatGPT conversation. You get the same spending summaries and budget analysis — you just paste a fresh statement each time you want up-to-date numbers. For people who only want a monthly budget check, this approach works well, and it means your live account data never touches OpenAI's servers at all. ## What to Try Next If you want a step-by-step guide to monthly budgeting with ChatGPT — including the paste-your-statement method and ready-to-use prompts — [How to Build a Monthly Budget with ChatGPT](/guides/chatgpt-monthly-budget) covers all of that. And if you are deciding whether a dedicated budgeting app might suit your needs better than ChatGPT, [Best AI Budgeting Apps Compared](/guides/best-ai-budgeting-apps) puts the main options side by side. --- # How ChatGPT Memory Works — and How to Control It Source: https://aiforregulars.com/guides/chatgpt-memory-settings ChatGPT can now carry information from one conversation to the next. If you mention that you're vegetarian, it will remember that the next time you ask for recipe ideas. If you say you're job hunting, it will factor that in when you ask for writing help. That's useful — but it raises a fair question: what exactly does it know about you, and how do you control it? Here's the full picture. Go to chatgpt.com and sign in to your account. In the bottom left corner of the screen, click your name or profile icon. A small menu appears — select Settings from it. A settings window will open. Look for a section called Personalization or Memory in the left-hand menu and click it. Inside the Memory section, you'll see a button or link labeled something like Manage memories or View memories. Click it. A list appears showing every fact ChatGPT has saved about you. Each entry is a short phrase — things like "prefers vegetarian recipes," "works in accounting," or "has two kids in elementary school." Some of these you mentioned intentionally; others the AI picked up from things you said in passing. Take a few minutes to scroll through the whole list. Scan through the list. For any item you want to remove, click the small trash icon or X button next to it. It's deleted immediately — ChatGPT will no longer use that fact in future conversations. You can also delete all memories at once with a Clear all button if you want to start completely fresh. If you want one private conversation that won't add to your memory or appear in your history, look for a Temporary Chat option. It's usually a toggle near the top of the chat window or accessible from the new chat menu. In a Temporary Chat, nothing gets saved to your memory and the conversation won't appear in your sidebar after you close it. It's a good choice for sensitive personal topics or anything you simply don't want remembered. If you'd rather ChatGPT not remember anything at all, go back to Settings → Personalization → Memory and toggle Memory off. When it's off, ChatGPT won't save any new facts from your conversations, and it won't use any previously saved ones. Your existing memories are not deleted when you toggle it off — they're just paused. If you turn Memory back on later, those saved facts become active again. You can also tell ChatGPT what to remember. Just say something like: "Remember that I'm vegetarian" or "Remember that I prefer short, direct answers." ChatGPT will confirm that it saved the fact. This is useful for making sure the AI knows things about you that it hasn't picked up yet — rather than waiting for it to notice on its own. I went through my memory list after a few months of using ChatGPT and found some things stored that surprised me — some accurate, one or two that were just wrong guesses from things I'd said in passing. Cleaning it up took about two minutes. I'd suggest checking yours every month or two, especially if you use ChatGPT for anything personal. ## What to try next If you want to review your broader ChatGPT privacy settings — like what gets stored in your account and how to manage your data — [ChatGPT Privacy Settings](/guides/chatgpt-privacy-settings) covers that in full. And if you're new to ChatGPT altogether, [How to Use ChatGPT](/guides/how-to-use-chatgpt) is the right place to start. --- # How to Build a Monthly Budget with ChatGPT Source: https://aiforregulars.com/guides/chatgpt-monthly-budget Most people do not have a budget — not because it is hard, but because sitting down to build one feels overwhelming. ChatGPT makes it into a conversation instead of a chore. You do not need a spreadsheet, a finance app, or any special knowledge. You just need to know roughly what comes in each month and roughly what goes out. ChatGPT handles the rest. **One important caution before you start:** share spending categories and totals with ChatGPT, not your full bank statement. Never type account numbers, card numbers, or login credentials into any chat tool. Before opening ChatGPT, take two minutes to note your take-home pay — the amount that actually lands in your bank account after taxes and deductions. If your income varies month to month, use an average of the last three months. If you have more than one income source (a second job, freelance work, Social Security, rental income), list each one separately. You will paste these into ChatGPT in the next step. Jot down the main buckets your money goes into each month. You do not need exact numbers yet — rough estimates are fine. Common categories include: - Housing (rent or mortgage) - Utilities (electric, water, internet, phone) - Groceries - Eating out and takeout - Transportation (gas, car payment, insurance, transit) - Subscriptions (streaming, gym, apps) - Medical and pharmacy - Kids or pets - Entertainment and hobbies - Clothing and personal care - Savings or debt payments Do not stress about perfection. A reasonable guess is better than nothing and ChatGPT will help you fill in any gaps you missed. Open ChatGPT and paste a message like this, filling in your own figures: ``` I want to build a monthly budget. Here is my situation: Monthly take-home income: $[AMOUNT] Approximate monthly spending: - Rent: $[AMOUNT] - Utilities: $[AMOUNT] - Groceries: $[AMOUNT] - Eating out: $[AMOUNT] - Transportation: $[AMOUNT] - Subscriptions: $[AMOUNT] - Medical: $[AMOUNT] - Kids/pets: $[AMOUNT] - Entertainment: $[AMOUNT] - Other: $[AMOUNT] My goals: [e.g., save $300 a month, pay off a credit card, build a 3-month emergency fund] Please add these up, show me how much is left over (or how much I'm over), and tell me where I might be spending more than average for my situation. ``` ChatGPT will do the math, show the total, and flag anything that looks out of proportion. Once you see the numbers laid out, ask ChatGPT to turn them into a monthly budget you can actually follow: ``` Based on those numbers, build me a simple monthly budget. I want to [save $300 / pay down debt / have some fun money — your goal here]. Show it as a table with category, current spending, and suggested limit. ``` ChatGPT will suggest which categories to trim and by how much. The suggestions are a starting point, not commands — you decide what fits your life. Ask directly about the categories that surprise you: ``` My subscriptions add up to $[AMOUNT] a month. What are some ways people cut this down without giving up the things they actually use? ``` Or if eating out is higher than you expected: ``` I'm spending about $[AMOUNT] a month on takeout and restaurants. What is a realistic target for a family of [SIZE] and how do I get there gradually? ``` ChatGPT does not judge. You can be completely honest about your spending and it will respond with practical suggestions, not lectures. I keep my budget summary in a plain notes app on my phone — just the category names and monthly limits, nothing sensitive. That way I can check it in the grocery store parking lot before I go in. Having the numbers somewhere visible is what makes a budget actually work. If saving money is your goal, give ChatGPT a specific number to work with: ``` I want to save $[AMOUNT] over the next [NUMBER] months for [vacation / emergency fund / car repair]. Based on my budget above, is that realistic? If not, what would I need to change to get there? ``` ChatGPT will tell you honestly whether the goal is reachable and what adjustments — skipping a streaming service, cooking one more dinner at home per week — would close the gap. Copy the finished budget out of ChatGPT and paste it into a Google Doc, a notes app, or a simple spreadsheet. Add a column to track what you actually spent at the end of the month. ChatGPT cannot access your bank account or auto-update your numbers. That part is still on you — but checking in takes about ten minutes a month once the plan is built. ## What to try next With a budget in hand, planning big purchases gets easier. The [vacation planning guide](/guides/plan-vacation-with-chatgpt) shows how to use ChatGPT to estimate trip costs before you book anything. And the [AI meal planning guide](/guides/ai-meal-planning) can help bring your grocery and takeout spending down with a week of dinners planned in minutes. --- # ChatGPT Parental Controls and Safe Setup for Kids Source: https://aiforregulars.com/guides/chatgpt-parental-controls A lot of parents arrive at this question the same way: their child already knows what ChatGPT is, probably already uses it, and you are playing catch-up. That is fine. The goal here is not to shut everything down — it is to set up something that is reasonable, visible, and safe for your family. This guide walks through the current state of ChatGPT's age requirements and available controls, and what you can actually do to create a safer setup. ChatGPT requires users to be at least 13 years old. In the European Union and some other regions, the minimum age is higher — 16 in some countries — due to local data privacy rules. For users between 13 and 17, OpenAI requires parental or guardian consent. This is part of the account creation process — the user is asked to verify their age, and if they indicate they are under 18, a consent step is triggered. In practice, age gates are not foolproof. A child who knows to enter a false birth year can create an account without triggering any additional steps. This is why knowing your child's situation matters more than assuming the system catches everything. If your child is under 13, ChatGPT is not appropriate for them under OpenAI's terms. Look at the section at the end of this guide for alternatives designed for younger children. Before setting anything up, find out what already exists. Ask your child directly if they have a ChatGPT account — and approach this as information-gathering, not interrogation. You can also check: - Browser history (ChatGPT runs at chat.openai.com or chatgpt.com) - Saved passwords in the browser or password manager - The app store on their phone — look for the "ChatGPT" app by OpenAI If they have an account, ask them to log in while you are with them. Look at the account settings to see what email address is associated with it and what age was entered during signup. The most visible arrangement for families is a shared account — one that you manage together and that the child uses with you logged in. This lets you see the full conversation history and keep the account settings under your control. To do this: 1. Go to chatgpt.com and create (or log in to) an account using a parent email address. 2. Set a strong password that you control. 3. Use this account with your child rather than creating a separate one for them. The downside is that this account will also contain your own conversations if you use ChatGPT. OpenAI launched built-in parental controls in September 2025. As an alternative to the shared-account approach, you can link your own account with your teen's: go to your profile icon → Settings → Parental controls → Add family member, then invite your teen by email or phone number. Once linked, you can set quiet hours, restrict voice mode and image generation, and receive safety alerts if OpenAI's systems detect signs of serious distress. Note that linked parental controls do not give you access to read your teen's conversations; the shared-account method above remains the only way to review chat history directly. Once logged in, go to **Settings** (the icon in the lower left corner of the screen) and look at a few key areas: **Memory:** ChatGPT can remember details from previous conversations. For a child's account, you may want to turn this off so conversations do not build on each other over time. Go to Settings → Personalization → Memory and toggle it off. **Data controls:** Under Settings → Data Controls, you can turn off the option to use your conversations to train OpenAI's models, and you can download or delete your conversation history. **Custom instructions:** If custom instructions are set, review them. These are instructions the account always follows, and they could affect how the AI responds. There is currently no content filter toggle in standard ChatGPT settings that parents can set separately from what OpenAI provides by default. The default settings already block explicit content, but ChatGPT is not filtered to the same standard as platforms designed for children. ChatGPT account settings are limited. Device-level controls give you another layer. **Screen time and app limits (iPhone/iPad):** Go to Settings → Screen Time → App Limits or Content & Privacy Restrictions. You can limit access to specific websites or apps by time of day. **Google Family Link (Android):** Google's Family Link app lets you approve app downloads, set screen time limits, and review activity on your child's Android device. **Router-level controls:** Many home routers let you set browsing restrictions by device. This is the most reliable method because it applies regardless of which browser or app the child uses. **Browser extensions:** Extensions like Google's Family Link for Chrome, or parental control software like Bark or Circle, can restrict specific websites or flag certain types of content. No single method is perfect. The combination of a supervised account plus device-level time limits plus open communication covers most situations. Honestly, the most effective thing I have seen is parents using ChatGPT alongside their kids — not over their shoulder, but together, trying things out. When kids see a parent use a tool, ask questions about it, and talk about when it helps and when it misleads, they build their own judgment. That is harder to install than a content filter, but it lasts longer. Controls slow things down. Conversation changes behavior. The most useful thing to establish with your child is not a list of rules, but a shared understanding of what ChatGPT is and is not. It is a tool that generates plausible-sounding text — it can be very helpful and it can also be wrong, biased, or shallow. Knowing that, and knowing when to trust it and when to check elsewhere, is a skill worth building early. A few things worth covering: - ChatGPT can make things up. It sounds confident even when it is wrong. Important facts need to be checked in other sources. - Anything typed into ChatGPT goes to OpenAI's servers. Personal information — full name, home address, school name, photos — should stay out of it. - Using AI to complete assignments may be against school rules, depending on the assignment. When in doubt, ask the teacher. You do not have to cover all of this in one conversation. Bringing it up when it is relevant — when homework comes up, when they mention something they heard about AI — keeps it low-pressure. ## If Your Child Is Under 13 ChatGPT is not appropriate for children under 13 under OpenAI's terms. Some alternatives designed for younger learners include: - **Khanmigo** by Khan Academy — an AI tutor built for students with educational guardrails - **Socratic by Google** — designed for homework help with stronger content controls - **Magic School** — designed for classroom use with educator oversight These tools are built with younger audiences in mind and include content filtering that ChatGPT does not. ## What to Try Next Once the practical setup is in place, the bigger conversation is about how your child uses AI for learning — not just whether they can access it. [My Kid Uses ChatGPT for Homework — A Parent's Guide](/guides/kids-chatgpt-homework-parents-guide) has a calm framework for that conversation, including how to tell the difference between AI helping with learning and AI replacing it. For a look at privacy settings for your own ChatGPT account, [ChatGPT Privacy Settings](/guides/chatgpt-privacy-settings) covers the key options. --- # Free vs Paid ChatGPT: Is Plus Worth It? Source: https://aiforregulars.com/guides/chatgpt-plus-worth-it ChatGPT offers a free version and a paid plan called Plus. The gap between them has shifted several times since launch, so it's worth looking at what's actually different right now — and whether it matters for how you use it. ## What the free tier actually does well The free version of ChatGPT is not a stripped-down demo. It's a genuinely capable AI assistant that millions of people use every day. It handles: - Answering everyday questions in plain language - Writing and editing emails, messages, and short documents - Brainstorming ideas for gifts, meals, or creative projects - Explaining concepts in simple terms - Summarizing text you paste in - Basic coding help and formula writing for spreadsheets For casual or occasional use, free is enough. Starting there before paying is the right move. ## What Plus adds Plus gives you access to OpenAI's most capable models, which noticeably improve performance on complex tasks. The practical differences: - **Smarter responses on hard questions** — longer documents, multi-step reasoning, nuanced writing - **Image generation** — create images from text descriptions using ChatGPT Images - **Higher limits on images, files, and voice** — free text chats are unlimited since August 2026, but images, file uploads, and voice mode still have daily caps that Plus raises - **Better file and document analysis** — upload and discuss PDFs, spreadsheets, and other files - **Faster responses** during peak hours when the service is busy - **Advanced data analysis** — the AI can run code to analyze datasets, create charts, and crunch numbers ## Side-by-side comparison | Feature | Free | Plus | |---|---|---| | Core AI chat | Yes | Yes | | Most capable model | Limited access | Full access | | Image generation (ChatGPT Images) | Limited | Yes | | File / document upload | Limited | Higher limits | | Web search | Yes (limited) | Yes | | Text message limits | Unlimited (since Aug 2026) | Unlimited | | Image / voice message limits | Lower | Higher | | Response speed at peak times | May slow down | Priority access | | Custom GPTs | Use only | Create and use | | Memory across conversations | Yes | Yes | | Voice mode | Basic | More options | ## When Plus is worth it **Plus is likely worth it if you:** - Use ChatGPT every day for work — drafting, research, analysis - Regularly hit the free tier's limits on images, file uploads, or access to the most capable model - Need image generation for presentations, social media, or creative work - Work with long documents or data files that the free tier struggles with - Notice that free-tier responses aren't quite good enough for complex tasks **Stick with free if you:** - Use ChatGPT a few times a week for simple tasks - Mainly need help writing emails, quick explanations, or light brainstorming - Are still figuring out what AI can do for you - Want to try before committing to a monthly cost My honest take: I started on free, hit the message limit after a few days of heavy use, and upgraded. (OpenAI removed that free-tier text limit in August 2026, so that specific reason no longer applies — but the quality and feature gaps below still do.) The quality jump on complex writing tasks was real. But during lighter weeks I caught myself thinking "I'm not using this enough to justify the cost." If you work in front of a computer all day and you're already reaching for ChatGPT constantly, Plus pays for itself in time saved. If you're an occasional user, free genuinely covers it. ## A simple self-test Ask yourself these four questions before deciding: 1. Have I hit the free tier's limits on images, file uploads, or the most capable model in the last month? If yes, Plus is worth considering. (Free text chats are unlimited since August 2026, so this no longer applies to plain conversations.) 2. Do I regularly work with long documents I want to upload and discuss? Plus handles this better. 3. Do I need to generate images? Free access to image generation is limited. 4. Am I still figuring out if AI is useful for me? Start free, upgrade later when you know what you need. The safest move is to use the free tier seriously for two weeks. If you keep running into limits or wishing for better results, that's your answer. ## What to try next If you haven't set up ChatGPT yet, [How to Use ChatGPT](/guides/how-to-use-chatgpt) walks you through the basics on the free version. If you're comparing tools beyond ChatGPT, [AI Search vs Google](/guides/ai-search-vs-google) helps you figure out which AI tool fits which job. --- # Is My Data Safe with ChatGPT? Privacy Settings Walkthrough Source: https://aiforregulars.com/guides/chatgpt-privacy-settings ChatGPT is a genuinely useful tool, but like most free services, it handles your data in ways that are worth understanding. The good news is that OpenAI provides real controls — and changing your settings takes only a few minutes. This walkthrough covers what to look at, what to turn off, and what habits to build. When you use ChatGPT with an account, every conversation is saved to your history by default. OpenAI states that it uses this data to improve its models — meaning your conversations may be reviewed by people or processes and used to train future versions of ChatGPT. This does not mean your conversations are publicly visible to other users. But it does mean they exist on OpenAI's servers and are not purely private. Knowing this helps you decide what is and is not appropriate to share in the chat window. This is the most important setting to check. Open ChatGPT and click your profile picture or name in the corner of the screen. Select **Settings**, then look for a section called **Data Controls** or similar. You should see an option labeled something like **Improve the model for everyone** — turn this off. This tells OpenAI not to use your future conversations to train the AI. In the same section, you may see an option to disable **Chat History** entirely. If you turn this off, new conversations will not be saved after you close them. This is the most private setting available, though it means you cannot go back and review a previous conversation later. If you have been using ChatGPT for a while, months of conversation history may already be stored — including anything personal you asked for help with. In the left sidebar, you can scroll through past conversations and delete individual ones by clicking on the three-dot menu next to each one. To clear everything at once, look for a **Delete all chats** option inside **Settings > Data Controls**. Deleting removes conversations from your view and from active storage. OpenAI's policies note that some data may be retained temporarily for safety and abuse prevention purposes even after deletion. Before deleting anything, you can download a copy of everything ChatGPT has stored for your account. This is useful if you want to see exactly what is there, or if you want a personal archive before making changes. In **Settings > Data Controls**, look for an **Export data** option. You will receive an email with a download link — exports can take up to 7 days to arrive, and the download link expires 24 hours after you receive it. The file contains your conversations in a readable format, along with other account information. Beyond the in-app export, OpenAI also runs a separate Privacy Portal (privacy.openai.com) for formal requests to get a copy of your data or have it deleted — useful if the Settings export isn't working, or if you want to exercise a GDPR right in writing. I went through this process after realizing I had months of conversations stored — including times I had asked for help thinking through health decisions and financial questions. Most of what I had shared was not truly sensitive, but seeing it all in one place made me more intentional about what I type. I turned off training, deleted the old history, and now treat ChatGPT more like a search engine than a private journal: useful for information and writing help, but not where I process genuinely personal details. If you use a paid ChatGPT plan or if the Memory feature has been enabled on your account, ChatGPT can remember things about you between conversations — your preferences, your name, your situation, details you have mentioned in passing. This makes responses more useful but means personal information accumulates over time. To see what ChatGPT has remembered about you, go to **Settings > Personalization > Memory**, then select **Manage memories**. You can delete individual memories you do not want kept, or turn Memory off entirely. If you turn it off, ChatGPT will start each conversation fresh with no memory of previous ones. One thing to watch for: deleting an entry from the Memory summary does not always remove it completely. If the same information also shows up in a saved or archived conversation, or in an uploaded file, it stays there — according to OpenAI, fully removing it means deleting it from each of those places separately, not just from the Memory summary. No privacy setting fully replaces good habits. Think of ChatGPT as a knowledgeable assistant in a shared office space — helpful for a wide range of tasks, but not the right place for your most private information. Things that are generally fine to share: questions about general topics, writing you want feedback on, fictional scenarios and creative projects, general advice on common situations, and anything you would comfortably say in a public forum. Things worth keeping out: your full name combined with your address or date of birth, financial account numbers, medical records or diagnosis details, passwords or security answers, and the personal details of family members who have not chosen to be discussed. This is not about distrusting ChatGPT specifically. It is about using any digital tool in a way that is proportionate to the sensitivity of what you are sharing. If you or someone in your household uses ChatGPT through a workplace or school account (ChatGPT Team, Enterprise, or Edu), those plans do not use conversations to train models by default — the opt-out described above applies to personal Free, Plus, and Pro accounts. **What to try next:** If children in your household have access to ChatGPT, [ChatGPT Parental Controls](/guides/chatgpt-parental-controls) covers the family-specific settings worth reviewing. For a broader introduction to how ChatGPT works and what it is actually doing, [What Is ChatGPT?](/guides/what-is-chatgpt) is a good place to start. ## Sources - [Data Controls FAQ](https://help.openai.com/en/articles/7730893-data-controls-faq) — OpenAI Help Center - [Memory FAQ](https://help.openai.com/en/articles/8590148-memory-faq) — OpenAI Help Center - [Exporting your ChatGPT history and data](https://help.openai.com/en/articles/7260999-how-do-i-export-my-chatgpt-history-and-data) — OpenAI Help Center - [Art. 17 GDPR – Right to erasure ('right to be forgotten')](https://gdpr-info.eu/art-17-gdpr/) — GDPR-info.eu - [Making a complaint](https://ico.org.uk/make-a-complaint/) — UK Information Commissioner's Office --- # ChatGPT Projects: Keep Your Recipes, Job Search, and Hobbies in Separate Folders Source: https://aiforregulars.com/guides/chatgpt-projects-guide If you've ever started a ChatGPT conversation about your job search, then switched to asking for dinner ideas, and then had to explain yourself all over again — this is the fix. ChatGPT Projects let you create separate, named workspaces for different areas of your life. Each one keeps its own chats, its own uploaded files, and its own rules that ChatGPT follows automatically every time you open it. ## What Is a ChatGPT Project? Think of a ChatGPT Project as a dedicated notebook for one subject. Inside that notebook, you keep three things together: - **Conversations** — all your chats on that topic, in one searchable place - **Files** — documents you upload for reference (your CV, a recipe collection, an insurance policy) - **Custom instructions** — rules ChatGPT follows every time you open that project That last part is what makes Projects genuinely useful. Without them, ChatGPT forgets your preferences the moment you start a new chat. With a project, you write your instructions once — "Always suggest recipes using grams and milliliters, and avoid shellfish" — and ChatGPT follows them every time, without you having to repeat yourself. Projects are available to both free and paid ChatGPT users. ## Creating Your First Project In the ChatGPT web app, look at the left-hand sidebar. Near the top you will see a **Projects** section with a **New project** button or a "+" icon. On the mobile app, tap the menu icon in the top-left corner to open the sidebar first, then look for Projects there. Click **New project** and type a clear name — "Recipes," "Job Search," or "Family Documents" works well. A good name makes it obvious at a glance which workspace you are opening. You can rename it later, so don't overthink this step. This is the most powerful step. Inside the project, look for an **Instructions** field or a settings icon. Write what you want ChatGPT to always know in this context. For a Recipes project, you might write: > Always suggest recipes using metric measurements (grams, milliliters). I am vegetarian and avoid gluten. Keep suggestions simple — no more than 30 minutes of active cooking time. For a Job Search project: > I am applying for mid-level marketing manager roles in the Netherlands. My background is digital marketing with five years of experience. Always tailor suggestions to that role and that market. You can come back and edit the instructions anytime — ChatGPT will apply the updated version the next time you open a chat inside the project. Click the paperclip or file upload icon inside the project to add documents. ChatGPT Projects support PDFs, Word documents (DOCX), PowerPoint files (PPTX), Excel spreadsheets (XLSX), JPG images, and more. Each project can hold up to 5 files (Free), 25 files (Plus), or 40 files (Pro/Business/Enterprise). File storage counts against your total Library quota, not a per-project cap. Files sit at the top level — there are no sub-folders inside a project. If you hit the limit, delete files you no longer need and upload new ones. Click **New chat** inside the project. ChatGPT automatically applies your instructions and has access to your uploaded files. You do not need to re-explain who you are or what you need — it already knows. When you want to refer to a file, just mention it naturally: "Based on my CV, which skills should I highlight for this role?" I keep a project called "Admin" for anything dull but important — insurance questions, utility bills, government letters I don't understand. When an official letter arrives that I can't parse, I photograph it, upload it to that project, and ask "what is this asking me to do and by when?" Having a dedicated project means those conversations don't get mixed up with everything else, and I can find them again when the next letter arrives three months later. ## Three Projects Worth Setting Up Today ### The Recipes Project **What to put in it:** A PDF of your favourite recipes, a list of dietary restrictions or allergies, and any notes about equipment you own (for example, "I don't have a stand mixer" or "I cook on an induction hob"). **Instructions to write:** Your unit preference (metric or imperial), allergies or restrictions, how many people you usually cook for, and your rough cooking skill level. **Try this prompt:** > What can I make tonight with chicken thighs, tinned tomatoes, and whatever herbs and spices are typical to have on hand? Keep it under 40 minutes and make it work for two people. ### The Job Search Project **What to put in it:** Your current CV as a PDF or DOCX, any cover letters you have already written, and copies of job postings that interest you. **Instructions to write:** Your target role, the industry and location you are aiming for, your experience level, and any specific requirements (salary expectations, remote-only, etc.). **Try this prompt:** > Here is a job posting I want to apply for. Compare it to my CV and tell me: what are my three strongest matches, and what is the one gap I should address in my cover letter? ### The Family Paperwork Project **What to put in it:** Scans of insurance policies, a list of medications for family members, lease or rental agreements, warranty documents — anything you might need to look up in a hurry. **Instructions to write:** Names of family members, any important medical or practical context, and the language you'd like answers in if it differs from English. **Try this prompt:** > My home insurance policy document is uploaded. We had a water leak in the kitchen and some cabinets are damaged. Does this policy cover that, and what do I need to do to make a claim? ## Projects vs. ChatGPT Memory: What's the Difference? ChatGPT has two separate ways to "remember" things, and it helps to know which is which. **ChatGPT's general memory** saves facts about you across all your conversations — things like your name, your job, or that you prefer bullet points over long paragraphs. It applies everywhere, including inside projects, unless you turn it off. **Project instructions** only apply inside that specific project. If you tell your Recipes project to avoid shellfish, that instruction disappears the moment you switch to your Job Search project — which is exactly what you want. You can also set a project to use **project-only memory**, meaning ChatGPT won't draw on your general memory inside that workspace. This is useful if you want to keep a project completely separate — for instance, if you're testing something without your personal background influencing the answers. ## What to Watch Out For **Be thoughtful about what you upload.** Files you add to a project are sent to OpenAI's servers and processed to answer your questions. Don't upload documents containing passwords, national ID numbers, or sensitive financial details unless you have read OpenAI's privacy policy and are comfortable with how that data is handled. **Projects don't replace your own filing system.** If you need a document to exist long-term and be accessible outside ChatGPT, keep the original file somewhere safe — a folder on your computer, a cloud drive, or printed and filed. A ChatGPT project is a useful working context, not a document archive. **ChatGPT can still be wrong.** Uploaded context and clear instructions make ChatGPT far more accurate for your specific situation, but they do not make it infallible. For anything with real consequences — medical decisions, legal questions, financial planning — treat ChatGPT's answer as a starting point and verify with a qualified person. **The file limits are per project.** If you need to add more files than the limit allows, remove documents you no longer need first. ## What to Try Next The quality of your results inside a Project depends a lot on how clearly you write your instructions — [How to Write Better Prompts](/guides/how-to-write-prompts) shows you exactly how to phrase things so ChatGPT understands what you actually want. If you want to understand how ChatGPT's general memory works and how to control what it remembers about you, [ChatGPT Memory Settings Explained](/guides/chatgpt-memory-settings) is the place to start. --- # ChatGPT Scheduled Tasks: Set Up a Daily Briefing or Weekly Reminder (No Tech Skills Needed) Source: https://aiforregulars.com/guides/chatgpt-scheduled-tasks If you've ever wished you could get a short briefing every morning without lifting a finger, ChatGPT can now do that for you. The Scheduled Tasks feature — which relaunched with a dedicated sidebar section in June 2026 — lets you write a one-sentence instruction once, set a time, and have ChatGPT deliver the result to your phone automatically, day after day. No app-building, no code, no complicated setup. Just tell it what you want and when. **Update, August 29, 2026:** Scheduled Tasks is now available on the Free plan too, with a lower task limit than paid plans. See [ChatGPT Scheduled Tasks Are Free Now](/news/chatgpt-scheduled-tasks-now-free) for what changed — the steps below work the same regardless of which plan you're on. ## How to Create Your First Scheduled Task In the ChatGPT app on your phone (or at chatgpt.com in a browser), look for **Scheduled** in the left sidebar. On a phone, tap the menu icon in the top-left corner to open the sidebar first. This section was added in the June 2026 update — if you don't see it, make sure your app is up to date. Tap or click **New task** (or the plus icon near the top of the Scheduled section). ChatGPT opens a short form asking for two things: what you want it to do, and when you want it done. Write your task in one or two sentences, the same way you'd ask a friend for help. You don't need any special format or commands. A few examples you can copy and adapt: > Every weekday at 8 a.m., give me a two-paragraph summary of the most important world news today. > Every Sunday evening, suggest three simple dinner ideas for the week ahead. > Every Monday morning, give me five Spanish words to practice, with an example sentence for each. ChatGPT reads your description and interprets it — write whatever feels natural. Type when you want the task to run, in plain language: "every weekday," "every Monday at 9 a.m.," "once a week on Friday at noon," or whatever fits your routine. ChatGPT confirms back what it understood, so you can catch any misreading before saving. One limit to be aware of: tasks can run at most about once per hour. If you try to schedule something more frequently, ChatGPT will let you know. If this is your first task, ChatGPT will ask permission to send you push notifications — that's how the results arrive on your phone. Tap **Allow** when the prompt appears. If you're on a desktop browser, click **Allow** in the browser's permissions pop-up. If you accidentally clicked "Block," you can fix it in your phone's Settings app under the ChatGPT entry, or in your browser's site settings for chatgpt.com. Tap **Save** (or **Create task**). Your new task appears in the Scheduled sidebar with its next run time shown below it. When that time comes, you'll get a push notification on your phone — tap it to open ChatGPT and read the full response. ## Everyday Recipes Worth Trying Once you know how to create a task, adding new ones takes under a minute. Here are a few genuinely useful ones to get you started: **Morning news digest** > Every weekday at 7:30 a.m., summarize the three most important world news stories in two sentences each. Keep it factual and brief. **Weekly budget check-in** > Every Friday evening, ask me three short questions about my spending this week to help me notice any patterns. **Meal planning prompt** > Every Sunday morning, suggest a five-day dinner plan using simple, affordable ingredients. Include one vegetarian option. **Medication refill reminder** > Every Thursday, remind me to check whether any of my prescriptions need refilling in the next week or two. This kind of reminder can be a helpful nudge — but ChatGPT is not a medical device and has no access to your pharmacy, health records, or dosing schedule. For anything health-critical, keep your doctor's instructions and a dedicated alarm as your primary safety net. **Daily stretch or hydration reminder** > Every day at 11 a.m., remind me to stand up, stretch for two minutes, and drink a glass of water. **Language learning practice** > Every morning at 8 a.m., give me five words or phrases in French to practice today, with a short example sentence for each. I set up a Monday morning task asking ChatGPT to summarize the past week's AI news in plain, non-technical language. What surprised me was the timing: having it arrive before I open social media means I've already got a frame for the week — not whatever's loudest on my feed right then. If you try just one task, make it a Monday brief on whatever topic you actually care about. ## Managing Your Tasks: Pausing, Editing, and Deleting All your tasks live in the **Scheduled** sidebar. From there you can: - **Pause a task** without deleting it — useful when you're on vacation or a recurring task isn't relevant for a few weeks - **Edit the description or timing** — tap the task, make your changes, and save again - **Delete a task** permanently — tap the task and choose Delete One thing to know: if you receive several notifications from a task without tapping any of them, ChatGPT will automatically pause that task to avoid sending alerts you're not reading. You'll see it marked as paused in the Scheduled section, and you can re-enable it whenever you want. ## Task Limits by Plan How many tasks you can run at once depends on your ChatGPT subscription: | Plan | Active tasks | |------|------| | Free / Go | 3 | | Plus | 5 | | Pro / Enterprise | 15 | If you've reached your limit, pause or delete an existing task to make room. You can't use uploaded files, documents, or Custom GPTs inside a scheduled task — it runs on ChatGPT's general knowledge only. And tasks can run no more frequently than about once per hour. One more difference on Free and Go: tasks can only run on a schedule, not on an event — automatically kicking off a task when something happens in Gmail, Slack, or GitHub is still limited to Plus, Pro, Business, and Enterprise accounts. ## What to Watch Out For **ChatGPT doesn't have access to real-time information by default.** Scheduled tasks work best for analysis, ideas, language practice, and text-based reminders — not for live weather, current stock prices, or sports scores. If ChatGPT can answer the question right now without browsing the web, it will work well as a scheduled task too. **It's a convenience tool, not a critical reminder system.** Push notifications can be delayed by your phone settings, a poor signal, or a brief ChatGPT service outage. For deadlines that truly matter — medication times, legal filings, important appointments — use a phone alarm or calendar event as your backup. **Keep task descriptions free of sensitive data.** What you type into a task description is stored and processed by OpenAI each time the task runs. Avoid including account numbers, passwords, medical details, or other private information. **When a phone alarm is simply the better tool.** If you just need a nudge to take action — leave for an appointment, take a pill, call someone back — a phone alarm or calendar entry is simpler, faster, and works without an internet connection. ChatGPT Scheduled Tasks are most useful when you want a *response*, not just a *ping*: a digest, a set of suggestions, or a short analysis waiting for you when you wake up. ## What to Try Next For more ideas on what to ask ChatGPT with well-written prompts, [100 Things to Ask ChatGPT](/guides/things-to-ask-chatgpt) has a wide range of copy-and-paste starting points. And if you haven't installed the ChatGPT app on your phone yet — which is where the push notifications land — [How to Set Up the ChatGPT App](/guides/chatgpt-app-setup) walks you through it in just a few minutes. --- # Sol, Terra, or Luna: Which ChatGPT Model Do You Actually Have? Source: https://aiforregulars.com/guides/chatgpt-sol-terra-luna-explained On 9 July 2026, OpenAI released GPT-5.6 — at the time the newest version of ChatGPT — with a new naming system that confused a lot of people. Instead of a number like "GPT-4" or a letter like "GPT-4o," there are suddenly three models with celestial names: **Sol**, **Terra**, and **Luna**. (OpenAI has since released a newer flagship, GPT-6 Astra, in September 2026 — see our news on [what GPT-6 Astra changes](/news/chatgpt-gpt-6-astra-explained) if you're on a paid plan. Note that OpenAI also made Luna, not Terra, the default for Free and Go accounts starting in August 2026 — see [is ChatGPT free?](/guides/is-chatgpt-free) for the current free-tier default.) Sun. Earth. Moon. That naming is not random. It is actually a pretty good way to understand what each model is for — and which one you are already using. ## Why Sun, Earth, and Moon? Think of the three names as a size and brightness scale. Sol (the sun) is the biggest and most powerful. Terra (the earth) is the comfortable middle ground where most of us live. Luna (the moon) is smaller and faster, good enough for many tasks but not as capable as the others. OpenAI chose this metaphor deliberately: the three models are different versions of the same underlying system, just optimized for different trade-offs between speed, cost, and depth of thinking. ## The Three Models at a Glance Here is what each one does in plain English — no benchmarks, just real-world examples of where each one shines. | | Sol ☀️ | Terra 🌍 | Luna 🌙 | |---|---|---|---| | **Best described as** | The deep thinker | The all-rounder | The quick responder | | **Speed** | Slower | Comfortable | Fastest | | **Great for** | Complex analysis, long documents, research, coding | Emails, recipes, planning, explaining things, most everyday tasks | Simple questions, quick translations, short to-do lists | | **Not ideal for** | Quick back-and-forth chat | Very long multi-step technical work | Anything that needs careful reasoning | | **Who gets it** | Paid plans only (Plus, Pro, Business, Enterprise) | Everyone — it is the default for Free and Go | Paid plans only | | **Special features** | Optional "max reasoning effort" and Ultra mode (see FAQ) | None needed | None needed | **The honest summary:** Terra handles the vast majority of what people actually use ChatGPT for. The gap between Terra and Luna matters more than the gap between Terra and Sol for most readers of this site. I tested all three on the same everyday task: "Help me write a polite but firm email to my landlord about a heating problem that has been ignored for two weeks." All three produced a usable email. Sol's version was a little more structured and persuasive. Terra's was perfectly fine. Luna's was slightly shorter but still good enough to send. For that task, any of them would have worked. ## Which Model Do I Have Right Now? The model you have depends on your ChatGPT plan. ![ChatGPT model menu showing GPT-5.6 Sol](/screens/chatgpt-sol-terra-luna-explained/model-picker.webp) *The model menu on a ChatGPT Plus account — our own screenshot, 31 July 2026.* **If you use ChatGPT for free:** You have Terra. You do not need to do anything to set it up — it is the default for free accounts. You will not see a model picker in your interface; Terra is what runs when you open ChatGPT and start typing. **If you pay for ChatGPT Plus, Pro, Business, or Enterprise:** You can choose. Look for a model selector — usually a small dropdown near the top of the chat window or next to the text box. Here is how to check, step by step: 1. Open ChatGPT at chatgpt.com (or the app on your phone) 2. Start a new conversation 3. Look at the top of the screen — if you see a name like "ChatGPT" with a small arrow next to it, tap or click it 4. A menu will appear. If you only see one option (Instant) — or no menu appears at all — you are on the free plan, which uses a fast default model for everyday questions. If you are on a paid plan (Plus, Pro, Business, or Enterprise), you will see additional choices labelled Medium, High, and possibly Extra High or Pro. These represent different levels of reasoning effort, all powered by GPT-5.6 Sol. Medium is the right choice for most tasks; reserve High or Extra High for lengthy documents, detailed analysis, or complex multi-step problems. ## Do I Need to Pay to Get a Better Model? For most people: no. Terra, the free default, handles everyday tasks very well. Writing emails, explaining documents, brainstorming ideas, drafting a letter to a company, finding a recipe from what you have in the fridge — Terra does all of this reliably. Where Sol becomes genuinely worth considering: - You are doing something that requires careful, multi-step reasoning — reviewing a contract clause by clause, asking for a detailed analysis of a medical summary before a doctor's appointment, working through a complex financial decision - You regularly upload long documents (research papers, lengthy reports) and need careful, thorough responses - You are using ChatGPT for work in a technical field like software, law, or finance Where Luna is the right choice (for paid users): - You are on a paid plan and want the fastest possible response for a simple question - You are in a rapid back-and-forth where depth does not matter — brainstorming names for something, translating a short sentence, getting a quick definition If you are on the free plan and Terra feels slow or limited for what you need, it is worth reading [Is ChatGPT Free? What You Get Without Paying](/guides/is-chatgpt-free) and [Is ChatGPT Plus Worth It?](/guides/chatgpt-plus-worth-it) before deciding. ## What About the New Voice Mode? GPT-5.6 also introduced an updated version of ChatGPT's voice feature, now called **GPT-Live**. It lets you have a real back-and-forth spoken conversation with ChatGPT — more like talking to a person than dictating commands to a phone. We have a separate guide that covers it in detail: [ChatGPT Voice Mode: How to Use It and What It Is Good For](/guides/chatgpt-voice-mode). ## What to Try Next If you are new to ChatGPT and want to get comfortable with the basics first, [What Is ChatGPT? A Plain-English Guide for Beginners](/guides/what-is-chatgpt) is the best starting point. Once you are ready to actually send your first message, [How to Use ChatGPT: Step-by-Step for First-Timers](/guides/how-to-use-chatgpt) walks you through the whole thing. And if you are wondering how ChatGPT stacks up against Google's AI, [ChatGPT vs Gemini](/guides/chatgpt-vs-gemini) gives you an honest side-by-side comparison. ## Sources - [GPT-5.6: Frontier intelligence that scales with your ambition](https://openai.com/index/gpt-5-6/) — OpenAI - [Models reference](https://platform.openai.com/docs/models) — OpenAI - [ChatGPT plans and pricing](https://openai.com/chatgpt/pricing/) — OpenAI --- # ChatGPT Voice Mode: Talk Instead of Type (Guide for Seniors) Source: https://aiforregulars.com/guides/chatgpt-voice-mode If typing on a small phone screen feels frustrating, Voice Mode is the feature you have been waiting for. You simply speak your question out loud, and ChatGPT answers — out loud. No typing, no tapping, just conversation. This guide walks through getting Voice Mode set up on your phone, step by step. If you have not yet downloaded the ChatGPT app, start with our [ChatGPT app setup guide](/guides/chatgpt-app-setup) first, then come back here. Voice Mode only works in the ChatGPT phone app — not on the chatgpt.com website on your phone's browser. Open your phone and look for the ChatGPT app icon. It is a simple white shape on a dark background, made by OpenAI. If you do not have the app yet, go to the App Store (iPhone) or Google Play Store (Android), search for **ChatGPT**, and install the one made by **OpenAI**. It is free to download. Once the app is installed and you are logged in, come back to this guide. Tap the ChatGPT app to open it. You will see the main screen with a text box near the bottom for typing messages. Before starting Voice Mode, begin a fresh conversation. Look for a button that opens a new chat — it is typically near the top of the screen, often shown as a pencil or compose icon. Tap it. You will see a clean, empty conversation screen. This is where you will start your voice conversation. Near the text box at the bottom of the screen, look for a small waveform icon on the right side of the message bar. This is the Voice Mode button. Tap it once. The screen will change. The interface will switch to a dedicated voice conversation view — you will see a Live label at the top and the text bar will disappear. This means Voice Mode is active and listening. The first time you use Voice Mode, your phone will pop up a message asking: "Do you want to allow ChatGPT to access your microphone?" Tap **Allow** or **OK**. Without this permission, ChatGPT cannot hear you. If you accidentally tapped "Don't Allow," you can fix this: - **iPhone:** Go to your phone's Settings → scroll to ChatGPT → tap Microphone → switch it to On. - **Android:** Go to Settings → Apps → ChatGPT → Permissions → Microphone → Allow. After allowing the microphone, try tapping the Voice Mode button again. The first time I used Voice Mode, I was surprised by how natural it felt — more like talking to a knowledgeable friend than using a computer. I asked it to explain a confusing paragraph from my health insurance paperwork. It read the paragraph back in simpler words and then offered to explain each part. I did not type a single letter. Once the Voice Mode screen is active, just speak. You do not need to press any button to start or stop — ChatGPT listens automatically. GPT-Live is full-duplex, meaning it is listening even while it is speaking. So if ChatGPT starts going in the wrong direction, you can simply speak up and redirect it without waiting for it to finish its sentence. Speak at your normal pace and volume, as if talking to someone in the room. Here are a few things you might try your first time: - *"What is a good way to lower my blood pressure naturally?"* - *"Explain my Medicare plan in simple terms — what does it cover?"* - *"Tell me a short joke."* - *"What should I make for dinner tonight if I have chicken and potatoes?"* ChatGPT will think for a moment and then speak its answer back to you through your phone's speaker. Make sure your phone volume is turned up. You do not need to start over after each answer. Voice Mode is designed for a real back-and-forth conversation, just like talking to another person. You do not need to wait for ChatGPT to finish before you speak. GPT-Live is full-duplex — it listens and speaks at the same time, so you can jump in naturally mid-sentence, just like in a real conversation. If ChatGPT says something you did not understand, cut in and ask right away. You can respond to what it said, ask a follow-up question, or change the subject entirely. **Examples of natural follow-ups:** - "Can you say that again more slowly?" - "What did you mean by that last part?" - "Give me one more suggestion." - "Okay, now explain the second option." If the voice is speaking too fast, say "Can you speak more slowly?" It will adjust. When you want to stop the voice conversation, look for an X button or a stop/end button on the Voice Mode screen. Tap it to go back to the regular text chat screen. Your voice conversation is saved in the chat — you can scroll up and read everything that was said, in text form. This is useful if you want to copy down something ChatGPT told you. To use Voice Mode again later, just open the app, start a new chat, and tap the voice button again. It remembers your microphone permission, so you will not need to allow it a second time. ## Tips for Getting the Best Results **Speak in complete sentences.** "What are the side effects of ibuprofen?" gets a better answer than just "ibuprofen." **Say exactly what you need.** "Explain this to me simply, I am not a doctor" helps ChatGPT use plain language. **Ask it to slow down or repeat.** ChatGPT will not mind — just ask. **Use earbuds or headphones** if you are in a public place or want more privacy. ## Live Translation During Voice Conversations GPT-Live can translate between languages in real time during a voice conversation. A few situations where this can come in handy: - **Talking with a relative abroad.** If a family member speaks a different language, ChatGPT can relay what each of you says — you speak in English, it speaks in their language, and vice versa. - **Traveling.** Ask for help at a hotel, restaurant, or pharmacy abroad by speaking in English and having ChatGPT speak back in the local language. **A fair note on quality:** Live translation is a brand-new feature and the quality is still catching up. Early users noticed that the translated speech can carry a noticeable accent — think American-accented Hindi, for example. It works well for simple, casual situations, but do not rely on it for anything high-stakes like a medical appointment. No separate setting or menu is needed — once you are in a GPT-Live voice conversation, simply ask ChatGPT what you want, for example: "Please translate everything I say into Spanish." It will start right away and keep translating for the rest of the conversation. Think of it as a helpful bridge, not a replacement for a professional interpreter. ## What to Try Next If you want more ideas for what to ask once you are comfortable with Voice Mode, [25 things you can ask ChatGPT today](/guides/things-to-ask-chatgpt) gives you copy-ready prompts across a wide range of topics. All of them work in voice — just speak them instead of typing. --- # ChatGPT vs Copilot: Which One for Office Work? Source: https://aiforregulars.com/guides/chatgpt-vs-copilot Both tools promise to save you time at work. But they go about it in very different ways — and using the wrong one for the wrong job means extra steps instead of fewer. ## What each tool actually is **ChatGPT** is a chatbot made by OpenAI. You open it in a browser or app, type what you need, and it responds. Since September 2026, a free ChatGPT add-in can also work inside the Word, Excel, or PowerPoint file you currently have open — see our [explainer on ChatGPT in Word](/news/chatgpt-in-word-explained) — but it still cannot see your inbox or any file you have not opened, unless you paste the text in yourself. **Microsoft Copilot** is built on similar AI technology, but it lives *inside* Microsoft 365 apps: Word, Excel, Outlook, Teams, and others. It can see your documents and emails directly, because it is already there. That single difference — separate app versus built-in tool — drives almost every practical comparison between the two. ## Side-by-side comparison | | ChatGPT | Microsoft Copilot | |---|---|---| | Where you use it | Browser tab or mobile app | Inside Office apps + copilot.microsoft.com | | Free tier | Yes (with limits) | Yes (basic web version) | | Works inside Word/Excel/PowerPoint | Yes (free add-in, current document only) | Yes (paid 365 plan required) | | Reads your emails | No | Yes (Outlook) | | Internet access | Yes | Yes | | Image generation | Yes (paid plan) | Yes (free, via Designer) | | Best for | Open-ended tasks, research, writing from scratch | Office document work, email, meeting summaries | ## Where Copilot clearly wins If you spend your day in Word, Excel, or Outlook, Copilot is the obvious pick for those specific tasks. **Summarizing long email threads.** In Outlook, you click one button and get a bullet-point summary of a 30-message chain. With ChatGPT, you would need to copy every message, paste it in, then ask. **Drafting in Word.** Copilot sees your existing document and can continue it, rewrite a paragraph in a different tone, or shorten a section — all without leaving the app. **Excel formulas.** Describe in plain English what calculation you want and Copilot drafts a formula and explains what it does — a useful starting point that typically gets you most of the way there, though Microsoft recommends reviewing generated formulas before relying on them for critical work. **Meeting recaps.** In Teams, Copilot can transcribe a meeting and pull out action items automatically. ## Where ChatGPT clearly wins **Flexibility.** ChatGPT handles almost anything — summarizing a PDF you upload, translating text, explaining a legal document, writing a cover letter, creating a quiz. It is not tied to one company's ecosystem. **No extra subscription for basic use.** The free version of ChatGPT is powerful enough for most daily tasks. The Copilot that works inside Office apps requires a paid Microsoft 365 Copilot plan on top of your regular subscription. **Better for creative and open-ended writing.** Starting a newsletter from nothing, brainstorming taglines, or writing a speech feels more natural in ChatGPT, where the whole interface is built around back-and-forth conversation. **Works without Microsoft.** If your company uses Google Workspace, or you work mostly outside Office apps, ChatGPT fits more naturally into your day. In my experience, I use Copilot for anything already living inside a document — revising, formatting, pulling key points from a meeting transcript. I switch to ChatGPT when starting from a blank page or when I need something creative that has nothing to do with an existing file. Keeping both available takes less effort than it sounds. ## The price question Copilot at copilot.microsoft.com is free and reasonably capable for general questions. But the version that lives inside Word and Excel costs extra. If you already pay for Microsoft 365 and do most of your work in Office apps, that upgrade can pay for itself quickly — especially if you spend a lot of time on emails and reports. As of mid-2026, the Microsoft 365 Copilot add-on costs around $30 per user per month for enterprise plans and $21 per user per month for small-business plans (annual commitment required; check Microsoft's pricing page at microsoft.com/en-us/microsoft-365-copilot/pricing for current figures, as these are updated periodically). If your tasks mostly happen outside Office, ChatGPT's free tier likely covers you. ## Which one should you choose? Pick **Copilot** if you live in Word, Excel, Outlook, or Teams and want AI that sees your documents without any copy-pasting. Pick **ChatGPT** if you need a flexible general-purpose assistant, want to start for free before committing, or your work goes beyond the Microsoft Office suite. Use **both** if you do a mix of document work and open-ended tasks — that is the most common setup among regular AI users. ## What to try next If you decide to start with Copilot, our step-by-step [Copilot in Word and Excel guide](/guides/copilot-word-excel-guide) will get you productive in under 15 minutes. Want to see how other AI tools compare? [ChatGPT vs Gemini](/guides/chatgpt-vs-gemini) covers another popular rivalry worth knowing about. --- # ChatGPT vs Gemini: Real-World Test (Same 20 Tasks) Source: https://aiforregulars.com/guides/chatgpt-vs-gemini Two of the most popular AI assistants right now are ChatGPT (made by OpenAI) and Gemini (made by Google). Both are free to try, both answer questions and help with writing, and both can feel a bit overwhelming if you are new to AI. So which one should you actually use? To structure this comparison, we mapped findings from published head-to-head evaluations across 20 everyday task types — the same categories independent reviewers consistently use to put AI assistants side by side. ## How This Comparison Is Structured The comparison covers five categories, four task types each: 1. **Writing** — draft a complaint email, write a birthday message, summarize a long article, improve a paragraph 2. **Everyday questions** — explain a medical term, help plan a weekly menu, calculate a tip, explain a lease clause 3. **Work tasks** — write a meeting agenda, create bullet points from notes, suggest interview questions, draft a job posting 4. **Voice and phone use** — ask for directions, set a reminder via voice, ask a follow-up question mid-conversation, use on a mobile screen 5. **Beginner experience** — how easy is the signup, how clear are the answers, how does it handle a confused follow-up, how does it explain its own limits Each category looks at: accuracy, readability, and whether the answer actually helped — not word count or technical complexity. ## Head-to-Head Results by Category ### Writing Tasks Both tools produced solid writing in published reviews. For emails and short messages, the quality was close enough that most people would be happy with either. Where reviewers consistently noticed a difference: ChatGPT's tone felt slightly more natural in longer pieces — less formal-sounding, more like something a real person would write. Gemini sometimes defaulted to a slightly stiff, corporate tone on the first try. That said, both tools respond well to feedback. Telling either one "make this sound warmer" or "shorter, please" produced noticeably better results in a second pass. **Edge: ChatGPT (slight)** ### Everyday Questions For factual questions — what does a term mean, how does something work, what should I eat this week — both performed well on common questions. Gemini has one meaningful advantage here: it can pull current information from Google Search. If you ask about something that happened recently, Gemini is more likely to give you up-to-date information. ChatGPT's free version now includes web search, though Gemini's tighter Google integration tends to deliver fresher, more citation-rich results. For timeless questions (medical terms, cooking, household tips), the answers were comparable. **Edge: Gemini (for recent or time-sensitive topics)** ### Work Tasks Meeting agendas, bullet-point summaries, job descriptions — both tools handled these well. This is where AI assistants genuinely save time, and neither one disappointed in published reviews. Gemini has a practical advantage for Google Workspace users: it is built into Google Docs and Gmail. You can ask it to draft something without leaving the app you are already in. If your job runs on Google tools, that integration is genuinely convenient. ChatGPT has similar integration through Microsoft's tools (Copilot in Word and Outlook) if that is your workplace environment. **Edge: Gemini (for Google Workspace users), ChatGPT (for Microsoft users)** ### Voice and Phone Use Both tools have mobile apps and voice input. In published reviews, both tools understood spoken questions accurately and responded clearly. Gemini's voice mode felt slightly more conversational — it handled follow-up questions mid-conversation a bit more smoothly. ChatGPT's voice mode has improved significantly and is comfortable for most tasks. On a small phone screen, both apps are easy to navigate. Gemini's app benefits from Google's polish in mobile design. **Edge: Gemini (slight)** I tested both tools on the same tricky request: "I got a letter from my insurance company saying my claim was denied — help me write a response to appeal it." ChatGPT gave me a ready-to-send letter on the first try that felt personal and firm. Gemini's first draft was more generic. Both got me to a good result, but ChatGPT saved me an extra round of editing. ### Beginner-Friendliness This category matters most for new users. **Signup:** Both are straightforward. Gemini requires a Google account, which most people already have. ChatGPT lets you sign up with any email or a Google/Microsoft account. **First experience:** ChatGPT's interface is slightly cleaner — a single message box, fewer menus. Gemini's interface is also simple but integrates with Google's wider ecosystem, which can feel like a lot if you just want to ask a question. **Handling confusion:** In published comparisons, both tools were asked a deliberately vague question ("Can you help me with something important?") and then a confused follow-up. Both were patient and helpful. Neither talked down to the user. **Explaining limits:** Both acknowledged when they might be wrong. ChatGPT was slightly more direct about saying "I'm not certain about this — please verify with a professional." **Edge: ChatGPT (slight)** ## Overall Comparison Table | Category | ChatGPT | Gemini | Edge | |---|---|---|---| | Writing quality | Excellent | Very good | ChatGPT | | Current/recent info | Good (with browsing) | Excellent (Google Search) | Gemini | | Work tasks | Excellent | Excellent | Tie | | Google Workspace integration | Limited | Built-in | Gemini | | Microsoft Office integration | Via Microsoft Copilot (requires M365 Copilot licence) | Limited | ChatGPT | | Voice and mobile | Very good | Very good | Tie | | Beginner friendliness | Excellent | Very good | ChatGPT | | Free plan availability | Yes | Yes | Tie | *Quality ratings reflect consensus across published 2026 AI assistant comparisons, not proprietary testing.* ## So Which One Should You Use? **Choose ChatGPT if:** - You are new to AI and want the simplest starting point - Writing, editing, or explaining things is your main use - You use Microsoft Office at work **Choose Gemini if:** - You live in Google's world — Gmail, Docs, Sheets, Drive - You want AI that can look up current news and recent information - You prefer everything in one Google account **Use both if:** - You want to compare answers on something important before acting - You are curious and like exploring The honest truth is that for most everyday tasks — drafting an email, getting a recipe, understanding a confusing document — either tool will serve you well. The best one is the one you will actually use. ## What to Try Next If you have not tried either tool yet, start with [How to Use ChatGPT: Step-by-Step for First-Timers](/guides/how-to-use-chatgpt) — it walks you through creating a free account in minutes. Once you are comfortable, you can try Gemini at gemini.google.com to see which one feels right for you. For a practical first task with either tool, [writing a professional email with AI](/guides/write-professional-email-with-ai) is a great place to start. --- # Circle to Search: The Feature Everyone Actually Uses Source: https://aiforregulars.com/guides/circle-to-search You are watching a recipe video on Instagram. The host uses a spice you have never heard of. You want to look it up, but if you leave the app, you lose your place and have to scroll back through everything. That used to be the trade-off. Circle to Search removes it entirely. Here is how to use it and what it is best for. ## Step 1: Make Sure Circle to Search Is Enabled Circle to Search is available on Pixel phones (Pixel 6 and newer) and Samsung Galaxy phones (Galaxy S21 series and newer, Galaxy Z Fold and Z Flip models from 2021 onward, and many Galaxy A-series models). To check: go to **Settings**, tap **Display** or **Navigation**, and look for **Circle to Search** as a toggle. On Samsung, it may be under **Advanced Features**. If you do not see it, your phone may not support it yet or may need a software update. Long-press the **Home button** (the button or gesture bar at the very bottom of your screen). On phones with gesture navigation (where you swipe instead of tapping), long-press the navigation bar at the bottom — a glow effect appears around the edges, signaling that Circle to Search is active. You do not need to leave the app you are in. The screen slightly dims and a search panel begins to slide up from the bottom. With the screen in Circle to Search mode, you can interact with anything visible: - **Circle** an object, product, or face by drawing a loop around it with your finger - **Tap** a specific spot to identify what is there - **Scribble** over text or objects if you want to select a larger area - **Highlight** a word or phrase by holding and dragging, just like selecting text Google then searches for whatever you indicated. The results appear in a sheet that slides up from the bottom — you can scroll through them and tap links without ever leaving the app you were in. Circle to Search uses Google Lens technology, which means it can identify many real-world things from a photo or screenshot. Practical uses: - **Shopping:** Circle a pair of shoes or a piece of furniture to find where to buy it and how much it costs - **Plants and animals:** Circle a houseplant or a bird outside the window to identify the species - **Landmarks:** Circle a building or monument in a travel photo to find out what it is - **QR codes and barcodes:** Tap a QR code to open the link without a separate scanner app It works on live content as well as static images — so you can pause a YouTube video on a product you noticed and circle it right there. This is one of the most useful applications. If you see text in another language — in a social post, a photo, a menu, a screenshot from a friend — activate Circle to Search and highlight the text. Google translates it in place, overlaid on the original text, without you opening a translation app. The translation quality is the same as Google Translate. This works for handwritten text too, as long as it is legible. I started using Circle to Search almost every day within a week of getting a Pixel 8. The most common use for me is shopping: I see something in an Instagram reel, circle it, and immediately see where to buy it and whether it is on sale elsewhere. It has also saved me on several hikes — I circled a mushroom I was not sure about and found enough information to decide not to touch it. When Circle to Search is active, a small music note icon appears at the bottom next to the text field. Tap it and hold your phone near any audio source — a video, a TV across the room, music from a speaker. Google identifies the song, shows the title and artist, and links to streaming services where you can save or play it. This feature works similarly to Shazam but from inside any app. ## What to try next Circle to Search is one of several AI tools built directly into Android — for a full tour of what else your phone can do without any extra apps, see [Hidden AI Features on Your Smartphone You Did Not Know About](/guides/hidden-ai-features-smartphone). If you are on Samsung specifically, [Galaxy AI: A Complete Guide](/guides/galaxy-ai-features) covers the full set of AI features Samsung has built on top of Android. --- # Claude vs ChatGPT for Everyday Writing Source: https://aiforregulars.com/guides/claude-vs-chatgpt Both Claude and ChatGPT are general-purpose AI assistants that have become go-to tools for everyday writing. They share a lot in common — you type a request, they respond in natural language, and both can help you write better. But they have real differences that show up depending on the task. Here is how they compare on the writing tasks most people actually do. ## The writing tasks tested Based on dozens of published head-to-head comparisons, here are the everyday writing tasks both assistants are most commonly evaluated on: 1. Write a professional email declining a meeting invitation 2. Summarize a 1,200-word article in three bullet points 3. Edit a paragraph to make it clearer and shorter 4. Write a bio for a LinkedIn profile 5. Draft a short complaint letter to a company 6. Explain a technical concept to a non-expert 7. Rewrite a casual text as a formal message ## Side-by-side comparison | Writing task | Claude | ChatGPT | |---|---|---| | Tone control (formal vs. casual) | Excellent | Excellent | | Long document handling | Excellent | Good | | Editing existing text | Excellent | Very good | | Writing from scratch | Very good | Excellent | | Following detailed instructions | Excellent | Very good | | Speed of response | Good | Excellent | | Free tier quality | Good | Good | *Ratings reflect consensus from published head-to-head comparisons; individual results vary by prompt, model version, and use case.* ## Where Claude stands out **Long documents.** Claude can take a very long piece of text — a report, a contract, a lengthy email chain — and work with the whole thing without losing the thread. If you have ever pasted a long document into a chatbot and gotten a response that seemed to ignore half of it, Claude's handling of length is noticeably better. **Careful, nuanced writing.** When the task calls for getting the tone exactly right — a sensitive message to a colleague, a resignation letter, a thank-you note that does not sound generic — Claude tends to produce drafts that need less revision. It is attentive to subtext in a way that stands out on delicate writing tasks. **Following detailed instructions.** If you give Claude a long prompt with multiple requirements ("write in a friendly but professional tone, keep it under 150 words, do not mention the price, and end with a question"), it tends to honor all the constraints. ChatGPT sometimes drifts away from one of them. **Explaining its reasoning.** Claude often explains why it made a particular word choice or structural decision, which is useful if you are trying to learn rather than just get a draft. ## Where ChatGPT stands out **A larger ecosystem.** ChatGPT has been available longer and has a much larger community of people sharing tips, prompts, and workflows. If you search for help with a specific task, there are more tutorials built around ChatGPT. **Integrations.** ChatGPT connects to a wide range of tools and plugins. If you want your AI assistant to interact with your calendar, search the web, run code, or generate images, ChatGPT currently has more options available. **Speed.** ChatGPT often responds faster, which adds up when you are going back and forth through multiple drafts. **More familiar interface.** Because ChatGPT launched earlier and got more press, many people find the interface more intuitive simply from familiarity and from the larger pool of getting-started guides. I use both regularly, and the honest version is that for most everyday writing tasks — drafting a quick email, editing a paragraph, rewriting something for a different audience — I cannot consistently predict which one will give me the better first draft. What I have noticed is that when a task is genuinely delicate — a message where tone really matters — I tend to reach for Claude first. When I need to use a specific integration or want something turned around very fast, I go to ChatGPT. ## Pricing comparison Both have free tiers that are genuinely useful for casual writing. The paid subscriptions are priced similarly and unlock higher usage limits and access to more powerful model versions. If you only write occasionally, the free tier of either will likely be enough. If you write daily and hit the usage limits, the paid tier is worth considering for whichever tool you prefer. ## Which one should you use? If you are choosing for the first time: **start with ChatGPT**. It has more getting-started resources, and the broader ecosystem means it is more likely to connect to tools you already use. If you work with long documents, do a lot of editing, or want writing that feels more considered: **give Claude a serious try**. The difference is most noticeable on longer or more nuanced tasks. If you write daily and want the best output: **try both for a week** on the same tasks and trust your own judgment. The gap between them is small enough that personal preference ends up mattering a lot. ## What to try next If you want to see how ChatGPT compares to Google's option, [ChatGPT vs Gemini](/guides/chatgpt-vs-gemini) covers that matchup in detail. Or if you are still getting started, [How to Use ChatGPT](/guides/how-to-use-chatgpt) walks you through your first conversation step by step. --- # Microsoft Copilot on Windows 11: The No-Nonsense Setup Guide Source: https://aiforregulars.com/guides/copilot-windows-guide Windows 11 has a free AI assistant that many people walk right past. Microsoft Copilot sits quietly in your taskbar, and once you know how to use it, it can save you real time on tasks you do every day — writing emails, summarizing long documents, adjusting settings, or just answering questions while you work. Here is how to get it set up and what to actually do with it. Look for the Copilot icon in the taskbar at the bottom of your screen, typically toward the right side. Click it to open the Copilot panel. If you do not see it, press **Windows key + C** on your keyboard. You can also search for "Copilot" in the Start menu. When Copilot opens, you will see a prompt to sign in. Click **Sign in** and use your existing Microsoft account (the one connected to Outlook, Xbox, or OneDrive), or create a free one at microsoft.com. Signing in lets Copilot remember context across a conversation and gives you access to more features. If you prefer not to sign in, you can still use a limited version. Type your question in the chat bar at the bottom of the Copilot panel. Write naturally — you do not need special commands or keywords. Good starting points: "What is the weather this weekend in [your city]?" or "Explain what a VPN is in simple terms." Copilot answers in a clear paragraph, not just a list of links. You can ask follow-up questions the same way you would in a normal conversation. Copilot is genuinely good at writing first drafts. Tell it what you need and it will produce a clean draft you can edit. Example prompts you can type directly: ``` Write a short email to my doctor's office asking to reschedule my appointment to next week. ``` ``` Draft a polite complaint letter to my internet provider about a billing error. Keep it professional and under 200 words. ``` Copy the result, paste it into your email client or Word document, and adjust anything that does not sound quite right. The draft gives you a starting point so you are not staring at a blank screen. If you have a long PDF or webpage you need to understand quickly, Copilot can read it and give you a summary. Open the document in Edge (Microsoft's browser), then open Copilot and type "Summarize this page" or "What are the main points of this document?" Copilot reads the content in the browser tab and gives you a short summary. You can follow up with questions: "What does section 3 say about refunds?" or "Explain the part about fees in plain English." I use Copilot most often to untangle confusing official letters. Insurance explanation-of-benefits forms, HOA notices, terms-of-service updates — I paste the text into the chat and ask "What does this actually mean for me?" The plain-English summary usually takes about ten seconds. It has saved me a lot of frustrated rereading. You can ask Copilot to adjust Windows settings instead of hunting through menus. Try requests like: - "Turn on dark mode" - "Open Bluetooth settings" - "Make the text on my screen bigger" Copilot either makes the change directly or shows you exactly where to go. This is especially useful if you have trouble remembering where a particular setting is buried in the Windows menus. Copilot stays in a panel on the side of your screen, so it does not take over your whole display. You can keep working in your email, browser, or any other program while Copilot is open on the right. Use it for quick lookups, to check spelling or grammar, to translate a short phrase, or to brainstorm ideas while you work. It is most useful when you treat it as a reference tool that is always one click away rather than a separate destination you have to visit. ## What Copilot cannot do Copilot is not connected to your personal files, calendar, or email by default. It works with content you paste into it or pages open in Edge. It also does not make purchases, send emails, or take actions on your computer without your confirmation. Like all AI assistants, it can occasionally give you incorrect information. For anything important — medical advice, legal questions, financial decisions — verify what it tells you with a real expert or a trusted source. ## What to try next If you use Microsoft Word or Excel regularly, [Copilot inside Word and Excel](/guides/copilot-word-excel-guide) shows how to use AI directly in your documents. And if you want to get better results from any AI tool, [how to write good prompts](/guides/how-to-write-prompts) is a short guide that makes a noticeable difference. --- # Copilot in Word and Excel: What That Button Actually Does Source: https://aiforregulars.com/guides/copilot-word-excel-guide That small purple Copilot icon has been sitting in your toolbar for months. Maybe you've been ignoring it, unsure what it actually does. You're in good company — most people who use Word and Excel every day still haven't clicked it. This guide walks you through exactly what happens when you do. Open any Word document or Excel spreadsheet. Look for the **Copilot** button — a floating icon anchored in the bottom-right corner of your document or spreadsheet canvas. Clicking it opens a chat panel on the right side of your screen. If you prefer a toolbar button, right-click the icon and choose **Move to ribbon**. If you don't see it, your Microsoft 365 account may not include the Copilot add-on license. Ask your IT department or visit your Microsoft account page to check. Open a report, contract, or meeting notes that would normally take you ten minutes to skim. Click the Copilot button, then type into the panel: *"Summarize this document in five bullet points."* Copilot reads the whole file and returns a condensed version in seconds. This is one of the most immediately useful things it does — you get the key points without reading every paragraph. Click in your document where you want to add text. A small Copilot icon appears in the left margin — click it, or use the Copilot panel directly. Type what you need, like: *"Write a short introduction for a report about our Q2 sales results."* Copilot generates a draft paragraph you can keep, edit, or throw away. Think of it as a first draft you didn't have to write yourself. Highlight any text you've already written. Right-click and look for a Copilot option, or paste the text into the Copilot panel and describe what you want. Requests that work well: - *"Make this shorter and easier to read."* - *"Rewrite this in a more professional tone."* - *"Simplify this for someone outside our industry."* Copilot offers a revised version. You decide whether to use it. I found the rewrite feature most useful when I had a blunt draft email I didn't want to send as-is. I asked Copilot to make it more diplomatic, and it came back with something I was happy to actually send — in about 30 seconds, instead of me staring at the screen trying to find the right words. Open a spreadsheet with rows and columns of data — a budget, a list of sales figures, anything. Open the Copilot panel and ask a plain-English question about it. Try: *"What does this data tell me about spending over the last three months?"* or *"Which product had the highest total?"* Copilot answers in normal sentences. No formulas needed to get basic answers. Instead of searching online for the right Excel formula, describe what you want to Copilot. In the panel, type something like: *"Give me a formula that adds up all sales in column B where column A says 'North.'"* Copilot writes the formula and explains what each part does. Copy it into your cell. Always test it on a few rows whose correct answers you already know — Copilot occasionally gets the cell range wrong. Select your data range in Excel, open the Copilot panel, and type: *"Create a bar chart showing monthly revenue."* Copilot can suggest chart types and insert one directly into your sheet. If it doesn't look right, describe the change: *"Show only the top five categories"* or *"Switch to a line chart."* Keep refining until it looks the way you need. ## What to try next Now that you know the basics, sharper prompts will get you much better results. The guide on [writing AI prompts that actually work](/guides/how-to-write-prompts) shows exactly how to phrase requests so you get what you want on the first try. If your workplace uses Google tools instead of Microsoft, [Gemini in Gmail and Google Docs](/guides/gemini-google-docs-gmail) covers the same ground for that ecosystem. --- # How to Create AI Images for Free: Text to Picture in 3 Steps Source: https://aiforregulars.com/guides/create-ai-images-free You don't need Photoshop, design skills, or an art degree to create a good-looking image anymore. You just need words. Type what you want to see, and an AI tool turns your description into a picture in seconds. Here's how to do it on any computer or phone — completely free. ## The three steps at a glance 1. Pick a free tool and open it 2. Type a clear description 3. Generate, pick the best result, and download Let's go through each one. All of these are free to use with a basic account: - **Adobe Firefly** (firefly.adobe.com) — clean, realistic images; free account with 25 monthly credits - **Microsoft Designer** (designer.microsoft.com) — powered by the same AI as DALL·E; free with a Microsoft or Outlook account - **Canva** (canva.com) — drag-and-drop friendly; AI image generation included in the free plan - **Google Flow** (flow.google) — Google's free creative tool for images and video, powered by the Nano Banana model; requires a Google account Any of these will work. If you already have one of these accounts, start there. The description you type is called a **prompt**. Think of it as telling a very skilled artist exactly what to paint. A bare-bones prompt: *"a cat"* A much better prompt: *"A fluffy orange tabby cat sitting on a sunny windowsill, looking outside, warm afternoon light, realistic photo style"* The difference is specificity. Try to include: - **What** is in the image (the main subject) - **Where** it is (the setting or background) - **What it looks like** (colors, mood, time of day) - **What style** you want (photo-realistic, watercolor, cartoon, oil painting) You don't need all four every time, but adding even one extra detail makes a noticeable difference. My go-to trick: describe the image the way you'd describe a photograph to someone on the phone. "There's a little girl in a yellow raincoat jumping in a puddle on a rainy street, she's laughing, it looks like an old film photo." That level of detail almost always gets me something I actually want to use. Most tools generate 4 images at once from your single prompt. Each one will be slightly different. Look through them and pick the one you like best. If none of them are right: - Click **Generate again** for a fresh set with the same prompt - Or go back and add more detail to your description Common tweaks that help: - Too dark? Add *bright*, *well-lit*, or *sunny* - Too cluttered? Add *simple background* or *minimalist* - Not realistic enough? Add *photorealistic* or *ultra-detailed photo* - Too realistic? Add *illustration*, *flat design*, or *cartoon* Once you have an image you like, click the download button. It will save as a JPG or PNG file. - JPG is fine for social media and sharing - PNG is better if you need a transparent background (some tools offer this) Your downloaded image is yours to use for personal projects, social media posts, greeting cards, and more. If you plan to use an image commercially — for a business, a product, or anything you sell — check the license terms for the tool you used, as rules vary by platform. ## Ideas for what to make Not sure what to try first? Here are some easy starting points: - A custom birthday card image (a birthday cake with a specific number of candles) - A picture of your pet in a funny scenario ("my golden retriever as an astronaut") - A landscape from your dream vacation destination - A simple logo or icon idea for a hobby project - An illustration to go with a story you're writing ## Prompt tips quick reference | What you want | Words to add | |---|---| | Realistic photo look | "photorealistic, high resolution" | | Painting style | "oil painting, impressionist" | | Simple cartoon | "flat design illustration, vector" | | Warm, cozy feel | "warm light, golden hour, cozy" | | Clean background | "white background, studio lighting" | ## What to try next Once you're comfortable creating still images, [making a free AI video](/guides/make-ai-video-free) works the same way — just a few words and a click. And if you want to use AI images in a designed layout, [Canva's AI tools](/guides/canva-ai-beginners) let you drop your generated images straight into invitations, cards, and social posts. --- # Is Your Match Real? How to Tell If You're Talking to an AI on a Dating App Source: https://aiforregulars.com/guides/dating-app-ai-bot-detect You matched, you've been chatting for a few days, and the conversation feels good — warm, interested, a little exciting. But something is slightly off. The replies come too fast, the messages are too smooth, and somehow you've talked for a week without learning a single specific thing about this person's actual life. You might be talking to an AI. This isn't a rare edge case anymore. Research into romance-scam operations suggests AI is already doing most of the conversational heavy lifting — a 2025 academic study (arxiv.org/abs/2512.16280, later presented at USENIX Security 2026) found that 87% of the conversational labor in romance-baiting scams consists of systematized tasks readily susceptible to LLM automation. And it isn't only scammers: some dating apps have been accused of running their own bots to keep users engaged and less likely to cancel subscriptions. A 2025 Norton survey of online daters in the US found that 60% believe they have had at least one conversation written by AI. ## Two Kinds of AI You Might Encounter Before getting into the signals, it helps to know what you might be dealing with — because the two main scenarios are different. **Platform bots** are run by the dating app itself. Their goal is to keep you swiping, messaging, and staying subscribed. They're not necessarily trying to trick you into sending money; the app already profits from your continued engagement. These are the AI encounters where a conversation can feel great for a while, then eventually you notice nothing is leading anywhere. **Chatfishing** is when someone creates a fake profile to deceive you personally — using AI to handle the conversation while building enough trust to eventually ask for money, personal information, or something else of value. A human operator may or may not be actively involved; the AI can run much of the conversation on its own. Norton reported blocking 17 million dating scam attacks in the fourth quarter of 2025 alone, a 19% increase year-over-year. The signals below apply to both. The difference matters mainly when you're deciding what to do about it. ## Signal 1: It Won't Get on a Video or Phone Call This is the clearest single indicator. A real person who is genuinely interested in you will eventually agree to a phone or video call — not necessarily the first time you ask, but at some point, the question of actually talking comes up and gets answered. An AI bot cannot do a live video call without a human behind it. A chatfisher using AI as a writing tool can do a call, but doesn't want to, because the illusion ends immediately. **Watch for the pattern of excuses.** One declined call is nothing. A consistent string of camera problems, bad WiFi, odd working hours, or always "just about to go somewhere" — across weeks of otherwise enthusiastic chatting — is meaningful. **Ask early.** You don't need to wait weeks before suggesting a quick call. A real person who is interested but genuinely can't right now will say "soon" and follow through. Watch whether they do. > "Hey, would you want to do a quick five-minute video call this week? Just to actually see each other — no pressure." ## Signal 2: Perfect Replies, Any Hour, Instantly Real people have lives. They reply when they're free — sometimes immediately, sometimes hours later. They type fast sometimes and slow other times. They occasionally misread a question, drift off-topic, or send a message that clearly got written in a hurry. They're tired, distracted, in the middle of something. AI doesn't have any of that. It's available at 3 a.m. on a Tuesday. It never takes longer than a few seconds. Its messages arrive clean and coherent every single time. **Consistent instant replies at unusual hours are a tell.** One late-night chat isn't unusual. A pattern of perfectly smooth, fast responses whenever you message — including 2 a.m., on public holidays, in the middle of workdays — starts to stand out. **The writing never varies.** No typos, no autocorrect errors, no trailing sentences that went somewhere and then didn't. Some people do write carefully, but if the style is identical at midnight and 9 a.m. and during their supposed lunch break, that's worth noting. ## Signal 3: Warm, Interested, and Weirdly Vague AI is genuinely good at seeming emotionally present. It reflects your feelings back at you, asks follow-up questions, says the supportive and interested thing. In a dating context, where you're hoping the other person genuinely likes you, this lands well — sometimes very well. What AI struggles with is specificity anchored in a real life. A real person who's told you they work at a bakery will eventually mention something specific — a burnt batch, a difficult customer, a coworker's name. Their life leaks into conversation in small, real ways. **Look for depth without detail.** If your match is warm and engaged but after many conversations you still don't know what their neighborhood looks like, what their apartment is like, or one specific story from their actual week — that vagueness is a signal. **Notice if it never interrupts.** Real people interject unexpectedly, circle back to something you said earlier in an odd moment, forget what they already told you, or occasionally say something that doesn't quite follow. AI maintains perfect conversational continuity and never forgets what the "character" has established. That consistency is, paradoxically, a red flag. When I'm trying to figure out if a conversation feels real, I look for moments of inconvenience — someone saying they can't talk right now, that they're bad at responding on weekends, or that their phone is dying. Those tiny frictions signal there's an actual person on the other side with an actual life. A match that is always available, always engaged, and never has any friction anywhere gives me pause. ## Signal 4: Photos That Look Almost Too Good AI-generated profile photos have improved dramatically. The giveaways of a few years ago — odd fingers, blurred ears, wavy backgrounds — are rarer now. But some patterns remain. **Look for an oddly consistent aesthetic.** Real people's photo collections are messy — some from a great day, some not, different lighting, different company. A profile where every photo looks like it came from the same professional shoot, with consistent soft lighting and the same flattering camera angle in every single one, might be AI-generated or curated from someone else's photos. **Try a reverse image search.** On a desktop browser, right-click any profile photo and choose "Search image" (Chrome or Edge) or save it and upload to Google Images or TinEye. If the same face appears elsewhere under different names, that's conclusive. ## Tests That Can Help Expose an AI None of these are foolproof, but they create situations AI handles poorly. **Ask about a very recent local event.** Something from the last 48 hours in a specific location — a local sports result, a neighborhood news story, anything from that city's local feed. AI has a training data cutoff and doesn't have live local news. A real person will either know it or say they haven't heard. An AI will deflect, give a generic answer, or make something plausible-sounding up. > "I saw something happened near downtown [their city] yesterday — did you catch that?" **The spontaneous specific selfie.** Ask for a photo right now with something specific and unplanned — holding up three fingers, standing next to whatever is on their left, or with their hand on their chin. Make it odd and unscripted. A real person can do this in 30 seconds. A bot cannot. A chatfisher with a stolen photo library cannot pull it off quickly and naturally. > "Random request — can you send me a selfie right now holding up two fingers? I just want to see your face today." **Absurdity and humor.** Drop in something genuinely odd mid-conversation: a non-sequitur, a terrible pun, a completely random question about whether they'd rather fight one horse-sized duck or a hundred duck-sized horses. Real people react unpredictably. AI responds smoothly and thoughtfully, treating the odd remark as a reasonable conversational contribution. The absence of genuine confusion or amusement is a tell. **Recall a specific detail unexpectedly.** If they mentioned something specific early on, bring it up again in an unusual way a week later. Real people misremember things, get details slightly wrong, or say "wait, when did I tell you that?" AI maintains perfect recall of every detail the character has established. ## Why Detection Is Genuinely Hard Now A 2025 study found that GPT-4.5 passed as a human 73% of the time in structured Turing test conditions — but only when it was instructed to play a casual, imperfect human persona; without that instruction, the rate was 36%. That's exactly the kind of setup that makes a chatfishing bot convincing. Scientific American has described chatfishing as "the modern Turing test" — a scenario where telling human from machine in conversation is no longer easy or obvious. In a dating context, where warmth and attention feel meaningful and you want the connection to be real, the difficulty is even higher. This is also why the gut feeling that something is off is worth taking seriously, even when you can't name exactly what it is. The signals described here often register before you can articulate them. ## What to Do When You're Suspicious **Stop sharing personal information right now.** Don't give out your phone number, home address, workplace, or anything financial until you've resolved the doubt. This costs you nothing if the person is real; it protects you if they're not. **Try the tests together.** A video call request, a spontaneous specific selfie, and a local-event question — all three at once are hard to fake simultaneously. Real people clear them easily. AI-driven conversations can't clear all three. **Report it.** If you're on a paid app and believe you've been matched with a platform bot, report it to the app's support team. In the US, you can also file a complaint at reportfraud.ftc.gov. **Trust the mismatch.** If the conversation feels emotionally warm but nothing in it is grounded in real, specific detail — no inconvenience, no specifics, no evidence of an actual life behind the messages — that mismatch is the signal itself. One scope note: if things have moved into money requests, emergency stories, or anything financial, that goes beyond detecting AI into romance scam territory. The guide on [AI romance scams on dating apps](/guides/ai-romance-scam-dating-app) covers that situation specifically. ## What to Try Next If you want to apply the same kind of detection to writing in general, [How to Tell If Something Was Written by AI](/guides/how-to-tell-if-text-is-ai) walks through the signals across any kind of text. And if you've been on a video call and wanted to know whether the face on screen was real, [How to Spot a Deepfake Video Call](/guides/verify-video-call-deepfake) covers exactly that test. ## Sources - [Large Language Models Pass the Turing Test](https://arxiv.org/abs/2503.23674) — Jones and Bergen, UC San Diego, 2025 - [Love, Lies, and Language Models: Investigating AI's Role in Romance-Baiting Scams](https://arxiv.org/abs/2512.16280) — Gressel et al., USENIX Security 2026 - [Match Group, Inc.](https://www.ftc.gov/legal-library/browse/cases-proceedings/172-3013-match-group-inc) — Federal Trade Commission - [Federal Trade Commission v. Match Group Inc — docket, N.D. Texas 3:19-cv-02281](https://www.courtlistener.com/docket/16244264/federal-trade-commission-v-match-group-inc/) — CourtListener / Free Law Project - [Match Group Agrees to Pay $14 Million, Permanently Stop Deceptive Advertising, Cancellation, and Billing Practices to Resolve FTC Charges](https://www.ftc.gov/news-events/news/press-releases/2025/08/match-group-agrees-pay-14-million-permanently-stop-deceptive-advertising-cancellation-billing) — Federal Trade Commission, August 2025 - [Report Fraud](https://reportfraud.ftc.gov) — Federal Trade Commission - [What Norton's Insights Reveal About AI Dating Scams](https://us.norton.com/blog/online-scams/ai-online-dating-scams) — Norton (Gen Digital) - [Romance Reimagined: How AI is Playing Cupid (and Catfish)](https://newsroom.gendigital.com/2025-02-04-Romance-Reimagined-How-AI-is-Playing-Cupid-and-Catfish) — Gen Digital newsroom --- # Someone Made a Deepfake of You: What to Do First Source: https://aiforregulars.com/guides/deepfake-of-me-what-to-do Finding out that someone has made a fake AI photo or video of you — sexual, embarrassing, or just designed to make you look like you said or did something you didn't — is frightening, and the instinct to panic or immediately confront whoever posted it is completely normal. But the order you do things in matters. Acting fast on evidence, not emotion, gives you the best chance of getting it removed and, if needed, holding someone accountable. Before you report, block, or confront anyone, capture what exists right now: screenshots of the image or video, the post's URL, the username or account that posted it, the date and time, and screenshots of any comments, shares, or messages tied to it. If you can, save or download the actual file — you will need a copy of it for some of the removal tools below. Content gets deleted, accounts get renamed, and posts move — evidence you didn't save in the first hour is often gone by the next day. If the fake shows you nude, partially nude, or in a sexual situation, you have a fast, private removal path. **If you are 18 or older**, go to [StopNCII.org](https://stopncii.org). It turns the image or video on your own device into a digital fingerprint called a hash — the actual file is never uploaded or seen by anyone — and shares that fingerprint with participating platforms, including Facebook, Instagram, TikTok, Threads, X, Snapchat, Bing, and OnlyFans, so they can find and remove matching content. This works on AI-generated deepfakes as well as real images. **If you are under 18**, use [NCMEC's Take It Down](https://takeitdown.ncmec.org) instead — it works the same way and is built specifically for minors. Alongside a hash-matching tool, report the specific post through the platform's own reporting flow — look for options like "non-consensual intimate image," "impersonation," or "synthetic or manipulated media" rather than a generic complaint. In the United States, the federal TAKE IT DOWN Act now requires platforms to remove nonconsensual intimate images — including AI-generated deepfakes — within 48 hours of a valid request, with penalties of over $53,000 per violation for platforms that don't comply. If a U.S. platform ignores a valid takedown request, you can file a complaint at the FTC's [TakeItDown.ftc.gov](https://takeitdown.ftc.gov). Even after a platform takes content down, it can still show up in search results or on sites that copied it. Google has a dedicated removal request for nonconsensual explicit or fake sexual imagery at [Remove explicit or intimate content](https://support.google.com/accounts/answer/17045517) — it also proactively filters similar results and removes duplicate copies it finds once one image is confirmed and removed. The instinct almost everyone has is to message the person who posted it and demand they take it down, or to argue with them in the comments. Resist that. It rarely works, it can tip them off to delete evidence you haven't saved yet, and on some platforms your own angry replies can get flagged instead of theirs. Screenshot first, report through the proper channel second, and let the platform's policy — not a back-and-forth — do the work. A fake video putting words in your mouth, a manipulated photo used to embarrass or scam someone, or an AI clip impersonating you for a prank or a con doesn't fall under the sexual-image tools above, but it's still a policy violation on nearly every major platform. Report it under "impersonation" or "synthetic and manipulated media" — most platforms including Meta, TikTok, and YouTube have specific categories for this. If people who know you may have seen it — coworkers, family, a school, an employer — a short, calm message letting them know it's fake and that you've reported it does more to limit damage than staying silent while it spreads. If a message comes with it — "pay me" or "send more or I'll post this" — that's sextortion, a crime, not a negotiation. Don't pay and don't send anything further; compliance rarely ends the demands and often invites more. Save every message and file as evidence, then report it to your local police and, if you're in the U.S., to the FBI's [IC3.gov](https://www.ic3.gov). Being targeted this way is not something you caused, and it's not something you have to handle by yourself. ## What to Try Next If you want to get better at spotting fakes before they fool you or someone you love, [how to spot a deepfake video](/guides/how-to-spot-deepfake-video) covers the visual signs, and [is that really your family on the video call](/guides/verify-video-call-deepfake) walks through verifying a live call. If the fake involves a cloned voice rather than an image, [how AI voice cloning scams work](/guides/ai-voice-cloning-scams) explains how those are made and reported. ## Sources - [FTC Begins Enforcing the TAKE IT DOWN Act](https://www.ftc.gov/news-events/news/press-releases/2026/05/ftc-begins-enforcing-take-it-down-act) — Federal Trade Commission - [Take It Down Act enforcement starts now: what to know about the FTC and TIDA](https://www.ftc.gov/business-guidance/blog/2026/05/take-it-down-act-enforcement-starts-now-what-know-about-ftc-tida) — Federal Trade Commission - [How StopNCII.org Works](https://stopncii.org/how-it-works/) — StopNCII.org - [Take It Down — Frequently Asked Questions](https://takeitdown.ncmec.org/faq/) — National Center for Missing & Exploited Children - [Ask Google to remove explicit or intimate content](https://support.google.com/accounts/answer/17045517) — Google Account Help --- # How to Tell If a Phone Call Voice Is AI (Not a Real Person) Source: https://aiforregulars.com/guides/detect-ai-voice-phone-call AI-generated voices have gotten good enough that they can fool people — at least for the first few seconds of a phone call. Scammers use them to impersonate bank representatives, family members in distress, or government officials. Knowing what to listen for gives you a real edge. I want to be honest with you: these clues are fading as the technology improves, and on today's best cloning tools, researchers say listening alone performs barely better than a coin flip. Your best protection is a verification process, not just your ears. But the clues below are still worth knowing, because most scam calls today are not using the very best voice technology — they're using cheap, fast, good-enough imitations that still have tells. Here's what to do from the moment you pick up. The first moments of a call are when AI voices give themselves away most clearly. Before you say anything beyond "hello," just listen. Notice how the caller starts. Real people begin with natural hesitations, brief acknowledgments, or small talk ("Oh good, I caught you" or "Hey, is this a good time?"). AI-generated calls often launch immediately into the reason for the call with a smooth, rehearsed opener — no natural warm-up. Don't feel you have to respond right away. A brief silence after you say hello is normal, and it can prompt a scripted system to reveal itself. While the caller is talking, run through this list mentally. You don't need to check off all six — two or three together should prompt you to test further. **1. Too-smooth rhythm.** Real speech has natural variations in pace — people speed up when excited and slow down when thinking. Cheaper AI voice tools still default to a consistent, slightly flat rhythm — but this is one of the first things premium cloning tools have fixed, so treat it as a supporting clue, not proof either way. **2. Absence of breath sounds.** This used to be one of the more reliable tells: natural speech includes faint inhales between sentences that AI voices used to skip. It no longer is — several current voice-cloning tools now insert artificial breath sounds specifically to sound more human, so don't lean on this one. **3. Unnatural pauses before answers.** When you ask something unexpected, a live AI system sometimes needs a beat to process before responding, creating a pause a real person wouldn't have. This still works reasonably well against real-time conversational scams, though it tells you nothing if the caller is just playing back a cloned voice reading a pre-written script. **4. Over-precise pronunciation.** AI voices tend to pronounce every word fully and evenly, while real speech has reduced vowels, swallowed syllables, and regional accent quirks. This gap is closing as cloning tools train on more casual, natural-sounding speech, so treat it as one clue among several, not a reliable test on its own. **5. Clipping at sentence edges.** Cheaper or older AI voice generators occasionally cut off the first or last syllable of a sentence when compressing audio for a phone call. Better-funded tools have mostly eliminated this artifact, so its absence doesn't mean the voice is real. **6. No background noise whatsoever.** Real calls from real people usually have some ambient sound — room echo, a TV in another room, traffic, keyboard clicks. Complete silence behind a voice is still a reasonable flag against today's typical low-effort scam call, though a more sophisticated scammer can add fake background noise on purpose. This is the most effective single test you can do. Ask something specific that a real person in the claimed role would know — but that a prepared script would not have anticipated. If the caller claims to be your bank: *"Can you tell me which branch I normally visit?"* or *"What was the last transaction I disputed with you?"* If the caller claims to be a family member: *"What's the name of the restaurant we went to for my birthday last year?"* or *"What did you get me for Christmas?"* A real person will answer directly or say they don't remember. An AI script will typically give a generic deflection, pause oddly, or loop back to its main purpose. If your family has set up a safe word — a specific word or short phrase only your family members know — now is the time to use it. Simply ask: *"What's our word?"* No legitimate caller who is actually your family member will be confused or offended by this question. A scammer won't know the word. This is the single most reliable verification tool for calls that claim to be from someone you love. If you haven't set one up yet, [Family Safe Words](/guides/family-safe-words) explains how to pick one that's memorable and hard for an outsider to guess. I once got a call that I was almost certain was a real person — the voice sounded warm and natural with what seemed like genuine nervousness. But something felt slightly rehearsed. When I asked which city the caller was in (the script had said "local branch"), the answer didn't match any branch I'd used. I hung up. When I checked the number later, it was flagged as a known scam line. The lesson: trust the nagging feeling, not just what your ears tell you in the moment. If anything about the call felt off — the voice, the request, the urgency — hang up. Then: 1. Find the person's or organization's number yourself — from your contacts, from their official website, or from the back of a card you have 2. Call that number directly 3. Do not call back using any number the suspicious caller gave you — that number may also be controlled by scammers This single step protects you from nearly all voice-based scam attempts. A real organization will always be reachable through their official channels. Hanging up is not rude — it's the right move. If you've confirmed or strongly suspect the call was AI-generated and fraudulent: - In the US, file a report at [reportfraud.ftc.gov](https://reportfraud.ftc.gov) - Outside the US: in England, Wales, and Northern Ireland, report to [Report Fraud](https://www.reportfraud.police.uk/), the national reporting centre (this is the service that used to be called Action Fraud), reachable online or on 0300 123 2040 — in Scotland, report to Police Scotland on 101 instead; in Canada contact the [Canadian Anti-Fraud Centre](https://www.antifraudcentre-centreantifraude.ca/report-signalez-eng.htm); in Australia report to [Scamwatch](https://www.scamwatch.gov.au/report-a-scam) - Report the number to your carrier — most have a short code or app feature for reporting spam calls - Block the number on your phone so it doesn't get through again Don't feel embarrassed if you weren't certain during the call. These systems are designed to be convincing. What matters is what you do before making any payment or sharing any personal information. ## What to Try Next If you'd like protection that starts before the call even reaches you, [AI Tools That Block Scam Calls Before They Start](/guides/ai-scam-call-blocker) walks through the built-in tools on iPhone and Android that most people haven't turned on yet. And if you're concerned about an older relative who gets frequent calls, [The Grandparent Scam 2.0](/guides/grandparent-scam-ai) explains the specific tactics being used and has a family safety plan you can put in place together. ## Sources - ['Grandparent' Scams Get More Sophisticated](https://www.fcc.gov/consumers/scam-alert/grandparent-scams-get-more-sophisticated) — Federal Communications Commission - [Scammers use AI to enhance their family emergency schemes](https://consumer.ftc.gov/consumer-alerts/2023/03/scammers-use-ai-enhance-their-family-emergency-schemes) — Federal Trade Commission - [Artificial Imposters: Cybercriminals Turn to AI Voice Cloning for a New Breed of Scam](https://www.mcafee.com/blogs/privacy-identity-protection/artificial-imposters-cybercriminals-turn-to-ai-voice-cloning-for-a-new-breed-of-scam/) — McAfee - [AI 'voice cloning' scams are on the rise. Here's how to protect yourself](https://www.cnn.com/2026/05/29/tech/ai-voice-cloning-scams-protect-yourself) — CNN --- # Should You Tell Employers You Used AI to Write Your Application? Source: https://aiforregulars.com/guides/disclose-ai-job-application AI is now part of how most people write — from fixing grammar to drafting full documents. Job applications are no different. The question of whether to tell employers about it comes up constantly, and the answer is less dramatic than most people expect. ## The short version Using AI to help write your resume or cover letter is not cheating, lying, or something you owe an employer a confession about — with two important exceptions: the content must be factually true, and the work must represent your actual abilities. If those two things are true, disclosure is usually a personal choice, not an ethical requirement. ## When disclosure helps you Some situations exist where mentioning AI use actually works in your favor. **When the job involves AI tools.** Many roles in marketing, communications, operations, and technology now expect employees to use AI regularly. If you used AI thoughtfully during your application — say, you ran your resume through multiple iterations to optimize it for ATS — that demonstrates exactly the kind of practical AI fluency employers want. Mentioning it briefly in a cover letter can position you ahead of candidates who say "I'm familiar with AI" without any evidence. **When you're asked directly.** Some companies include AI usage questions in their screening process. This is most common at tech companies, AI startups, and forward-thinking marketing teams. Answer honestly. A good answer shows you understand both the value and the limits of AI tools. **When it explains something unusual.** If your resume is in a format or style noticeably different from your previous applications on file, a simple line acknowledging that you refreshed it with AI assistance can prevent confusion. ## When disclosure hurts you — or doesn't matter **Most general professional roles.** A hiring manager reviewing 200 resumes for an operations coordinator position is not thinking about whether you used AI. They want to know if you have relevant experience, if your application is clear, and if you seem like a strong candidate. Volunteering that you used AI adds no information and risks distraction. **Roles where writing is not the job.** If you're applying for an accounting, nursing, warehouse management, or sales role, the quality of your prose is a minor factor. Disclosure about your writing process is simply not relevant. **When the AI just helped you fix errors.** Running your cover letter through an AI grammar tool is functionally identical to using spell-check. No one discloses spell-check use. I've spoken with hiring managers across several industries about this, and the consistent answer was: "I don't care if they used AI — I care if they can do the job." The concern they actually have is about candidates whose written application doesn't match how they present in an interview. That's the thing worth worrying about. ## The line you shouldn't cross The ethical concern with AI in job applications is not about tools — it's about honesty. Two things cross the line: **Inventing qualifications.** AI tools sometimes embellish. If a resume builder adds skills you don't have, or your AI-generated cover letter claims an accomplishment you can't back up in an interview, that's a problem — not because AI was involved, but because the information is false. Read every line of AI output and delete anything that isn't true. **Hiding inability.** If a writing job asks for a writing sample and you submit AI-generated work as your own, you're misrepresenting a skill the employer is specifically evaluating. Similarly, if your interview requires you to demonstrate knowledge you don't actually have because AI wrote your application without your real understanding behind it, you're in a difficult position regardless of what you disclosed. ## A practical rule of thumb Ask yourself: if the hiring manager asked me about any part of this application in the interview, could I speak to it comfortably from memory? If yes, you're fine. If the application describes a version of yourself you'd struggle to represent in person, fix the application — not the disclosure question. ## What to try next If you're ready to build that application, the [comparison of AI resume builders](/guides/best-ai-resume-builders) walks through which tools work best for which situations. For cover letters specifically, the [ChatGPT cover letter guide](/guides/chatgpt-cover-letter) has prompts that produce first drafts you can actually edit into your own voice. --- # Duck.ai: Chat with AI for Free, Without an Account, Without Being Tracked Source: https://aiforregulars.com/guides/duckduckgo-ai-chat If you have ever wanted to ask an AI a question but felt uneasy about creating yet another account and handing over your email address, Duck.ai was built for exactly that situation. It is DuckDuckGo's free AI chat tool — no signup, no password, no personal information required. You open the site and start talking. ## What Makes Duck.ai Different Most AI chatbots — ChatGPT, Gemini, Claude — require you to create an account before you can use them. That links your conversations to your identity, which means they can be used to train future AI models, stored for years, and tied to a profile that knows your name and email address. Duck.ai works differently. DuckDuckGo acts as an intermediary between you and whichever AI model you choose. When you type a question, DuckDuckGo strips your IP address and personal identifiers before passing the message on to the AI provider. The provider — OpenAI, Anthropic, Mistral, or Tinfoil.sh depending on your model — receives your question, but without the information that would connect it back to you. By default, Duck.ai saves your recent chats locally on your device only — not on DuckDuckGo's servers — so they're waiting for you again the next time you open Duck.ai. You can turn chat history off or clear it at any time. DuckDuckGo also offers an optional Sync & Backup feature that keeps an encrypted copy of your chat history across your own devices, but it stays off until you switch it on yourself. ## How to Use Duck.ai Type **duck.ai** into your browser's address bar and press Enter. You do not need to download anything, install an extension, or create an account. The chat window loads immediately. At the top of the chat window you will see a model selector. Free users can choose from several options: - **Claude 4.5 Haiku** — fast responses, good for everyday questions - **GPT-5.4 nano** — OpenAI's lightest model, very fast - **GPT-5.4 mini** — OpenAI's capable and quick option - **Mistral Small 4** — a European-made open model - **gpt-oss-120b** — open-source model hosted by Tinfoil.sh, strong for explanations - **Gemma 4 31B** — open-source model hosted by Tinfoil.sh If you are not sure where to start, Claude 4.5 Haiku or GPT-5.4 mini are solid choices for general use. Paid DuckDuckGo subscribers also get access to stronger models. Just type normally — Duck.ai works like any other AI chatbot. Ask a question, read the answer, and follow up with more questions in the same conversation. There is no limit on back-and-forth within a single session. > "Explain to me in simple terms what a home equity loan is." > "What are some good questions to ask my doctor before starting a new medication?" > "Help me write a short thank-you note for a neighbor who watched my cat." If a response from one model feels too short or misses the point, you can switch to a different model and ask the same question again. Each model has slightly different strengths. gpt-oss-120b often gives more detailed explanations; Claude 4.5 Haiku tends to be more concise. You can switch at any point during the conversation. When you are finished, you can just close the browser tab. Duck.ai keeps your recent conversations saved locally on your device by default, so you'll find them again in the sidebar next time you open it. If you'd rather nothing was kept, open Settings and turn chat history off, or use the Fire Button to clear everything at once. If you want to start a fresh conversation while the tab is still open, look for a "New Chat" button. I use Duck.ai when I want to ask something I would rather not have attached to my Google or OpenAI account — researching a medical symptom, understanding a legal situation, or looking into a financial decision I have not shared with anyone yet. It is not that those companies would necessarily do anything harmful with the information, but the question feels easier to type when I know it is not going into a profile with my name on it. ## Who Is Duck.ai Best For? Duck.ai is a strong fit if you: - Want to try AI without handing over an email address - Are researching something private — a health concern, a financial question, a sensitive family situation — and do not want it linked to an account - Have read about AI privacy and want a tool that puts those values into practice - Just need a quick answer to a one-off question without setting up a new service It is less suited for tasks that need the AI to remember you between sessions, work with uploaded files or images, search the web, or do anything more than text-based conversation. For those, ChatGPT or Gemini with an account are better equipped. ## What to Watch Out For **Private from the provider does not mean invisible to everyone with access to your device.** By default, Duck.ai keeps your recent chats — not just your browsing history — saved locally on the device and browser profile you used. Anyone who can get to that browser, such as a family member, a housemate, or your employer if you are on a work computer, can open the sidebar and read them, the same as they could with any other saved browser data. Duck.ai shields your identity from OpenAI and Anthropic; it does not shield your conversation text from someone with physical or logged-in access to your device. If that matters to you, turn off chat history in Settings, or clear it with the Fire Button after sensitive conversations. **Local history is not the same as the AI remembering you.** Duck.ai's Recent Chats let you reopen a past conversation, but the model itself does not build a profile of you across separate chats the way a logged-in assistant's memory feature can. If you want the AI to carry context from one topic into another automatically, you still need to restate it — for example, "Last time we established that I am looking for a part-time remote job in marketing, and I have ten years of experience." **Free models have real limits.** Claude 4.5 Haiku and GPT-5.4 mini are capable, but they are not the most powerful models available. If you need deep analysis, long documents summarized, or nuanced writing feedback, the paid versions of ChatGPT or Claude will outperform what you get here for free. **Still read the privacy policy.** DuckDuckGo's privacy commitments are public, readable, and better than most. But if you are handling genuinely sensitive information — legal, medical, or financial matters with real stakes — a conversation with an actual professional remains irreplaceable. Duck.ai can help you prepare smart questions to ask; it cannot replace the person who answers them. ## What to Try Next Knowing what Duck.ai keeps private is only half the picture — [What You Should Never Tell an AI Chatbot](/guides/what-not-to-tell-ai-chatbot) covers the information you should hold back even from private tools. If you are already using ChatGPT and want to tighten up its privacy settings to bring it closer to Duck.ai's approach, [How to Change Your ChatGPT Privacy Settings](/guides/chatgpt-privacy-settings) walks you through every option. ## Sources - [How does Duck.ai protect your privacy?](https://duckduckgo.com/duckduckgo-help-pages/duckai/ai-chat-privacy) — DuckDuckGo - [Duck.ai Privacy Policy and Terms of Use](https://duckduckgo.com/duckai/privacy-terms) — DuckDuckGo - [Is my Duck.ai chat history private?](https://duckduckgo.com/duckduckgo-help-pages/duckai/recent-chats) — DuckDuckGo - [What AI chat models are available?](https://duckduckgo.com/duckduckgo-help-pages/duckai/chat-models) — DuckDuckGo - [DuckDuckGo Sync & Backup](https://duckduckgo.com/duckduckgo-help-pages/sync-and-backup) — DuckDuckGo - [Duck.ai](https://duck.ai) — DuckDuckGo --- # The EU AI Act Explained: What Rights You Actually Get as a Regular User Source: https://aiforregulars.com/guides/eu-ai-act-your-rights The EU AI Act is the world's first major law specifically about artificial intelligence. It covers every AI product sold or used in the European Union — which affects hundreds of millions of people. If you have ever chatted with a bot, seen AI-generated content online, or had a loan or job application processed automatically, this law is meant to protect you. Here is what it actually says, in plain English. ## The Big Idea: Not All AI Is Equally Risky The law sorts AI systems into four categories based on how much harm they could cause. Think of it like food-safety ratings — a hospital kitchen and a home kitchen face very different rules. **Unacceptable risk — banned outright.** These are AI uses the law considers too dangerous to allow at all. Examples include systems that secretly manipulate people against their own interests, or that score people's behaviour to restrict their rights (sometimes called "social scoring"). These bans came into force on 2 February 2025. **High risk — allowed, but tightly controlled.** This covers AI that affects important parts of your life: a system that screens your job application, assesses your loan, manages your medical records, or decides on your access to public services. These systems are not banned, but companies using them must follow strict rules — about accuracy, documentation, and your right to know what is happening. **Limited risk — must label itself.** Chatbots and AI-generated images or videos must tell you they are AI. You should not have to guess whether you are talking to a person or a machine, or whether a photo is real or synthesised. This requirement is part of the rules taking effect progressively. **Minimal risk — mostly free to use.** Spam filters, video game AI, and most recommendation engines fall here. The law leaves these largely alone. ## "You Must Be Told When You Are Talking to an AI" One of the most concrete things the EU AI Act does for regular people is the labelling requirement. If you are chatting with an AI assistant on a company's website, that company must make it clear the assistant is AI — not a human employee. If a social media post, a video, or an advertisement was substantially created by AI, it must be marked as such. This matters because AI-generated media — images, voices, video — can be very realistic. The law's goal is that you always have the basic facts to make up your own mind. ## Your Right to Know — and to Push Back — on Automated Decisions Here is where the EU AI Act most directly touches your everyday life. If a high-risk AI system makes or contributes to an important decision about you — say, rejecting your mortgage application, screening you out of a job, or assessing your eligibility for a benefit — you are entitled to: - **A meaningful explanation.** Not just "the system said no," but information about what factors were considered. - **Human review.** You can ask for a human being to look at the decision, rather than simply accepting the automated outcome. This builds on rights you may already have under GDPR, and the two laws are designed to work alongside each other. When I read through the Act's consumer-facing provisions, the human-review right stood out most. Many people accept an automated rejection letter and never question it. This law makes the right to push back explicit and legally enforceable — which is a genuine shift from how most automated systems have operated until now. ## How the AI Act and GDPR Work Together You may have heard of GDPR — the EU's data-protection law. GDPR is about your personal data: who holds it, why, and your rights to see or delete it. The AI Act works alongside GDPR, not instead of it. GDPR still covers all the data questions. The AI Act adds a new layer: rules about the AI systems themselves — how they must be designed, tested, and documented, and what they are forbidden to do. Think of it this way: GDPR protects your data. The AI Act protects you from what AI does with that data. If you have not yet explored your GDPR rights — including the right to request your data or opt out of certain processing — our [guide to AI and your data under GDPR](/guides/ai-gdpr-your-data) covers those steps clearly. ## The Timeline: What Is Already in Force and What Is Coming The EU AI Act entered into force on 1 August 2024. It is being phased in over time: - **2 February 2025** — Bans on unacceptable-risk AI took effect. Social scoring and subliminal manipulation systems are prohibited. - **2 August 2025** — Rules for general-purpose AI models (the large AI systems that power many products) apply. - **2 August 2026** — Most rules for high-risk AI systems and the main consumer-facing protections — transparency requirements and human-review rights — become enforceable. If you are reading this in 2025 or early 2026, the most significant protections for ordinary users are not yet fully in force. They are coming — but they are not here in full yet. ## What to Watch Out For The EU AI Act is a real law with real obligations. But enforcement is still taking shape, and some details are still being worked out by regulators. A few things to keep in mind: - **The timeline is set, but implementation takes time.** Companies are still adapting their systems. Do not assume every AI tool you encounter is already following all the rules. - **Enforcement runs through national authorities.** If you think your rights are being violated, you would contact your country's designated AI supervisory authority. The process is not instant, and complaint systems are still being established in many EU member states. - **The law covers AI used in the EU.** If you are dealing with a company headquartered outside the EU that targets EU users, the law still applies in principle — but enforcement can be more complex. - **Check current guidance.** Because rules are phasing in and details are still settling, always check an up-to-date official source before relying on a specific right. The European Commission publishes updates at its official website. The best mindset right now: know that these rights are coming, understand the protections in principle, and stay alert to news as 2026 approaches. ## If an AI Decision Has Already Affected You Even before the AI Act is fully in force, you already have rights. GDPR gives you the ability to challenge automated decisions that have a legal or similarly significant effect on you. If you were rejected for a loan, an apartment, or a job by an automated system, you can already ask for a human to review that decision. Our guide on [how to appeal an AI decision](/guides/ai-decision-appeal-rights) walks through exactly how to do that, step by step. ## What to Try Next The EU AI Act protects you from how AI is used. GDPR protects you from what is done with your data. Understanding both gives you the full picture: - [What AI companies know about you and how GDPR protects your data](/guides/ai-gdpr-your-data) — your data rights in plain English. - [How to appeal an AI decision that affected you](/guides/ai-decision-appeal-rights) — a practical step-by-step walkthrough. --- # European Alternatives to ChatGPT: Mistral, DeepL Write, and Aleph Alpha Explained Source: https://aiforregulars.com/guides/european-ai-alternatives A lot of people in Europe have started asking a simple question: is there an AI tool that works well AND is made in Europe? The concern is understandable. In a Bitkom study published in February 2026, 68% of Germans said their country depends on the US and China for AI, and 69% said they would prefer an AI provider based in Germany. Many people across Europe feel similarly. The good news is that European AI tools do exist — and some of them are genuinely excellent for specific tasks. The honest answer, though, is that they are generally smaller and less capable than ChatGPT or Gemini for open-ended questions. This guide walks you through the main options, what each one is actually good at, and how to choose. ## The European Options at a Glance Here are the three main tools worth knowing about as of mid-2026: - **Mistral / Le Chat** (France) — a general-purpose chatbot, the closest European equivalent to ChatGPT - **DeepL and DeepL Write** (Germany) — the best European tools for translation and writing improvement - **Aleph Alpha** (Germany) — an AI platform for businesses and governments, not for everyday consumers All three are based in the EU and therefore subject to GDPR. That is a real benefit — it means you have legal rights over your data. But it does not automatically make them "more private" in practice. More on that in the "What to Watch Out For" section below. ## Mistral and Le Chat — The Closest European ChatGPT Mistral is a French AI company founded in 2023 and based in Paris. It has become the most prominent European player in general AI. Its research and models are widely respected internationally. For everyday users, the entry point is **Le Chat** — Mistral's free web-based assistant, available at chat.mistral.ai. You create a free account, and then you can chat with it just as you would with ChatGPT. It can answer questions, help you draft text, summarise documents, and explain things. Le Chat supports multiple languages including French, English, German, Spanish, and others. It performs well on writing tasks and handles European languages better than some US tools. > Try this: "Write a short, polite email to my energy provider asking them to explain my last bill in plain language." There is a free tier and paid plans with more features and higher limits. The free version is enough to get a clear sense of whether Le Chat suits you. One thing to know: Mistral is building fast, and Le Chat is improving quickly. It is not as well-known as ChatGPT, which means fewer tutorials and third-party guides — but it is a real, capable tool. I tried Le Chat for drafting a formal letter in French and was impressed. The tone was natural and it felt less American than ChatGPT's output. For anything involving French language specifically, it's worth a try before defaulting to a US tool. ## DeepL and DeepL Write — Built for Language DeepL is a German company based in Cologne, and it has been quietly beating Google Translate on European language pairs for years. If you have ever tried to translate something into or out of German, Polish, Dutch, or French and found the result stilted or odd, DeepL is worth trying. Many professional translators use it as a starting point. **DeepL Translate** is free to use on the web (with character limits per translation). You paste your text, choose your languages, and get a translation. You can also click on any translated word to see alternatives. Paid plans allow larger documents and more usage. **DeepL Write** is a separate tool. Instead of translating, it improves your writing. Paste in a sentence or paragraph and it will fix grammar, improve word choices, and adjust tone. It currently works in English, German, French, Spanish, Italian, and Portuguese. > Try this in DeepL Write: "I am writting to ask about the staus of my application and wonder if you have recieved all the documents I sended last week." DeepL Write will clean up the spelling errors and improve the phrasing. It is narrower than ChatGPT — you cannot ask it questions or have a conversation — but within its lane it is one of the best writing tools available in any language. ## Aleph Alpha — European AI for Businesses, Not You Aleph Alpha is a German AI company based in Heidelberg. It is worth mentioning because it represents the most ambitious attempt to build sovereign AI infrastructure in Europe — AI that runs entirely within EU borders, on EU servers, under EU rules. However, Aleph Alpha's products are aimed at enterprises, government agencies, and large organisations — not individual people. You cannot sign up and start chatting with it. If you work for a hospital, a government department, or a large company that handles sensitive data and needs AI it can control completely, Aleph Alpha is relevant. For everyone else, it is more a name to recognise than a tool to use. In 2026, Aleph Alpha announced it was merging with Cohere. Worth noting: Cohere is a Canadian company, not European, even though it often appears alongside European AI discussions. ## Which Should You Pick? Here is a simple steer based on what you want to do: **For translating text** → Use DeepL. It is the best translation tool of the lot, European or otherwise, for most European language pairs. **For improving your writing** → Use DeepL Write. Excellent for fixing grammar and polishing text in European languages. **For general questions, drafting, and chatting** with a European-made tool → Try Le Chat (Mistral). It is free, capable, and genuinely competitive for everyday use. **For business or government use with full EU data control** → Aleph Alpha is the name to investigate, but it requires direct contact with their sales team. **For most everyday tasks in general** → ChatGPT, Gemini, and Claude are still more capable and have more features. There is no shame in using them if they serve you well. Being European is a nice bonus, not a reason to use a less suitable tool. ## What to Watch Out For **"European" does not automatically mean "more private."** Where a company is based tells you which laws apply. Under GDPR, European companies must let you access and delete your data, and they face real penalties for mishandling it. That is meaningful. But actual privacy in daily use depends on what you have agreed to: does the service use your conversations to train its models? Can you opt out? Where is your data physically stored? Check the privacy settings of any tool — European or otherwise — before sharing sensitive information. **Pricing and features change frequently.** All three tools have changed their plans, limits, and included features multiple times. Any specific details about what is free or paid in this article should be verified against the provider's current pricing page before you make decisions. **Smaller companies can pivot or be acquired.** The European AI scene is consolidating. A tool you rely on today might be part of a different company — or discontinued — in a year. It is worth keeping an eye on the news for any tool you make a regular part of your workflow. **Do not share sensitive personal data with any AI tool** — address, national insurance number, bank details, medical records — regardless of where it is made. No chatbot, European or American, is the right place for that information. ## What to Try Next If you want to compare the major free AI chatbots side by side, see [Free AI Chatbots Compared: What You Get Without Paying](/guides/free-ai-chatbots-compared). If you are not sure which chatbot to start with as a complete beginner, [Best AI Chatbot for Beginners](/guides/best-ai-chatbot-beginners) gives a plain-English recommendation for getting started with confidence. ## Sources - [Künstliche Intelligenz in Deutschland: Perspektiven aus Bevölkerung und Unternehmen (PDF)](https://www.bitkom.org/sites/main/files/2026-02/bitkom-studienbericht-ki.pdf) — Bitkom - [Künstliche Intelligenz in Deutschland](https://www.bitkom.org/Bitkom/Publikationen/Kuenstliche-Intelligenz-in-Deutschland) — Bitkom --- # How to Explain AI to a Skeptical Parent or Grandparent Source: https://aiforregulars.com/guides/explain-ai-to-elderly-family Explaining AI to a skeptical older family member is different from explaining it to a coworker. The skepticism is usually not about the technology itself — it is about trust. They have seen enough promises from tech companies to be careful, and they are right to be. The goal of this guide is not to sell them on AI. It is to give them one honest, useful experience so they can decide for themselves. Before any demo, say something like: *"I want to show you something that's actually useful. Before I do, let me clear up a few things it's not — because there's a lot of hype and fear around it, and most of it is wrong."* Then cover the three biggest misconceptions: - **It is not a person.** There is no one on the other end. It is software that has processed enormous amounts of text and learned to respond in a natural way. - **It does not know who you are.** Unless you type your name or personal details, it has no idea who it is talking to. You can use it without identifying yourself at all. - **It is not always right.** It sounds confident even when it is wrong. Anything important needs to be verified — the same as something you would read on the internet. Give them time to ask questions before you open any app. These three points answer most of the concerns older adults bring up, and raising them yourself builds more trust than waiting for the objections to come. The single biggest mistake in first AI demos is starting with something too abstract or too personal. Start with something low-stakes and immediately useful to them specifically. Good first questions for older adults: - "What are some gentle exercises I can do if I have knee pain?" - "Can you suggest a simple recipe using chicken, rice, and whatever vegetables I have?" - "What's a good book for someone who loved [a book they mention]?" - "Explain what [medical term from a recent appointment] means in plain English." Avoid anything involving finances, passwords, family situations, or health information they would not want anyone else to see. The goal is a useful answer to a safe question — something that makes them think "I could actually use this." Sit next to them, not across from them. Let them type if they are comfortable, or type together with them watching. Do not take over the keyboard. The 15-minute session outline: 1. Open ChatGPT, Gemini, or Copilot (whichever is easiest on their device) — 2 minutes 2. Ask the first question you prepared together — 3 minutes 3. Read the answer together and talk about it — 5 minutes 4. Let them ask one question of their own choosing — 5 minutes That last step is the most important. Their question, their curiosity, their control. If you script the whole session, it feels like a sales demonstration. If they choose what to ask, it becomes their discovery. Some explanations land better than others for older adults who did not grow up with the internet. These tend to work well: **The encyclopedia analogy:** "It's like a very fast encyclopedia you can have a conversation with. But just like an encyclopedia, it can have outdated or wrong information, so you check the important stuff." **The well-read friend analogy:** "Imagine a friend who has read everything — every book, newspaper, and website — and can answer your questions in plain language. Smart, but not infallible. You'd still call a doctor about your health." **The autocomplete analogy:** "You know how your phone suggests the next word when you're texting? This is the same idea, but scaled up enormously — it suggests whole paragraphs and ideas." Avoid saying it "thinks" or "understands." That framing tends to either make it sound more human than it is (which later leads to disappointment) or more threatening than it is. "It generates responses based on patterns" is more accurate and usually less unsettling once they hear it. End the session with two honest statements: First: "There will be times it gives you a wrong answer. That doesn't mean it's broken — it just means you verify, the same way you'd check anything important." Second: "If something ever feels off — if it asks for personal information, if an answer seems strange — just close the tab. You don't owe it anything. You can always start a new conversation." Then follow up a week later. Did they try it again? What did they use it for? What confused them? That check-in matters more than the first session. It shows that you introduced them to something real, not just a party trick, and that you are available if they run into something unexpected. My mother-in-law used AI for the first time to look up a Czech idiom she half-remembered from childhood. The AI gave her three versions, explained the origins, and she spent twenty minutes just reading about the history of the phrase. That was several months ago and she uses it almost every day now. The key was that the first question was entirely hers. ## What to Try Next If your family member is ready to go deeper on their own, [the beginner's guide to AI for seniors](/guides/ai-for-seniors) is written for independent reading — clear, no jargon, practical from the first paragraph. And if they want a structured first hour with ChatGPT specifically, [the ChatGPT first-hour guide for seniors](/guides/chatgpt-for-seniors-first-hour) walks through setup and the first few uses step by step. --- # Fake Customer Service Bots: How Scammers Impersonate Real Brands Source: https://aiforregulars.com/guides/fake-customer-service-scam You search for your airline's customer service number. The first result has the airline's logo and a phone number with real-looking formatting. You call. A helpful agent asks for your booking reference — and your full card number "to look up your reservation." This is a fake customer service scam, and it is one of the most effective fraud techniques in use today. AI makes the agents more conversational, the websites harder to distinguish, and the whole operation easier to run at scale. ## How Fake Support Gets in Front of You The scam usually reaches victims through one of three paths: **Fake search results.** Scammers buy paid search ads using real brand names as keywords. When you search "Comcast customer service number" or "Amazon refund help," a paid listing with a fake phone number can appear above the actual company's result. These ads appear at the top of results, grouped under a "Sponsored results" label on both Google and Bing — a label many users scroll past without registering it — and the fraud phone number can sit above the real company's own listing. **Fake chat widgets on lookalike sites.** Scammers build websites that closely resemble real brand support pages — a similar URL, a copied logo, a matching color scheme. A chat widget appears within seconds, staffed by an AI bot or a human handler trained to extract your credentials. **Social media impersonation.** Fake accounts on X (formerly Twitter), Facebook, and Instagram pose as official brand support accounts. They watch for users posting public complaints about a brand and reach out offering to help through direct message. ## What Fake Agents Actually Want The conversation usually starts helpfully — they want to verify your identity. Then the requests get more specific: - Your account password ("so I can look into the issue for you") - A one-time code sent to your phone — this is how they bypass two-factor authentication - Remote access to your computer via a program like AnyDesk or TeamViewer - Your credit card number "to process a refund" No legitimate company support agent will ever ask for your password or a one-time code sent to you. Real refunds don't require you to provide card numbers — the company already has your payment information on file. ## How to Find Real Customer Support **Always start from the official website, navigated by you.** Type the company's address directly into your browser or use a bookmark you created yourself. Find the contact page from there. Never use a phone number or chat link from an email, a search ad, or a social media message you didn't expect. **Ignore the first paid search result.** When searching for a company's support contact, scroll past any sponsored or ad listings to find the organic result below. Or simply go directly to the company's known web address. **Verify social media accounts.** Real brand accounts are usually verified with a checkmark and have a large follower count built over years. If a support account messaged you first after you posted a complaint, be very skeptical — real brand support teams almost never initiate outbound direct messages. **Hang up on pressure.** If a support agent you called tries to rush you, asks for your password, or suggests downloading remote-access software, end the call. Real support agents don't do any of these things. I once nearly called a fake tech support number I found in a search result. The listing had the company name, the right-looking phone format, and appeared completely official. What saved me was that I happened to already have the company's app open, so I used the in-app support option instead. Now I treat in-app support and direct-navigation as my default — never a search-result phone number. ## If You Think You Contacted a Fake Agent Act quickly. Change the password for the affected account on the real company's website. If you gave a one-time code, that session may already be compromised — log out all devices from the account's security settings if that option is available. If you installed remote-access software, uninstall it immediately and run an antivirus scan. Contact your bank if payment details were shared. Report the scam to the FTC at reportfraud.ftc.gov and to the real company — they want to know their brand is being impersonated and can sometimes get fraudulent search ads taken down quickly. ## What to Try Next Many fake customer service scams begin with a phishing email that sends you to the fake site in the first place. [How to spot AI phishing emails](/guides/spot-ai-phishing-emails) will help you intercept that first step before you ever land on a fake page. For a look at how phone-based scams use convincing AI-cloned voices on top of this playbook, [AI voice cloning scams](/guides/ai-voice-cloning-scams) covers what the technology sounds like and how to recognize it. ## Sources - [Scammers and "customer service" — another imposter scam](https://consumer.ftc.gov/consumer-alerts/2020/08/scammers-customer-service-another-imposter-scam) — Federal Trade Commission - [How To Spot, Avoid, and Report Tech Support Scams](https://consumer.ftc.gov/articles/how-spot-avoid-and-report-tech-support-scams) — Federal Trade Commission - [Scammers Using Computer-Technical Support Impersonation Scams to Target Victims and Conduct Wire Transfers](https://www.ic3.gov/PSA/2022/PSA221110) — FBI Internet Crime Complaint Center (IC3) --- # I Was Falsely Accused of Using AI: What to Do (Students' Guide) Source: https://aiforregulars.com/guides/falsely-accused-ai *Note: The story below is a composite based on common experiences students report with AI detection disputes. Names and details are fictional.* --- Maya, a high school junior, spent three weeks on her history essay. She outlined it by hand, wrote three full drafts, and even asked her older sister to read it before she turned it in. A week later, her teacher called her in and said the paper had flagged as AI-generated. Maya felt sick. She hadn't used AI at all — not even to brainstorm. But the detector said otherwise, and her teacher was taking it seriously. What happened to Maya is happening to students across the country. AI detectors are not reliable. They flag human writing — especially careful, formal writing — as AI-generated on a regular basis. That doesn't make the situation less stressful, but it does mean you have a real case to make. Here's how. ## Step 1: Stay calm and ask for specifics The worst thing you can do is panic, argue, or immediately send an angry email. Take a breath first. Then ask your teacher or administrator one specific question: "Can you show me the exact detector result you received, including the tool used and the percentage score?" This matters for two reasons. First, it gives you something concrete to respond to. Second, it signals that you intend to address this seriously rather than just deny it. Some accusations quietly go away at this step because the teacher realizes the evidence is thin. Write down the date and time of this conversation and what was said. ## Step 2: Gather your writing process evidence This is the most important step. AI detectors produce a single number — they cannot show your process. You can. **Look for:** - **Version history in Google Docs or Word.** Google Docs keeps every saved version with timestamps. Go to File > Version history > See version history. Screenshot this and save it. - **Early drafts or outlines.** Even a handwritten outline scanned with your phone is powerful evidence. It shows thinking that happened before the final text. - **Research notes, browser history, or bookmarks** from when you were working on the assignment. - **Emails or messages where you discussed the assignment** — a text to a friend asking for their opinion, an email to a teacher asking a question about the topic. - **The timestamp on the document file itself** — this shows when it was created and last edited. If your school uses a learning management system like Canvas or Schoology, check whether it logged your submission time and any prior saves. When I looked into how AI detectors actually work, I found that some of them flag writing as AI-generated simply because it's clear and grammatically correct. Non-native English speakers, students who write formally, and anyone who edits their work carefully tends to get flagged more often. The detectors see "too polished" and call it suspicious. That's a serious flaw, and it's worth saying out loud when you defend yourself. ## Step 3: Understand why detectors fail — and explain it calmly You don't need a computer science degree to make this point. Here's how to explain it in plain terms: AI detectors work by looking for patterns in word choices and sentence structure. They compare your writing to what AI-generated text tends to look like. The problem is that careful human writers — students who revise, students who write formally, students who are not writing in their first language — write in patterns that overlap with AI patterns. The result is false positives. Researchers who tested major AI detectors on essays written by real students found that a meaningful percentage of clean human writing gets flagged. This is documented and widely reported. You can reference the AI for Regulars guide on [how AI detectors are tested](/guides/ai-detectors-tested) if you want to show your teacher a plain-language source on detector reliability. ## Step 4: Request a formal meeting and bring your evidence If the informal conversation doesn't resolve it, request a formal meeting in writing (email is fine — it creates a record). Ask that a parent or guardian be present if possible. At the meeting: - Present your process evidence: version history, drafts, notes. - Calmly state that you understand the teacher's concern, but that you did not use AI. - Ask what the school's formal process is for resolving academic integrity disputes. - Ask whether the detector result alone is sufficient evidence under school policy — in most cases, it is not. Keep your tone factual, not defensive. "Here is evidence of my writing process" is more persuasive than "I would never do that." ## Step 5: Escalate if needed — and know your options If the meeting doesn't resolve the situation and discipline is being pursued, you have options. Most schools have a written academic integrity policy. Ask for a copy and read it carefully. Specifically look for what constitutes evidence of AI use and what the process is for appealing a finding. If the school is a public school in the US, you have procedural rights under school disciplinary policies. A parent can request that any decision be reviewed through the school's formal grievance process before any punishment is applied. If discipline is applied based solely on a detector result with no other supporting evidence, that is worth challenging — in writing, through the school's official appeal process. ## Protecting yourself going forward Even if this situation resolves in your favor, it's worth changing a few habits to protect yourself from future false accusations: - Write major assignments in Google Docs, where version history is automatic. - Keep your research notes, even rough ones. - Send yourself early drafts by email — a timestamped email to yourself is hard to argue with. - If you do use AI for any part of your process (brainstorming, grammar checking, research), note exactly what you did and check your school's policy on AI assistance. ## What to try next Understanding how AI detectors work — and why they fail — is genuinely useful knowledge. The guide on [AI detectors tested](/guides/ai-detectors-tested) walks through the major tools in detail. If you're a parent trying to help a child navigate this situation, [how to tell if text is AI-generated](/guides/how-to-tell-if-text-is-ai) explains the real signals (and the fake ones) that teachers and software look for. --- # Safe Words for Families: The 5-Minute Setup That Beats Voice Scams Source: https://aiforregulars.com/guides/family-safe-words Voice scams are fast, emotional, and built to prevent you from thinking clearly. A safe word flips that dynamic. Instead of trying to decide in real time whether a voice sounds right, you ask one simple question that only a real family member can answer. Here is how to set one up in about five minutes. Pick a short phrase you would never use in everyday conversation — two to four words that are genuinely odd together. Things like "orange typewriter," "blue harbor seal," or "Grandma's toolbox" work well. Avoid pet names, family nicknames, or anything that might come up naturally in a normal conversation. The word does not need to be clever. It just needs to be specific, private, and easy to remember under stress. Shorter and more concrete is better than elaborate. Scammers who learn about safe words will sometimes try to work around them. They may say "I can't talk freely right now" or "there are other people in the room." Agree in advance that this does not matter — the rule still applies. If a caller cannot say the safe word for any reason, you hang up and call your family member directly on a number you already have. There is no version of the rule where not saying the word is acceptable. Close that loophole before it can be exploited. Tell each family member who might receive or make an emergency call. Do this in person or in a direct, private one-on-one message — not in a family group chat, not in an email thread, and not anywhere a screenshot could easily be shared. If you have a family member who lives far away, a brief private phone call where you both repeat the word back to each other is the most secure method. Confirm that everyone has it and knows what it is for. The rule is simple: if someone calls claiming to be a family member in an emergency and cannot say the safe word when asked, hang up and call that person directly on a number already saved in your contacts. No exceptions. No matter how urgent the story sounds. No matter how convincing the voice is. Say this rule out loud with everyone together. Scammers specifically craft stories designed to make you feel that the usual rules should not apply — "don't tell anyone yet," "I need help in the next five minutes." Agreeing on the rule in advance is what makes it hold up in those moments. Setting up the safe word in my family took less time than I expected. The harder part was convincing everyone that the rule needed to be firm — no exceptions. Once I played a short clip from a news story demonstrating what a cloned voice actually sounds like, everyone understood immediately. The demonstration made the rule feel real rather than hypothetical. If you can, find a short audio demo online and play it during your family conversation. Within a day or two of setting the word up, make one short call where someone asks for the safe word and someone else says it correctly. This step is easy to skip, but it matters. Practicing even once makes the interaction feel natural and confirms that everyone actually has the right word. The first time you need the safe word in a real situation is not the moment to be second-guessing whether you remembered it correctly. If a caller cannot say the word, hang up calmly. Then call your family member back on a number you already have saved — never use any number the caller provided. If your family member picks up immediately and is completely fine, report the scam call to the FTC at reportfraud.ftc.gov. If you cannot reach your family member and begin to worry it could be real, call a second person who would know — another relative, a close friend — before doing anything involving money. The key is that money never moves based on a single unexpected call. Verify first. You do not need to rotate the safe word regularly like a password. Only change it if someone outside your immediate family learned it through an overheard conversation, a screenshot, or a mistake in how it was shared. Keep a physical note of the current word somewhere only your family can access — inside a kitchen cabinet, a home office drawer, or a physical address book. Do not store it in a shared cloud document or a group chat message. **What to try next:** This setup protects against suspicious phone calls. The same AI technology also appears in fake videos — see [How to Spot a Deepfake Video](/guides/how-to-spot-deepfake-video) to learn what to watch for there. And for the full picture of how voice scams are built, [AI Voice Cloning Scams](/guides/ai-voice-cloning-scams) explains how just a few seconds of audio is enough to fake a voice you love. ## Sources - [Scammers use AI to enhance their family emergency schemes](https://consumer.ftc.gov/consumer-alerts/2023/03/scammers-use-ai-enhance-their-family-emergency-schemes) — Federal Trade Commission - [Family Emergency Scams](https://consumer.ftc.gov/all-scams/family-emergency-scams) — Federal Trade Commission - [Artificial Imposters: Cybercriminals Turn to AI Voice Cloning for a New Breed of Scam](https://www.mcafee.com/blogs/privacy-identity-protection/artificial-imposters-cybercriminals-turn-to-ai-voice-cloning-for-a-new-breed-of-scam/) — McAfee - [Scammers Target Older Adult Victims](https://www.fbi.gov/news/stories/scammers-target-older-adult-victims) — Federal Bureau of Investigation --- # Your Kids Want You on ChatGPT — How to Respond on Your Own Terms Source: https://aiforregulars.com/guides/family-wants-me-to-use-ai Somebody in your family — a grown child, a grandchild, a neighbor who's very enthusiastic about things — has been suggesting you try ChatGPT. "You'd love it," they say. Or: "It's so easy." Or: "Just try it once." And maybe you're a little curious, maybe you're skeptical, maybe you're just tired of the topic. All of those reactions are completely reasonable. This guide is written for you — not for the person nudging you. It won't tell you that you're missing out, or that you need to get on board with something new. It will tell you honestly what AI is useful for, what the real concerns are, and how to decide for yourself — including deciding that you'd rather not bother at all. ## Why So Many People Feel the Same Way You Do If you're not enthusiastic, you're in good company. An AARP survey found that AI use among adults 50 and older nearly doubled between 2024 and 2025 — from around 18% to 30% — but that still means the majority of people in that group aren't using it. The same research found the top barriers were privacy concerns, distrust of the technology, and worry that AI would reduce real human contact. In fact, 68% of respondents said they were concerned that AI would replace meaningful human interaction. Those aren't irrational fears. They're questions that researchers, technology ethicists, and policymakers are actively working through. Being skeptical doesn't mean you're behind. It might mean you're paying attention. There's also research suggesting that family pressure to adopt technology creates a particular problem: many older adults go along with something out of embarrassment or because they don't want to seem out of touch, then quietly stop using it when no one's watching. That's a frustrating outcome for everyone involved. This guide is an attempt at something different — enough honest information to make a real choice, and the words to say it out loud. ## You're the One Who Gets to Decide What You Try Your family may be framing AI as one big thing you either adopt or don't. It isn't. ChatGPT is one tool made by one company. There are dozens of others. You don't have to take on a category; you can ask whether any specific tool solves a specific problem you actually have. A more useful question than "should I try AI?" is this: **Is there anything in my week that takes longer than I'd like, or something I find confusing or annoying, where a little help would feel like genuine relief?** If something comes to mind — read on. If nothing comes to mind when you sit with that question, there may not be a reason to bother. That's a reasonable answer, and you don't need to apologize for it. ## Three Things AI Is Actually Useful For Right Now These aren't impressive-sounding demonstrations. They're tasks that match problems people actually run into. **Reading a confusing letter.** Insurance explanations, Medicare notices, legal documents, HOA rules — these are written in language that almost no one finds easy. You can paste the confusing paragraph directly into ChatGPT and ask it to explain what it means in plain English. This is one of the things AI is genuinely good at. > "Can you explain what this paragraph means in plain English? I'm not sure what it's asking me to do." > *(Paste the confusing text before or after your question.)* **A recipe from what you have on hand.** If you're standing in front of the fridge wondering what to make for dinner, describe what you have and ask for suggestions. It won't always produce a winner, but it's faster than searching through cookbooks and gives you something to work with. > "I have chicken thighs, some potatoes, an onion, and half a lemon. What could I make for dinner tonight that's not too complicated?" **Wording something you're not sure about.** A complaint to a company that you want to be firm but polite. A thank-you note for someone who helped you through a hard time. A message to a neighbor about something awkward. If you know what you want to say but can't quite find the right words, describe the situation and ask for help. > "I need to write a short note to my neighbor about their dog getting into my garden. I don't want to be rude, but I do want it to stop. Can you help me write something?" The most useful thing I tell anyone trying AI for the first time: treat it like asking a knowledgeable friend who's always available and never makes you feel embarrassed for asking. If the first answer isn't right, say "that's not quite what I meant" and describe it again. It usually gets there within a try or two. ## What to Say to the People Nudging You If you're willing to try it but want company the first time, say so directly: > "I'm open to trying it, but I'd like you to sit with me while I do — so you can help if I get stuck." That's a completely reasonable request. Someone who genuinely wants to help you will say yes immediately. If they hand you a phone and walk away, that tells you something useful about the kind of support you were actually going to get. If you want a single demonstration before committing to anything: > "Show me one thing it's actually useful for — something I would personally use — and then let me decide." And if you've already decided you're not interested: > "I've heard enough to know it's not for me right now. I'll let you know if that changes." You don't owe anyone a long argument or a detailed explanation. You've been making decisions about which technologies are worth your time for decades, and you've gotten good at it. ## What to Watch Out For If you do decide to try it, a few things are worth knowing before you start. **Don't put in sensitive personal details.** Your social security number, bank account information, passwords, and full medical history should stay out of any AI chatbot, including ChatGPT. General questions are fine; personal identifying details are not. See [What Not to Tell an AI Chatbot](/guides/what-not-to-tell-ai-chatbot) for a clear rundown of what to keep to yourself. **AI makes confident mistakes.** If you ask about a medication dose, a legal deadline, or a specific historical fact, don't treat the answer as definitive. Use it as a starting point, then verify anything important with a professional or a reliable source. **It starts fresh every time.** AI has no memory of previous conversations unless you're using a specific paid feature that saves your history. Every new session starts blank — it won't remember what you told it last week. **It's not company.** If the concern your family has is that you might be lonely or isolated, AI won't solve that. A chatbot is useful for specific tasks. It is not the same as a conversation with someone who knows and genuinely cares about you. ## If You Decide It's Not for You You're allowed to say no. Completely, without needing to justify it further. If you try it and don't find it useful: "I tried it, it's not for me." If you try it and find it annoying or uncomfortable, that's also a complete answer. If you've decided before trying that you'd rather not, that is your decision to make. The technology will keep changing, and you can revisit that choice whenever you want — next month, next year, or never. Nobody else gets to set that timeline. *(If you have a family member who's been trying to figure out how to introduce AI to you without making it awkward, there's a companion guide written from their side of the conversation: [How to Explain AI to an Elderly Family Member Without Frustrating Each Other](/guides/explain-ai-to-elderly-family).)* ## What to Try Next If you've decided you want to explore further, [ChatGPT for Seniors: Your First Hour](/guides/chatgpt-for-seniors-first-hour) walks through a first session step by step, with no assumed knowledge. If you want to understand what the real privacy and safety concerns are before you start, [AI for Seniors](/guides/ai-for-seniors) covers the things the enthusiastic introductions tend to leave out. --- # Which AI Health Apps Are Actually Doctor-Approved? FDA-Cleared vs. Wellness Apps Source: https://aiforregulars.com/guides/fda-cleared-ai-health-apps Your phone's app store has thousands of apps that call themselves health tools. Some track your sleep, some suggest supplements, some will tell you what your chest pain probably means. The problem is that most of those apps are regulated the same way a fitness tracker or a meditation app is — which is to say, almost not at all. The small category of apps that have gone through a formal US government review process as medical devices is a different thing entirely, and knowing which category you are looking at could matter a lot. ## What "FDA-Cleared" Actually Means When a health app (or the device it runs on) is described as "FDA-cleared," it means the manufacturer submitted it to the US Food and Drug Administration as a Software as a Medical Device and the FDA concluded it was substantially equivalent to an already-approved medical device. This process is called 510(k) clearance. It is not a rubber stamp. To get cleared, a manufacturer has to demonstrate what the software actually does, submit clinical or technical evidence, and describe its intended use. The FDA can and does reject submissions or ask for more data. The FDA has cleared more than 1,300 AI-enabled medical devices so far. That number sounds large, but against the tens of thousands of health apps on the market, it is a small fraction. **What clearance means in practice:** An app with 510(k) clearance can legally claim it detects or helps manage a specific medical condition. The claim has a defined scope — the app is cleared for a specific intended use, not for health in general. ## What FDA Clearance Does Not Mean **It does not mean your doctor endorsed it.** Clearance is a regulatory decision, not a clinical recommendation. Your cardiologist might find the Apple Watch ECG app useful or might have reservations about how it applies to your particular situation — clearance does not settle that. **"FDA-approved" and "FDA-cleared" are not the same thing.** Approval (used for drugs and some high-risk devices) involves a more intensive review with randomized clinical trial data. Clearance (the 510(k) path) is used for lower-risk devices that are substantially similar to something already approved. Many news reports use the terms interchangeably; they mean different things. **Clearance covers the intended use, not everything the app might do.** If an app is cleared to detect atrial fibrillation and you use it to monitor a different condition, you are outside the cleared use case. ## FDA-Cleared AI Health Apps That Actually Exist Here are real examples so you know what this looks like in practice. **Apple Watch ECG app (Series 4 and later)** — FDA-cleared to generate a single-lead ECG and classify it as showing signs of atrial fibrillation (AFib) or sinus rhythm. This is a regulated medical function built into a consumer device. The app states upfront: it does not detect heart attacks or other conditions. **UpDoc V1.0** — Cleared by the FDA on December 23, 2025 and publicly launched on June 25, 2026, this became the first FDA-cleared AI medical device built on a large language model (LLM). It is designed to help adults with type 2 diabetes manage insulin dosing. The clearance is notable because it is the first time the FDA has cleared an LLM-based tool as a medical device — most cleared AI health tools use simpler pattern-recognition algorithms rather than a generative language model. Other cleared AI tools exist in areas like diabetic retinopathy screening, dermatology image analysis, and radiology — but most of those are used by clinicians rather than downloaded by patients on a phone. When I am not sure whether an app is actually cleared or just wellness-branded, I search the FDA's 510(k) database with the company name. If it is not in that database, it is not cleared as a medical device — and I treat any medical-sounding claims the same way I would treat a supplement label with no studies behind it. ## The Wellness App Zone: Health Claims Without FDA Clearance The vast majority of health apps are wellness apps. These include sleep trackers that give you a sleep score, apps that analyze your voice or face to estimate mood, symptom checkers that list possible conditions, weight-loss apps that predict health outcomes, and AI coaching apps that give dietary advice. None of these are regulated as medical devices. That does not make them useless — a well-designed sleep tracker or food log can be a genuinely helpful habit tool. But it does mean the company behind it has not had to demonstrate clinical accuracy to any regulator before putting it in your hands. In January 2026, the FDA relaxed its oversight rules for certain wearable wellness apps, allowing more products to avoid the device review process entirely. That makes the cleared versus not-cleared distinction even more important to understand yourself, because regulators are intentionally stepping back in the wellness space. ## Red Flags That a Wellness App Is Overstepping These signals suggest an app is making claims that belong in regulated medical territory. **It tells you what you have.** An app that says "you have sleep apnea" or "this looks like melanoma" is making a diagnostic claim. Wellness apps are not cleared to diagnose. **It recommends specific drug doses.** Suggesting you take more vitamin D is one thing; telling you to adjust your insulin or blood pressure medication dose is another. Only a cleared medical device — or a licensed clinician — should be doing the latter. **It uses "clinically proven" or "medically validated" without linking to actual evidence.** These phrases are often marketing language with nothing published behind them. **It asks for sensitive information it does not need.** An app that collects your full medical history or lab results but is not connected to any healthcare provider warrants caution about how that data is stored and shared. ## How to Check an App in Under a Minute 1. Go to fda.gov and search for "510(k) database," or go directly to accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm 2. Search by the company name or the device name as it appears in the app store 3. If there is a 510(k) entry in the results, the device has been cleared — click through to see the exact intended use it was reviewed for 4. If nothing comes up, the app is not cleared as a medical device The app itself may say "FDA-cleared" in its description or in the About section. That claim should be checkable in the database — if it is not there, the claim is questionable. ## Questions to Ask Your Doctor Before Trusting Any Health App The most useful thing you can do with a health app — cleared or not — is bring it up with your doctor before acting on what it tells you. **"Have you heard of this app, and do you think it is right for my situation?"** A cleared app is cleared for a general population; your doctor knows your specific history. **"If this app flags something unusual, how should I respond?"** Know the protocol before you are staring at an alert at 11pm that says your heart rhythm looks abnormal. **"Would the data from this app actually help you as my doctor?"** Some cleared apps export data clinicians can use. Others generate enough noise that it adds more confusion than clarity. Your doctor can tell you which situation applies. ## What to Watch Out For **Privacy.** Health data is sensitive, and even a cleared medical device app may share your data with third parties. Read the privacy policy, especially the section on whether data is sold or shared with advertising partners. **False reassurance.** An ECG app that shows "sinus rhythm" does not mean your heart is healthy overall — it means that specific reading, at that moment, did not show signs of AFib. These are narrow tools with specific claims, and that is actually how they are supposed to work. **Anxiety from continuous monitoring.** Regular health alerts can cause real worry, especially for people already managing a chronic condition. Cleared does not mean anxiety-free, and it is worth talking to your doctor about whether continuous monitoring makes sense for you. This guide is specifically about apps and their regulatory status. If you want to know how to handle AI chatbots that answer medical questions, see [How to Ask AI About Health Symptoms](/guides/ai-health-symptoms). If you are looking at AI-powered health hardware for an older family member, [The Best AI Health Gadgets for Seniors](/guides/ai-health-gadgets-seniors) covers that ground. ## What to Try Next If a health app is recommending a medication or supplement, [How to Use AI to Understand Your Prescription](/guides/ai-prescription-explainer) walks through what AI can help you understand and what still requires a pharmacist or doctor. Before you download any health app that asks for personal information, [Is That App Actually AI, or Just a Scam?](/guides/is-ai-app-scam) covers the basic safety check. --- # Free AI Chatbots Compared: What You Get from ChatGPT, Gemini, Claude, and Copilot Without Paying Source: https://aiforregulars.com/guides/free-ai-chatbots-compared You've probably heard of ChatGPT, Gemini, Claude, and Copilot. What's harder to know is what you actually get from each one if you never pay — and whether the free version is good enough for how you'd actually use it. The short answer is yes, for most everyday tasks. What takes a little more explaining is which free option fits your situation best. One important caveat up front: free tiers change frequently. The specifics in this guide were accurate as of July 2026, but usage limits, model access, and included features shift without much warning. Before choosing based on a specific feature, it's worth a quick check on each provider's pricing page. ## The Four Chatbots at a Glance All four chatbots are free to use with an account. All four can answer questions, help you write, brainstorm ideas, and explain complicated things in plain language. Where they differ is how generous the free tier is, which features are included, and whether the chatbot can search the internet. A note on scale: ChatGPT is still far and away the most-used chatbot, holding around 64.5% of the chatbot market compared to Gemini's 21.5%, according to SimilarWeb data from January 2026. That dominance means ChatGPT has the largest collection of third-party guides, YouTube tutorials, and shared prompts — which matters if you ever want to look something up or learn from someone else's workflow. ## ChatGPT Free — The Biggest Ecosystem **What's included:** Access to a capable model by default, with occasional access to the best available model when traffic is low. Web search is available on free but may be limited at peak times. Image generation is included with a small daily allowance. Voice mode is available. **What you don't get without paying:** Consistent access to the top model at all times, advanced data analysis, unlimited image generation, and longer context for processing large documents. Text conversations themselves are unlimited on free since August 2026 — the caps that remain are on images, file uploads, and access to the top model. **Best for:** People who want the widest compatibility — whether it's finding tutorials, working with tools built around ChatGPT, or just wanting the chatbot that most guides are written about. The free tier is surprisingly capable for general questions and writing tasks. > Try this: "Write a short, friendly email to cancel a dentist appointment and ask about getting my deposit back." If ChatGPT tells you it has hit the usage limit for its best model, don't stop — it usually switches to a slightly less capable model automatically, and that model is still perfectly good for most everyday tasks. The limit on the top model typically resets within a few hours. ## Gemini Free — The Most Generous Free Tier PCMag named Gemini the best free AI chatbot option in 2026, with the main reason being how much it includes without a subscription. **What's included:** Gemini searches the web in real time as part of every response — not as an optional add-on, but as the default. This means it can tell you what happened recently, look up current information, and give you up-to-date answers rather than relying only on what it was trained on. Image generation is available for free. Voice and image input are supported. **What you don't get without paying:** Access to the most powerful Gemini model, deeper integration with Google Workspace files like Docs and Drive, and the ability to process very long documents. Since October 9, 2026, free personal accounts get only the lightest model, Flash-Lite; Flash and Pro are no longer included without a paid plan ([what changed](/news/gemini-free-plus-model-access-cut-october-9)). **Best for:** People who regularly need current information — recent news, current prices, things that happened in the last few weeks — and want a free chatbot that doesn't feel artificially limited. Also a natural fit if you already use Google products, since Gemini is built into Gmail, Google Docs, and Android. > Try this: "What are the most common reasons health insurance claims get denied, and what should I do if mine is?" ## Claude Free — Best Writing Quality, Tightest Limits **What's included:** Access to one of Anthropic's Claude models on the free tier (Claude Sonnet 4.5 or 4.6 as of mid-2026). Web search is included on the free plan. Claude is consistently praised for the quality of its writing — letters, summaries, creative pieces, and explanations that actually sound like a thoughtful person wrote them. File uploads (images and documents) are supported, up to 20 files per chat. **What you don't get without paying:** Daily message limits are the tightest of the four chatbots here. No image generation. **Best for:** Writing tasks where quality matters more than speed or volume. If you're drafting something important — a complaint letter, a personal statement, a professional email — Claude's output tends to feel more considered than what the other free tiers produce. > Try this: "Help me write a respectful but firm letter to my landlord about a heating issue that hasn't been fixed for three weeks." ## Copilot Free — Strongest Inside Microsoft Apps Microsoft Copilot is built into Windows 11, Microsoft Edge, and Microsoft 365 apps — and that integration is what sets it apart on the free tier. **What's included:** Web search via Bing is standard in Copilot responses. Image generation is included. Inside Edge, Copilot can see and interact with what you're looking at — it can summarize the webpage you're reading or draft a reply in Outlook without you having to copy and paste anything. **What you don't get without paying:** Copilot in Word, Excel, PowerPoint, and OneNote now requires a paid Microsoft 365 Copilot license — Microsoft removed it from the free tier in April 2026. Copilot Pro also unlocks priority access to the fastest models and more image generation. **Best for:** People who already use Edge, Windows 11, or Outlook. If you spend your day in Outlook or Edge, Copilot's in-context help inside those apps is a genuine time-saver that the other chatbots can't replicate without manual copying and pasting. > Try this (in Microsoft Edge): Open a long article, click the Copilot icon in the sidebar, and ask: "Can you give me the three most important points from this page?" ## What to Watch Out For **Free tiers change, sometimes without notice.** Features that are free today may move to a paid plan tomorrow. This has happened with ChatGPT and other platforms over the past few years, and it is likely to happen again. If you're building a habit around a specific feature, bookmark the pricing page and check it occasionally. **Your conversations may be used for training.** On free tiers, most providers use conversation data to improve their models unless you opt out. The setting to turn this off is usually in the privacy or data controls section of your account. If you're sharing anything sensitive — medical details, financial documents, personal messages — check those settings first. **Free doesn't mean less capable for most people.** The main differences between free and paid are limits on how often you can use the best model, not fundamental differences in what the AI knows or can do. For everyday questions and routine writing, free is often indistinguishable from paid. **None of these chatbots are always right.** On any plan, AI chatbots can confidently state incorrect information. The more specific the claim — a date, a statistic, a medical fact — the more worth it is to verify from another source before acting on it. ## Which Free Tier Is Right for You? Here's a simple way to think about it: - **You need current information regularly** — start with Gemini. Real-time web search is built in and free. - **You want the biggest ecosystem and most tutorial support** — ChatGPT is the safe default for most people. - **Writing quality is your top priority** — Claude produces the most polished prose, even with its tighter limits. - **You use Edge or Outlook** — Copilot's in-context integration makes it the practical choice. If you're not sure, pick any one and use it for a week. All four are free to try, and you can use more than one — many people keep two or three bookmarked and switch depending on the task. ## What to Try Next If you're just getting started and want help picking the right AI chatbot overall — paid and free together — [Best AI Chatbots for Beginners](/guides/best-ai-chatbot-beginners) walks through beginner-friendly picks with less assumed knowledge. If you're specifically weighing whether to pay for ChatGPT, [Is ChatGPT Free? What You Get Without Paying](/guides/is-chatgpt-free) covers that plan in detail. --- # Best Free Writing Helpers: Grammarly Alternatives That Cost Nothing Source: https://aiforregulars.com/guides/free-grammarly-alternatives Grammarly is the name most people know, but its free plan catches only the basics — and the constant "upgrade to fix this" nudges can feel frustrating. The good news is that several free tools do the same job well, and one of them (ChatGPT) goes much further than Grammarly ever could. Here's a plain-language comparison, plus a sample paragraph with common writing errors so you can see what each tool typically catches. ## The test paragraph Here is a poorly written paragraph with common errors — and what each tool is documented to flag on text like this: > *"The report was wrote by our team last week. It contains alot of important informations about the sales figures, which is very high this quarter. We should of included more details but ran out of time so the report is somewhat short."* ## The free tools, compared | Tool | Best for | Works inside browser/Word? | Languages | Rewrites whole text? | |---|---|---|---|---| | LanguageTool | Grammar + style | No (web editor only — free plan) | 25+ | No | | Hemingway App | Clarity and readability | No (paste in) | English only | No | | ChatGPT (free) | Full rewrites + suggestions | No (paste in) | Most major languages | Yes | | Microsoft Editor | Office + Edge users | Yes (built-in) | Many | No | | ProWritingAid (free) | Detailed style reports | Limited | English | No | ## LanguageTool LanguageTool is the closest direct swap for Grammarly. It finds grammar errors, flags awkward phrasing, and works via its web editor. Its documented checks cover all the main errors in a paragraph like this — "was wrote" (should be "was written"), "alot" (two words), "informations" (not countable in English), and "should of" (should be "should have"). These are well within its published capabilities. **What's free:** Grammar and spelling via the web editor, support for 25+ languages, personal dictionary. **What's paid:** Style suggestions like passive voice overuse, advanced punctuation, and the ability to check longer documents are behind the paid plan. **Best for:** People who write in multiple languages, or anyone who wants an always-on grammar checker similar to Grammarly without the cost. ## Hemingway App Hemingway (hemingwayapp.com) takes a different approach. It doesn't just catch errors — it highlights sentences that are hard to read, flags passive voice, and scores your text by reading level. Paste your writing in and the screen lights up in color: yellow means a sentence is complex, red means it's very hard to follow. Hemingway App highlights exactly these patterns — it flags passive voice constructions and hard-to-read run-on sentences, both of which appear in the paragraph above. It won't fix grammar errors the way LanguageTool does, but it makes you a better writer over time by teaching you to see your own hard sentences. **Best for:** Emails, blog posts, and anything where plain, direct writing matters. Also great for anyone who wants to improve their own style, not just fix mistakes. ## ChatGPT (free tier) ChatGPT takes more effort to use — you paste your text in and type a request — but it's the most powerful option on this list. It doesn't just fix errors; it rewrites your text the way a human editor would. On the test paragraph, a simple prompt like this works well: ``` Please correct all grammar and spelling errors in this paragraph, and rewrite any awkward sentences in a clear, professional tone. Keep the meaning the same. ``` Reviewers and published comparisons consistently find that ChatGPT fixes every error in a paragraph like this and goes further — rephrasing vague constructions to something more formal, smoothing run-on sentences, and tightening the whole piece. The prompt above is a reliable starting point for this kind of request. **Best for:** Cover letters, emails you want to sound polished, rewriting a paragraph that isn't working, or checking writing when you're not sure what exactly is wrong. My personal setup is LanguageTool in the browser for everyday writing — it catches errors as I type without me thinking about it. When I need a full piece rewritten or I want to make an important email really shine, I paste it into ChatGPT. Together they cover everything the paid Grammarly plan does, for free. ## Microsoft Editor If you use Microsoft Word, Outlook, or the Edge browser, Microsoft Editor is already built in and free. It flags spelling, grammar, and basic style issues, and it integrates tightly with Office apps. It's not as thorough as LanguageTool on pure grammar, but if Microsoft products are your main writing environment, it's the path of least resistance. ## ProWritingAid (free tier) ProWritingAid has a powerful paid plan, and its free tier gives you a taste of the style analysis. The free version lets you run up to 500 words at a time through a web editor and shows you a summary report. For casual use it's fine; for regular use you'll hit the word limit quickly. ## How to choose - You write in more than one language → **LanguageTool** - You want to improve your writing style over time → **Hemingway App** - You need full rewrites or have irregular writing needs → **ChatGPT free tier** - You live in Microsoft Office → **Microsoft Editor** - You want deep reports on a specific document → **ProWritingAid free tier** for occasional use ## What to try next If you're using ChatGPT for writing help, you might be surprised how much more it can do. The guide on [using AI to write professional emails](/guides/write-professional-email-with-ai) has copy-paste templates for common workplace situations. If you're job hunting, [writing a cover letter with ChatGPT](/guides/chatgpt-cover-letter) walks through the whole process step by step. --- # Galaxy AI: The 5 Features Worth Using (and How to Turn Off the Rest) Source: https://aiforregulars.com/guides/galaxy-ai-features Samsung has packed a lot of AI features into recent Galaxy phones. Some are genuinely useful from day one. Others sound impressive in a store demo but add friction to things you could do faster without them. This guide covers the five Galaxy AI features that are worth your time, how to use each one, and — equally important — how to turn off what you do not want. Circle to Search lets you draw a circle around anything on your screen and instantly search it. You can circle a product in a photo, a word you do not recognize, a restaurant sign, a QR-style barcode, or a plant in an image. **How to use it:** Press and hold the home button or navigation bar at the bottom of your screen. A glow appears around the edges. Draw a circle, scribble, or tap anything on the screen. Search results appear without leaving the app you were in. **Why it is useful:** You stay in whatever you were doing — a message, a photo, a web article — while you look something up. No need to screenshot, open a browser, and type. **To turn it off:** Settings → Display → Navigation bar → Circle to Search → toggle off. Live Translate works during phone calls and translates both sides of the conversation in near real time. You speak in English, the other person hears a translated version, and their reply is translated back to you. **How to use it:** During a phone call, tap Call Assist on the call screen, then tap Live translate. Select your language and the other person's language. Both of you hear the call through the translator. **Also works for text:** Open the Samsung Keyboard in any app and look for the translation button. Type in English and it converts your text to another language before you send it. **Why it is useful:** Makes calls with family or service providers who speak a different language much less stressful. It is not perfect — accents and fast speech can trip it up — but it handles slow, clear conversation well. **To turn it off:** Settings → Galaxy AI → Call assist → Live translate → toggle off. Writing Assist (formerly Chat Assist) appears inside Samsung's Messages app and some other apps. It can rewrite what you typed in a more professional, casual, or polite tone with one tap. **How to use it:** Type your message in the Samsung Messages app or another supported app. Look for the Writing Assist or Chat Assist icon near the text field. Tap it and choose a tone: Professional, Casual, Polite, or options like #social and Emojify. A rewritten version appears. Accept it, edit it, or ignore it. **Why it is useful for some people:** If you are writing a message to a landlord, a school, or a doctor's office and want it to sound more formal without a lot of effort, this saves time. **Why some people skip it:** If you already write naturally, it adds an extra step. The rewrites can sound a little generic. **To turn it off:** Settings → Galaxy AI → Writing assist (or Chat assist) → toggle off. Note Assist works inside Samsung Notes. After you write or record a long note, it can summarize it, pull out action items, and format it more clearly. **How to use it:** Open Samsung Notes and create or open a note. Tap the Notes Assist icon (three stars) above the keyboard. Options appear: Auto format, Summarize, Spelling and grammar, Translate. **Where it shines:** Meeting notes, shopping lists that got long, recipes you jotted down quickly. It turns a messy paragraph into a clean bulleted list in one tap. **Less useful if:** You write very short notes, or you do not use Samsung Notes at all. **To turn it off:** You can leave Note Assist installed and simply not use it. If you want to remove the icon from the Notes toolbar, open Samsung Notes, tap the three-dot menu in the top right, go to Settings, and look for an AI or Note Assist option to turn it off. I used to ignore Samsung Notes entirely and kept everything in Google Keep. I switched over just to try Note Assist and now I use it for anything meeting-related. The summary feature alone saves me five minutes of re-reading after a long call. Your notes sync to your Samsung account, though — so if you prefer to keep notes in Google's ecosystem, stick with Keep. Generative Edit is in the Samsung Gallery app. It lets you circle an object in a photo — a photobomber, a trash can, a parked car — and erase it. The AI fills in the background behind it. **How to use it:** Open a photo in Gallery. Tap Edit, then tap the Photo Assist icon (a blue circle with stars). Choose AI Eraser to mark objects for removal. Circle or tap the thing you want to remove. Tap Erase. The AI removes the object and reconstructs the background. **Why it is useful:** It works surprisingly well on simple backgrounds like grass, sky, sand, or plain walls. For busy or complex backgrounds, the fill can look patchy. **Honest limitation:** Do not use Generative Edit to remove or add people in a way that misrepresents what happened in a photo. AI-edited photos shared as real can mislead people — a small but real consideration. **To turn it off:** There is no global toggle — it lives inside the Gallery edit screen and only appears when you choose it. Just do not tap it if you prefer unedited photos. ## How to turn off Galaxy AI features you do not want If you find Galaxy AI suggestions popping up in places you did not expect, you can manage everything from one place: Settings → Galaxy AI Here you will see a list of AI features with toggles. Turn off what you do not use. The phone works exactly the same without them — they are additions, not requirements. You can also go to Settings → Privacy → Permission Manager if you want to review which apps Galaxy AI features have access to and tighten permissions further. ## What to try next If you are curious whether the AI features on your phone are sharing your data in ways you are not comfortable with, the [ChatGPT privacy settings guide](/guides/chatgpt-privacy-settings) explains the kinds of settings to look for — many of the same principles apply to phone AI tools. And if you want to understand AI more broadly, the [beginner's guide to ChatGPT](/guides/what-is-chatgpt) is a plain-English starting point. ## Sources - [Use features with Galaxy AI on your Galaxy phone and tablet](https://www.samsung.com/us/support/answer/ANS10000753/) — Samsung - [Galaxy AI](https://www.samsung.com/us/galaxy-ai/) — Samsung --- # Samsung Galaxy AI vs Apple Intelligence: Which Phone Is Smarter for Everyday Use? Source: https://aiforregulars.com/guides/galaxy-ai-vs-apple-intelligence The ads make it look simple: your phone understands everything, translates conversations instantly, and fixes your photos automatically. The reality is a bit more nuanced — both Samsung Galaxy AI and Apple Intelligence have genuinely useful features, but they work differently, they're strong in different areas, and what matters most depends on how you actually use your phone. Here's a task-by-task look at what each one does, so you can decide for yourself. ## Editing a Photo: Generative Edit vs. Clean Up This is where the difference between the two platforms is easiest to see. **Samsung's Generative Edit** lets you tap on any object in a photo and erase it. The AI fills in the gap with what it thinks should be there — and for common backgrounds like grass, sky, or a plain wall, it usually looks convincing. You can also move objects around the frame, swap the sky behind a landscape, and stretch the edges of a photo to change its proportions. For a family photo where a stranger walked through the shot at the wrong moment, it's a genuinely impressive tool. **Apple's Clean Up** does something similar but stays in a narrower lane — it's designed for removing small distractions from photos, like a piece of litter on the ground or an object sitting on a table. It works reliably within that scope, but it doesn't handle large objects or complex background replacements the way Generative Edit does. **Winner: Samsung.** If photo editing matters to you, Galaxy AI's tools go further and handle more types of edits. ## Translating a Call in Real Time: The Feature Apple Doesn't Have This is the most significant gap between the two platforms for everyday users. **Samsung Live Translate** works during a regular phone call. You speak in your language, the other person hears a translated version of your words played through their phone, and you hear a translation of what they say — all in real time, without a separate app, without a conference call setup, without anything unusual on the other end. For anyone with family in another country, international work calls, or occasional calls to businesses in other languages, it's one of the most practically useful phone features in years. **Apple Intelligence does not have a live call translation feature.** Apple offers text translation in many apps and has a dedicated Translate app, but it doesn't intercept and translate a voice call the way Live Translate does. **Winner: Samsung, clearly.** This is a real functional gap, not a marketing difference. ## Summarizing Emails and Long Messages Both platforms offer text summarization, and this is where they're closest to each other. **Apple Intelligence** summarizes emails automatically and shows summaries in your notification shade, so you can see the gist of a long email before you open it. It also summarizes web pages in Safari and groups related notifications together. The integration feels natural — it happens inside the apps you already use, without you doing anything extra. **Samsung's Galaxy AI** summarizes messages, emails, and notes as well. It also offers Circle to Search, which lets you long-press the home button, draw a circle around anything on your screen — a word, a photo, a product — and get information about it immediately. This works across more surfaces, including video frames. **Winner: Roughly equal.** Apple's version is slightly more polished and automatic; Samsung covers more surfaces. Either will serve you well for basic summarization. On both phones, I've found AI summaries helpful for deciding whether to open something now or later — but not as a substitute for actually reading. Summaries occasionally miss the emotional tone of a message or misread sarcasm. For anything that matters, read the original. Treat summaries as a filter, not a replacement. ## Getting Help with Writing Both phones offer writing assistance — cleaning up a draft, adjusting tone, making text shorter or more formal. The difference is in how consistently available the tools are. **Apple's Writing Tools** appear almost everywhere you type on an iPhone. Highlight any text in a note, email, or message, and you can ask it to rewrite it, make it shorter, more professional, or more casual. Because this works system-wide, it's available in third-party apps too, not just Apple's own. **Samsung Galaxy AI** offers writing help inside Samsung's apps — Messages, Notes, and email — and integrates Google's Gemini AI for broader assistance. Chat Assist can adjust the tone of messages before you send them. The coverage is solid within Samsung's apps but less consistent across everything on your phone. **Winner: Apple, slightly.** Writing Tools are available more consistently across more apps on iPhone. ## The Privacy Difference: Apple's Biggest Advantage This one is less visible in daily use but genuinely matters. **Apple Intelligence processes most requests on your device**, using the chips inside your iPhone rather than sending your words and photos to a server elsewhere. For tasks that need more computing power than your phone can handle alone, Apple uses Private Cloud Compute — a system designed so that not even Apple employees can access your data during processing. Apple has published technical documentation and invited independent security researchers to verify these claims. Whether you fully trust that or not, the architecture is genuinely more privacy-protective than most alternatives. **Samsung's Galaxy AI sends more tasks to the cloud** by default — to Samsung's servers, and for Gemini features, to Google's infrastructure. This is common practice and not inherently unsafe, but more of your data leaves your device. Samsung is expanding its on-device AI capabilities, so this balance is shifting over time. **Winner: Apple.** If keeping your data on your device matters to you, Apple Intelligence's design is more aligned with that goal. ## What the Usage Numbers Show A survey reported by Sammy Fans in July 2026 found that Galaxy AI beats Apple Intelligence in user awareness and real-world usage — more Samsung users knew their phone had AI features, and more of them were actually using those features day-to-day compared to iPhone users with Apple Intelligence. That finding matters in a practical sense. A feature that's invisible to most users isn't actually helping anyone. Samsung has placed AI tools in more prominent spots and marketed them more visibly, and users are finding them. Apple's implementation is often more polished, but it may also be less discoverable to someone who isn't actively looking for it. ## What to Watch Out For **Both AIs make mistakes.** Confident-sounding answers can be wrong. Photo edits can look unnatural on complex scenes. Call translation can mishear words or add an awkward delay. Don't rely on these features for anything where being wrong has real consequences. **You need a supported phone.** Apple Intelligence requires iPhone 15 Pro or later (or any iPhone 16 or later) running iOS 18.1 or later. Galaxy AI needs a recent Galaxy S, Z, or A series phone. Older phones don't get these features built in. **Neither is a reason to switch platforms.** Samsung and Apple update their AI features continuously, and they regularly trade places in specific areas. The gap between the two platforms is not large enough to justify the cost and disruption of switching. Use the phone you have, explore what it can do, and let the competition keep improving both. ## What to Try Next For a deeper look at what Samsung's AI can do across your phone, [Galaxy AI: The 5 Features Worth Using (and How to Turn Off the Rest)](/guides/galaxy-ai-features) covers the full picture. If you're on iPhone and want to understand all of Apple's AI tools in one place, [Apple Intelligence Setup Guide: What It Does](/guides/apple-intelligence-guide) is where to start. --- # Gemini AI on Your Android Phone: A Plain-English Guide Source: https://aiforregulars.com/guides/gemini-android-guide If you have an Android phone, you probably have Google Gemini already available — you may just not know it yet. Gemini is Google's AI assistant, and it goes well beyond the old "OK Google" voice searches. It can have a real conversation, analyze photos you take, understand what is on your screen, and help you get through everyday tasks faster. This guide starts from scratch and walks you through setup, then shows you the most useful things Gemini can do in real daily use. On most recent Android phones, the Gemini app is either pre-installed or available for free in the Google Play Store. Search for "Gemini" in the Play Store and install it, or look for the app icon in your app drawer. When you open Gemini for the first time, it will ask you to sign in with your Google account. Use the same account you use for Gmail or YouTube. The setup takes about a minute. To get the most out of Gemini — including the ability to hold the home button and have it appear on any screen — you need to set it as your phone's default digital assistant. Go to your phone's **Settings → Apps → Default apps → Digital assistant app** and select Gemini. On Samsung phones, tap **Settings → Apps → Choose default apps → Digital assistant app**. Once set, holding the home button or swiping inward from a bottom corner will bring up Gemini. Open the Gemini app and tap the message bar at the bottom. Type a question or a request just like you would in a text message. Good first things to try: "Can you help me write a short thank-you note to my neighbor?" or "Explain what inflation means in simple terms." Gemini gives you a full, readable reply. You can tap the microphone icon instead of typing if speaking is easier. Tap the microphone icon in the Gemini app, or hold your phone's home button if Gemini is set as your default assistant. Speak your question naturally — you do not need any special trigger phrase. Gemini keeps the conversation going. After it answers, you can ask a follow-up without starting over: "What about for vegetarians?" after a recipe question, or "Make it shorter" after it drafts a message for you. In the Gemini chat, tap the **+** button on the left side of the input bar to open the attachment sheet. Select a photo from your gallery or take a new one. Then type a question about it. Useful examples: "What plant is this?" with a photo of a leaf, "Is this a real document or does it look fake?" with a scanned form, or simply "Describe what is in this photo" if you want to understand something you photographed in a hurry. The screen context feature surprised me the most. I was reading a long restaurant menu on a website and held the home button. Gemini popped up and I asked "What on this menu is gluten-free?" — it read the screen and listed the options. I did not have to copy-paste anything. That kind of shortcut genuinely saves time. Hold your phone's home button (or use your phone's gesture shortcut) while something is on your screen — a webpage, an email, a photo, a map. Gemini appears as an overlay and can see what you are looking at. Ask it things like "Summarize this article," "Translate this text," or "What does this error message mean?" It reads the screen content so you do not have to copy anything. This works in most apps, though some apps can restrict screen reading — if Gemini does not respond to what is on screen in a particular app, that app may have blocked the feature. In the Gemini app, tap your profile icon and go to **Connected Apps**. Here you can allow Gemini to access Gmail, Google Calendar, Google Docs, and Google Maps. Once connected, you can ask things like "What meetings do I have tomorrow?" or "Find the email from my dentist last week." Gemini searches your actual data and summarizes what it finds, rather than making things up. ## Tips for better results Short, direct questions tend to work well. If Gemini gives you a confusing answer, try rephrasing with more context: instead of "help me write something," say "help me write a two-sentence reply to a friend who canceled our plans." Gemini can also make mistakes on facts, especially recent news. Treat its answers as a helpful starting point, not a final word. ## What to try next If you use Google's apps at work or school, [Gemini inside Google Docs and Gmail](/guides/gemini-google-docs-gmail) is worth exploring next. And if you are deciding between Google's AI and OpenAI's, our [ChatGPT vs. Gemini comparison](/guides/chatgpt-vs-gemini) breaks down the differences clearly. --- # Google Gemini Gems: Build a Personal AI Helper for Cooking, Fitness, or Anything Else Source: https://aiforregulars.com/guides/gemini-gems-guide > **Update, October 1, 2026:** Google is replacing Gems with "Skills", which work the same way and are free for everyone 18+ with a personal Google account. Your existing Gems are converted automatically. On personal accounts Gems end in November 2026; work and business accounts follow in March 2027 and school accounts in June 2027. Menus may look different depending on your account while the rollout continues. See [what's changing](/news/gemini-skills-replace-gems-free). If you've ever started a Gemini conversation by typing the same long explanation — your dietary rules, your job title, the tone you want — you already know the problem a Gem is designed to fix. A Gem stores that context once. Every conversation you open with it starts from the same briefing, automatically. You stop explaining and start asking. ## What Is a Gem? A Gem is a saved setup inside Gemini: a short name, a set of instructions you write, and optionally some files you upload. Think of it as a version of Gemini that's been briefed on your situation before you type a word. Gems don't replace regular Gemini — they sit alongside it. Open a plain Gemini chat and it starts fresh every time. Open a Gem, and it already knows you cook for a family of four, avoid nuts, prefer grams over cups, and want weeknight-friendly meals. You just describe what's in the fridge and ask what to make. Gems are free on all Gemini plans, including the free tier. Google Workspace organizations can also share Gems with colleagues, which makes them useful beyond personal use. ## The Premade Gems Worth Trying First Gemini comes with several ready-made Gems you can use immediately, without any setup. **Brainstormer** helps when you need more ideas than you can generate alone — for a gift, a project name, a travel itinerary, or a fundraising idea. It's designed to push past the obvious and get you to a longer list. **Career guide** handles job search questions: how to present a career change on a resume, how to prepare for a particular type of interview, how to turn a list of tasks into a strong bullet point. **Coding partner** helps when you're working on code and want a programming assistant that can explain concepts, help debug a problem, review a snippet, or walk you through an unfamiliar language or framework. **Learning coach** works well for anyone who wants to understand something rather than just get an answer. Ask it to explain a concept and it will check your understanding by asking questions back. **Writing editor** reviews text you paste in and suggests improvements in plain language — useful for emails, appeals, cover letters, or anything you want to sound more polished before you send it. To find these, look for the Gems section in the left sidebar when you're in Gemini. ## How to Create Your Own Gem Trying a premade Gem requires no setup. Building your own takes about five minutes, and the payoff starts the second time you use it. Sign in to Gemini at gemini.google.com. On the left side, look for a "Gems" section in the sidebar. You may need to scroll down or expand a menu to find it. Inside the Gems panel, you'll see the premade Gems plus a button to create your own. Click "New Gem" or the "+" button to open the Gem editor. Choose a name you'll recognize at a glance: "Sunday Cook," "Letter Helper," "Fitness Check-In," "Project Tracker." Make it specific enough that you can tell your Gems apart without opening each one. This is where you describe the role you want Gemini to play and the rules it should follow. You don't need formal language — clear, direct sentences work best. Think of it as briefing a new assistant: explain your situation, your preferences, and anything they should always remember. See the worked examples below for what good instructions look like. If you have a document the Gem should always reference — a recipe list, a house style guide, a template letter, an address block — you can upload it here. The Gem will have access to it in every conversation. Click Save. Your Gem now appears in the sidebar. Click it to open a new conversation — it already knows everything you wrote in the instructions, so you can start talking about your actual question straight away. When I write Gem instructions, I imagine I'm briefing a capable assistant on their first day with me. I spell out what I want, what I don't want, and the rules that matter most. If I later find myself re-explaining something in a Gem conversation that the Gem should already know, I go back and add it to the instructions. ## Two Worked Examples ### The "Sunday Cook" Gem This Gem handles weekly meal planning without requiring you to re-explain the same dietary constraints every week. **Instructions to paste in:** > You are my weekly meal planning assistant. I cook for two adults. We avoid all nuts and shellfish. I prefer metric measurements (grams, milliliters, Celsius). Meals should take 30 to 45 minutes of active cooking. I like Mediterranean and East Asian flavors, but variety is welcome. When I say "plan this week," suggest five dinners and a shopping list grouped by category. Flag any ingredient that needs advance preparation. Once the Gem is saved, you can open it any Sunday and say: "I have chicken, zucchini, and two cans of chickpeas — plan three meals." No setup, no re-explaining. ### The "Letter Helper" Gem This Gem drafts formal letters, complaints, or requests — things you don't write often enough to have a ready template for. **Instructions to paste in:** > You help me write formal letters and emails in English. Use a professional but warm tone — not stiff or overly legal-sounding. My name is [Your Name] and my address is [Your Address]. When I describe what I need to say, draft the full letter: greeting, body, and closing. If I don't specify a recipient, leave a placeholder. Ask me one clarifying question before you start if anything is unclear. Replace the bracketed details with your own. You can also upload a file with your common recipients' addresses as a Knowledge file so the Gem can fill them in. ## When a Gem Beats a Plain Chat A Gem earns its keep when you find yourself typing the same preamble over and over. Good fits include: - **A recurring creative project** — you're writing a novel and keep briefing Gemini on the world, the characters, and the rules each time. - **A consistent professional task** — you draft weekly reports in the same format for the same audience. - **A personal specialty** — you're a nurse and want health questions answered at a clinical level, not a general-audience one. - **Household rules** — dietary restrictions, home renovation measurements, your car make and model for maintenance questions. A Gem is probably overkill when you have a one-off question, you're exploring a topic fresh and don't yet know what context matters, or the task varies so much each time that standing instructions wouldn't help. ## What to Watch Out For **Gems don't verify facts.** A cooking Gem will confidently suggest a substitution that doesn't work in practice. Test before serving. **Instructions can fade over a long conversation.** As a conversation grows very long, Gemini may start weighting the Gem's instructions less. If it stops following your rules, open a fresh conversation in the same Gem. **File uploads have limits.** Very long documents may not be fully processed. If your Gem isn't referencing an uploaded file, try summarizing the key points in the instructions instead. **Shared Gems expose their instructions.** If you share a Gem within a Workspace organization, anyone who uses it can see your instructions. Don't include passwords, private addresses, or confidential details. **Gemini is still an AI.** For health, legal, financial, or safety decisions, treat a Gem's output as a starting point, not a final answer. It's a well-read assistant — helpful, but worth a second opinion when the stakes are high. ## What to Try Next If you want to get more from the instructions you write for your Gems, [How to Write Better AI Prompts](/guides/how-to-write-prompts) has the practical techniques that transfer directly. If you're still deciding whether Gemini or ChatGPT is the right fit for you, [ChatGPT vs. Gemini: Which AI Assistant Is Right for You?](/guides/chatgpt-vs-gemini) has the comparison you need. --- # Gemini in Gmail and Google Docs: 8 Things to Try Right Now Source: https://aiforregulars.com/guides/gemini-google-docs-gmail Google has quietly added Gemini AI to the apps you likely already use every day — Gmail and Google Docs. Most people haven't explored what it can do. Here are eight things worth trying, starting with the ones that are available on free accounts. In Gmail, look for a small star or sparkle icon inside the compose window, or a **Gemini** button on the right side of your inbox. In Google Docs, look for the same icon near the top of the page or in the right-side panel. Clicking it opens a chat panel where you can type requests in plain English. If you don't see it, your Google account settings or your organization's Workspace policy may have it turned off. Open any email thread that has gone back and forth several times. Click the Gemini icon and type: *"Summarize this thread for me."* Gemini reads the whole conversation and returns a short summary — who said what, what was decided, and what's still open. This is a real time-saver for long chains where you need to catch up fast. Open an email you need to respond to. In the compose window, click the Gemini icon or look for a **Help me write** option. Type what you want to say in rough terms, like: *"Thank them for the proposal and say we need two more weeks to review it."* Gemini writes a full, polished reply. You can accept it, edit it, or ask Gemini to adjust the tone — *"Make it shorter"* or *"Sound a bit warmer"* both work. Click **Compose** in Gmail. In the compose window, click the Gemini icon and describe the email you need. For example: *"Write a follow-up email to a client reminding them that their invoice is two weeks overdue, but keep it polite."* Gemini drafts the whole email. This works especially well for messages you dread writing — complaints, difficult requests, or anything where tone matters. I tested the "help me write" feature by asking Gemini to draft a cancellation request for a subscription service. It produced something clear and firm without sounding rude — exactly the balance I struggled to find myself. I sent it with one small edit. Open a long document — a report, a policy file, anything with several pages. Open the Gemini panel on the right and type: *"Give me a three-sentence summary of this document."* or *"What are the main action items?"* Gemini reads your document and answers based on what's actually in it, not just a guess. This is useful when you need to quickly understand a file someone sent you. You don't have to limit yourself to summaries. With the Gemini panel open on a Docs file, you can ask specific questions like: *"Does this document mention a deadline?"* or *"What does it say about the refund policy?"* Think of it as a search tool that understands meaning, not just keywords. Click **Help me write** at the top of a blank Google Doc (or use the Gemini panel) and describe what you need. For example: *"Write a one-page project proposal for a new employee onboarding process. Include a problem statement, goals, and three action steps."* Gemini writes a structured draft with headings and paragraphs. It won't be perfect, but it's a real starting point that usually saves 20–30 minutes of blank-page staring. Paste text you've written into the Gemini panel (or highlight it in Docs) and ask it to check your work. Try: *"Is this clear and professional? Suggest any improvements."* or *"Rewrite the second paragraph so it's easier to understand."* Gemini will point out awkward phrases, suggest clearer wording, and can rewrite sections on request. It's faster than waiting for a colleague to review a draft. ## What to try next If you want to get more out of any AI writing tool, the key is knowing how to phrase your requests. The guide on [how to write better AI prompts](/guides/how-to-write-prompts) explains the difference between a vague instruction and one that gets great results. If your workplace uses Microsoft tools, [Copilot in Word and Excel](/guides/copilot-word-excel-guide) covers the equivalent features in that ecosystem. --- # Gmail's New AI Inbox: How to Talk to Your Email Instead of Searching It Source: https://aiforregulars.com/guides/gmail-ai-inbox-guide Most people search their Gmail the same way they did ten years ago: type a keyword, scroll through results, hope the right email shows up. In January 2026, Google changed that. The new Gmail AI Inbox — powered by Gemini 3 — lets you ask your inbox questions the way you'd ask a person. You don't have to remember a subject line or a sender's name. You can just describe what you're looking for. Here is how each feature works and where to find it. ## What's free and what costs money Before getting into the steps, a quick overview. Conversational search, AI Overviews in search results, thread summaries, Help Me Write, and Suggested Replies are all free with any Google account. Two features require a paid Google One AI Premium subscription: the AI Inbox personalized daily briefing and Gmail Live (the voice interface, announced at Google I/O on May 19, 2026). Google turned on all AI features for users by default when the AI Inbox launched on January 8, 2026 — instructions for opting out are at the end of this guide. Click the search bar at the top of Gmail and type a question in plain English, the way you'd describe it to someone else. > What did my doctor's office send me last month? > Is there an email about my car insurance renewal? > Find emails with attachments from the plumber who gave me a quote Gmail's AI Overviews show a summary answer at the top of the results, followed by the matching emails underneath. You don't need to remember the sender's exact name or the subject line — the AI reads the content of your messages to find what you're asking about. Open any email thread that has grown beyond a few messages. At the top of the conversation, look for a "Summarize this email" chip or button, or a summary card that may appear automatically with a brief recap. Tap or click it to get a short summary of the whole exchange: who said what, what was decided, and what (if anything) still needs a response. This is particularly useful for long family group threads, work emails you were copied on, or any chain that grew while you were away. Instead of reading 20 messages in order, you read one paragraph and then decide whether you need to go deeper. Start composing a new email (the pencil icon or Compose button, usually at the bottom left). In the compose window, look for a small icon near the bottom of the text area — it may look like a pen, a star, or a sparkle, depending on your version of Gmail. Click it and choose "Help me write." Then type a short note about what you want the email to say: > Politely reply that I can't make the meeting on Thursday and suggest Friday afternoon instead > Write to the tradesman who gave me a quote last week and ask if the price is still valid Gmail will draft a full email based on your note. Read it through, fill in any details the AI left as placeholders, and edit anything that doesn't sound like you. You can also ask it to adjust the tone — more formal, friendlier, or shorter — before you send. When you open an email that ends with a question or a request, Gmail often shows two or three short reply options below the message: something like "Sounds good, I'll be there" or "Thanks, I'll take a look." Tap one to open it as a draft you can edit before sending, or ignore them if you want to write your own reply. The 2026 AI Inbox update made these suggestions more context-aware — they now draw on the full thread, not just the last message. If you have a Google One AI Premium subscription, look for the AI Inbox section near the top of your Gmail. It shows a personalized summary of what needs your attention today: packages arriving, upcoming appointments confirmed by email, bills due, and priority messages from people you hear from regularly. The briefing puts emails from your frequent contacts first. You can dismiss it once you have read it, or tap individual items to open the underlying email directly. Google enabled AI features for all users by default, but turning them off takes less than a minute. **On Gmail on the web:** click the gear icon (top right) > **See all settings** > **General** tab > scroll to **Smart features and personalization** > toggle it off. **On the Gmail app (Android or iOS):** tap your profile picture > **Manage your Google Account** > **Data & privacy** > scroll to **More options** > **Other Google activity** > **Gmail smart features and personalization** > turn it off. Disabling this turns off thread summaries, Help Me Write, AI Overviews in search, and Suggested Replies. Your email continues to work exactly as before — you just see no AI assistance. The most practical use I have found is asking conversational search about real-life admin: "what did the electrician quote me," "when does my gym membership renew," or "find the booking confirmation for our trip." It works even when you can't remember who sent the email or what the subject was, which is most of the time. ## What to watch out for **Summaries can miss nuance.** A thread summary gives you the shape of a conversation, but tone and context can get lost. A sarcastic "fine, if that's what you want" might be summarized as agreement. For anything important — medical, legal, financial — read the actual emails before acting on a summary. **Help Me Write doesn't know your specifics.** If you ask it to reply to a tradesman about a quote, it won't know the quoted amount, the job details, or the contractor's name unless you put them in your prompt. Read the draft before sending and fill in anything it left as a generic placeholder. **Your emails are being processed by Google.** The AI reads your email content on Google's servers to make these features work. Google says it does not use personal Gmail content for advertising, but if you use Gmail for sensitive work, check your employer's data policies before using AI drafting features. **Gmail Live is still rolling out.** The voice interface was announced at I/O in May 2026 for Google AI Ultra subscribers. If you cannot find it, it may not have reached your region or account type yet. **Proofread is a paid feature.** The Proofread tool, which reviews your draft for grammar and tone, requires a Google One AI Premium subscription — it is not part of the free features. ## What to Try Next If you also use Google Docs or Google Sheets, [How to Use Gemini in Google Docs, Sheets, and Gmail](/guides/gemini-google-docs-gmail) covers the same AI assistant applied across the full Google Workspace. For a broader look at using AI to write better emails — not just drafts but structure and tone — [How to Write a Professional Email with AI Help](/guides/write-professional-email-with-ai) is the practical next step. --- # Google AI Mode: What Changed in Search and What to Trust Source: https://aiforregulars.com/guides/google-ai-mode-explained Google Search now does something it never did before: it writes you an answer. Instead of just showing a list of links, it reads dozens of those pages and tries to summarize what they say in plain language. That's Google AI Mode in a nutshell. This is a big shift. Here's what you need to know to use it well — and when to trust it. ## What Google AI Mode actually does When you type a question into Google today, you may see a shaded box at the top of the page before the regular blue links. That box contains an AI-generated answer written specifically for your question. ![Google AI Mode answering whether cooked chicken bones are safe for dogs](/screens/google-ai-mode-explained/ai-mode-answer.webp) *Google AI Mode with its source list on the right — our own screenshot, 31 July 2026.* The AI looks at many web pages at once — things like news articles, help guides, and product pages — and tries to pull out the key points. It then writes a short answer in plain English, often with a few source links you can tap to read more. In AI Mode (sometimes a separate tab labeled "AI"), you can also ask follow-up questions and have a real back-and-forth conversation, much like you would with ChatGPT. ## What changed from the old Google Old Google showed you a ranked list of links. You decided which to click. You read the page. You formed your own answer. AI Mode flips that. The AI forms the answer first. You see conclusions before you see sources. That's faster and more convenient. But it also means the answer can be wrong before you even know to doubt it. Other things that changed: - **Results feel more conversational.** You can ask things like "What's a good beginner yoga pose for lower back pain?" and get a direct, friendly answer rather than ten blue links. - **Follow-up questions work.** After the first answer, you can type "What about if I have knee problems?" and Google remembers what you already asked. - **Source links are smaller.** They're still there, but you have to look for them. Clicking the small citation chips reveals which sites the AI pulled from. ## What to trust — and what to double-check AI Mode is most reliable for: - **General explanations.** "How does a deductible work?" or "What is sourdough starter?" - **Step-by-step how-to questions.** Basic cooking, home repairs, or software instructions. - **Definitions and background information** on topics that don't change often. Be careful with: - **Prices and product details.** AI answers can be out of date by months. - **Medical, legal, and financial advice.** Always verify with a licensed professional or official source. AI Mode is not a doctor. - **Recent events.** If something happened in the last few days, the AI may not know yet — or may mix it up with older news. - **Specific numbers.** Statistics, dosages, opening hours, phone numbers — these go wrong more often than anything else. The single habit that has saved me the most headaches: I click at least one of the small source links whenever the AI answer tells me something specific — a price, a date, a medical fact. It takes ten seconds and has caught several mistakes. Treat the AI answer as a first draft, not a final source. ## How to read the AI answer box When you see the AI answer box, look for these elements: **The summary text** — the main answer, written in plain language. Read this, but don't stop here. **Citation chips** — small labeled boxes (often numbered or with site names) that tell you where the information came from. Click them to read the original source. **"Show more" or expand links** — some answers are trimmed. Expanding reveals more details and more citations. **A thumbs-down button** — you can flag the answer if it looks wrong. Google uses this feedback to improve. ## What to do when you're not sure If an AI answer surprises you, contradicts what you already know, or covers something important, treat it like a headline you need to fact-check: 1. Click a citation chip and read the source page yourself. 2. Look at the regular search results below the AI box. Do they agree? 3. For health or legal topics, go directly to a trusted organization's website — like the CDC, NHS, or a government agency. ## Can you turn it off? If you prefer the classic link list, scroll past the AI box — everything you'd normally see is still below it. There is no setting to disable AI Overviews — Google does not offer one. The most reliable workaround is the **Web** tab in the search results bar (you may need to click "More" to find it); selecting it shows only traditional text links with no AI summary. ## What to try next If Google AI Mode left you curious about other AI search tools, [How to Use Perplexity AI](/guides/how-to-use-perplexity) walks you through a search tool built entirely around AI answers and citations. And if you want a head-to-head breakdown, [AI Search vs Google](/guides/ai-search-vs-google) compares your main options side by side. ## Sources - [Get AI-powered responses with AI Mode in Google Search](https://support.google.com/websearch/answer/16011537) — Google - [Find information in faster & easier ways with AI Overviews](https://support.google.com/websearch/answer/14901683) — Google - [Expanding AI Overviews and introducing AI Mode](https://blog.google/products-and-platforms/products/search/ai-mode-search/) — Google --- # Ask Maps: How to Use Google Maps' New AI to Plan Trips in Plain English Source: https://aiforregulars.com/guides/google-maps-ask-maps Planning a trip used to mean switching between at least three different screens — Google Maps for navigation, a separate search for what's worth stopping at, maybe a travel blog for spots that don't show up in a standard search. Ask Maps is Google's answer to that: a Gemini-powered feature built directly into Google Maps that lets you type a real question and get a useful, location-aware answer, all without leaving the app. It launched on March 12, 2026, first in the US and India (reported by Google's blog, TechCrunch, and CNBC), and it works noticeably differently from the keyword search bar you're used to. Ask Maps rolled out to Google Maps on Android and iOS starting March 12, 2026, beginning in the US and India. Desktop support was planned but was not available at launch. To check: open the Google Maps app on your phone and tap the search bar. If Ask Maps is available, you'll see a conversational prompt or an icon that looks different from the standard "Search Google Maps" placeholder. If you only see the regular search bar, the feature hasn't reached your region yet — skip to the "If Ask Maps Isn't Available Yet" section near the end of this guide. Make sure you're signed into your Google account. Ask Maps is tied to your account, both to confirm your region's eligibility and to personalize answers based on your saved places and search history. Tap the search bar and type a question the way you'd say it to a friend. You don't need to use keywords or abbreviations. > "What are some interesting things to do within an hour of Denver?" > "Is there a good farmers market near me this weekend?" > "Suggest a scenic route from Portland to the coast with a stop for lunch." Ask Maps reads your full question rather than treating it as a keyword lookup, so you can include context: how much time you have, what kind of experience you're after, distance preferences. It also uses your current location automatically, so you don't need to specify your city if you're already there. One of the most practical uses for Ask Maps is planning what to see on a longer drive — something that previously meant opening a separate app or digging through travel forums. > "Suggest interesting stops on my drive to the Grand Canyon." > "What's worth seeing on a road trip through the Texas Hill Country?" Ask Maps can suggest places to eat, scenic overlooks, local landmarks, and quirky spots along your route. Tap any suggestion to open its full listing in Google Maps — photos, reviews, and hours all load there — so you can decide whether it's worth a detour before committing. If the first suggestions aren't quite right, follow up in the same conversation: > "Those are a bit too touristy — any smaller towns worth stopping in instead?" Ask Maps handles the kinds of neighborhood questions that used to require posting on a travel forum or asking a local. > "Which of these two neighborhoods is more walkable — Back Bay or South End in Boston?" > "What's the vibe in the Mission District in San Francisco?" > "Is the waterfront area in Savannah easy to get around on foot?" These questions give you a starting point rather than a definitive verdict — "walkable" means different things to different people, and Ask Maps reflects general patterns rather than your specific situation. But it's a fast way to get oriented before you arrive somewhere new. Ask Maps handles conditional questions much better than a standard keyword search, because you can stack multiple requirements in one sentence. > "Quiet café near the museum with parking nearby." > "Family-friendly beach in San Diego that allows dogs." > "Rooftop bar in Nashville open on Sunday afternoons." If Ask Maps can't find something that matches every condition, it will usually say which requirement was the limiting factor, and you can loosen it in a follow-up question. When I'm not sure exactly what I want, I ask an open question first — "what is this neighborhood known for?" — and then use the answer to build a more specific follow-up. Starting broad and narrowing down tends to work better than trying to pack every requirement into the first question. If you've been using Google Maps for a while and have saved places — starred restaurants, custom lists, places you've reviewed — Ask Maps can draw on that when making recommendations. > "Suggest somewhere for dinner in Austin that matches the kind of restaurants I usually save." This personalization is one of the main things that separates Ask Maps from asking a general AI like Gemini or ChatGPT in isolation, which has no idea what you've liked in the past. It won't always get it right, but if you've built up a reasonable history in Google Maps, the suggestions are more likely to match your taste than a generic top-ten list. You can manage what data Google uses under your account's My Activity and Location History settings. Turning off location history limits Ask Maps' ability to personalize, but the feature still works for general questions. ## What to Watch Out For **Verify opening hours before you drive.** Ask Maps pulls hours from Google's Places database, which can be out of date — just as a standard Google Maps listing sometimes shows incorrect hours. Before making a trip specifically to visit somewhere, tap through to the actual place listing and check the hours, or call ahead. Never rely solely on hours shown by an AI feature. **Availability is still rolling out.** Ask Maps launched in March 2026 in the US and India. Other regions were expected to follow, but on a timeline Google didn't specify at launch. If you don't see the feature yet, it may arrive with an app update. **It's a starting point, not a source of record.** Ask Maps gives you suggestions based on Google's data and your history — it won't know about recent closures, newly opened spots, or seasonal hours. Use it to get oriented, then verify the specifics through the actual place listing, the business's website, or a quick call. **It personalizes using your Google account data.** If you'd prefer more privacy, you can limit this under your Google account's Activity settings. The tradeoff is less personalized suggestions, but Ask Maps still works for general location questions. ## If Ask Maps Isn't Available in Your Region Yet You can get similar results with a two-app approach while you wait: 1. Open Google Maps normally for navigation and the map view you're familiar with. 2. Open Gemini (gemini.google.com) or ChatGPT in a separate tab or on a second device. 3. Ask your planning question there — "what are some interesting stops on a drive from Seattle to Portland?" — and note the place names it suggests. 4. Search for those places directly in Google Maps to see photos, hours, reviews, and get directions. It takes a few extra steps, but it works anywhere in the world. Both Gemini and ChatGPT are quite good at this kind of planning question. If you want a fuller walkthrough of using ChatGPT this way, [Plan Your Vacation With ChatGPT](/guides/plan-vacation-with-chatgpt) covers it in depth. This guide focuses on Ask Maps inside Google Maps specifically. If you're looking for dedicated AI trip-planning apps — tools built from the ground up for travel rather than a navigation app with AI added — [Best AI Travel Planners](/guides/best-ai-travel-planners) covers those options separately. ## What to Try Next For the bigger picture of planning a full trip — building an itinerary, finding accommodations, organizing a day-by-day schedule — [Plan Your Vacation With ChatGPT](/guides/plan-vacation-with-chatgpt) walks through how to use a general AI assistant for the whole process. If your phone runs on Android and you want to get more out of Gemini across all your Google apps, [Getting Started With Gemini on Android](/guides/gemini-android-guide) is the place to start. --- # Google NotebookLM: The AI Research Tool Anyone Can Use Source: https://aiforregulars.com/guides/google-notebooklm-guide Most AI tools answer questions using the whole internet. NotebookLM is different: it only uses the documents you give it. Upload a PDF, a Google Doc, or a web article, and then ask anything about it. The AI stays focused on your material — nothing more, nothing less. That makes it especially useful for research, work projects, school reading, and anything where you want answers grounded in a specific source rather than a general AI guess. Go to notebooklm.google.com in your browser. Sign in with your Google account — the same one you use for Gmail or Google Drive. NotebookLM is free to use. You'll land on the home screen, which shows any existing notebooks and a button to create a new one. Click New Notebook. Every notebook starts empty. You need to give NotebookLM something to read. Click Add source (or the large + button in the Sources panel on the left). You'll see options to upload a file, paste a web URL, link a Google Doc or Slides file, or paste text directly. Choose the option that fits your material: - For a PDF (a report, a book chapter, a contract): click Upload and choose the file. - For a web article: paste the URL. - For a YouTube video: paste the video link and NotebookLM will read the transcript. You can add multiple sources to one notebook — up to 50 sources on the free plan (NotebookLM Plus raises this to 100). Each individual source can be up to 500,000 words or 200 MB, whichever comes first. After you add a source, NotebookLM processes it. This usually takes a few seconds to a minute depending on the file size. A progress indicator shows when it's ready. When the source appears in the left panel with a checkmark, it's done. You can add more sources while others are still loading. Once your source is loaded, click in the chat box on the right side and ask anything about your document. Write naturally: - "What are the main points of this report?" - "Explain section 3 in simple terms." - "Does this mention anything about deadlines?" NotebookLM will answer using only what's in your uploaded sources. It also shows inline citations — small numbered markers that link to the exact passage it used. Click any citation to jump directly to that spot in the source document. Look for the Studio panel on the right side of the page. It offers one-click outputs from your sources, including Audio Overview, Video Overview, Mind Map, Study Guide, Briefing Doc, FAQ, Flashcards, Quiz, Slide Deck, Infographic, and Data Table. All are available on the free tier; free users have daily limits (for example, 10 reports, flashcards, or quizzes per day and 3 audio overviews per day). - **Study Guide** — key concepts and review questions from your sources - **Briefing Doc** — a plain-language summary designed to get someone up to speed fast - **Flashcards** — auto-generated question-and-answer cards for memorizing key facts - **FAQ** — automatically generated questions and answers based on the content These are generated fresh from your documents. Click any of them to get a structured overview without typing a single question. One of NotebookLM's most popular features is Audio Overview. Find the Audio Overview button in the Notebook Guide panel and click Generate. Two AI voices will have a conversation about the content of your documents — explaining key ideas, asking each other questions, and summarizing what they found. The result is an audio file whose length depends on the format you select. The Brief is a single-speaker summary that runs under two minutes. The Deep Dive (the default) is a longer two-host conversation that can run up to 20 minutes. The Critique and The Debate are additional two-host formats. Generation typically takes a couple of minutes. Free accounts can generate up to 3 Audio Overviews per day. This is great for: - Getting a feel for a long document before reading it in full - Reviewing material while commuting, cooking, or doing something else - Sharing a spoken summary of a complex document with someone who won't read the full text Create a new notebook for each separate topic or project. NotebookLM keeps notebooks completely isolated — sources from one notebook don't affect another. Some examples of how to organize: - One notebook per work project - One for a trip you're planning (add itinerary PDFs, articles, hotel pages) - One for a health topic you're researching (add doctor's notes, articles from trusted sources) - One for a book or course you're working through I used NotebookLM to work through a long government document that would have taken me hours to read carefully. I uploaded the PDF, asked "What does this say about eligibility requirements?" and got a clear summary with citations pointing to the exact pages in under a minute. It didn't replace reading the document, but it helped me know exactly which pages actually mattered. ## What to try next If you want to see how NotebookLM compares to other AI research tools, [AI Search vs Google](/guides/ai-search-vs-google) gives you a clear breakdown of what each tool does best. And if you're new to AI tools overall, [What Is ChatGPT](/guides/what-is-chatgpt) is a friendly place to start. --- # Grammarly vs ChatGPT: Which Writing Tool Should You Actually Use? Source: https://aiforregulars.com/guides/grammarly-vs-chatgpt Both Grammarly and ChatGPT improve your writing. But they work in fundamentally different ways, and choosing the wrong one for the job means extra friction and worse results. This comparison breaks down which tool wins for each common writing task. ## How They Work Differently **Grammarly** sits alongside your text as you write. It underlines issues and suggests fixes inline — similar to spell-check but smarter and more context-aware. You stay in control of the content; Grammarly just highlights the problems. **ChatGPT** is a conversation. You give it instructions — "rewrite this more formally" or "draft a cover letter for this job description" — and it produces full text. You can keep refining through follow-up messages. The core difference: Grammarly edits. ChatGPT writes. ## Head-to-Head Comparison | Task | Better Tool | Why | |------|-------------|-----| | Fixing grammar and spelling | Grammarly | Inline suggestions, no copy-paste needed | | Writing a first draft from scratch | ChatGPT | Generates full text from a prompt | | Rewriting a weak paragraph | ChatGPT | Better at restructuring whole ideas | | Checking email tone | Grammarly | Tone detector works in real time | | Writing a cover letter | ChatGPT | Drafts to your specific job description | | Proofreading a long document | Grammarly | Reviews the whole doc inline | | Fixing unnatural-sounding phrasing | ChatGPT | Better at making sentences sound native | | Plagiarism check | Grammarly Pro | ChatGPT has no plagiarism check | | Working inside Gmail / Google Docs | Grammarly | Browser extension integrates directly | | Explaining why something sounds off | ChatGPT | Will explain the grammar rule if asked | ## When Grammarly Wins Grammarly is fastest for **editing text you've already written**. Its browser extension works inside Gmail, Google Docs, LinkedIn, and most other web apps — you don't need to copy anything out or open a separate tab. The tone detector is useful when you're writing something sensitive: a complaint, a resignation letter, feedback to a colleague. It shows whether your message reads as confident, formal, or accidentally aggressive, and lets you adjust before hitting send. The free tier handles most everyday needs. Basic spelling, grammar, and clarity suggestions are enough for emails and short documents. I use Grammarly mainly to check the tone of emails before I send them. More than once it flagged a sentence that would have come across harsher than I intended. That real-time warning has saved some awkward conversations. ## When ChatGPT Wins ChatGPT wins whenever you need to **create something from nothing** or **heavily rewrite** something that isn't working. Cover letters, professional bios, complaint emails, speech drafts, social media posts — these all start better in ChatGPT. You describe what you need, it produces a draft, and you edit from there. That's faster than staring at a blank page trying to get the first sentence right. ChatGPT is also better at explaining *why* something sounds off, not just that it does. If you ask "why is this sentence awkward?" it will tell you in plain English and suggest alternatives. Grammarly underlines the problem; ChatGPT teaches you about it. ## The Combination Approach For important writing, using both tools together gets the best results: 1. Draft in ChatGPT 2. Paste the result into a Grammarly-monitored editor (Google Docs or Gmail) 3. Accept Grammarly's grammar fixes and check the tone score This takes a few extra minutes but produces noticeably better writing for anything high-stakes: job applications, client-facing emails, performance reviews. ## Pricing Summary Grammarly offers a free tier and a paid Pro plan (starting at $12/month billed annually; formerly called Premium). ChatGPT offers a free tier (with usage limits), a Go plan ($8/month), a Plus plan ($20/month), and Pro plans ($100–$200/month). For most everyday users, the free versions of both are sufficient. The paid tiers add features aimed at professional and heavy users, such as plagiarism checking for Grammarly Pro and higher usage limits for ChatGPT. ## What to Try Next Ready to put ChatGPT to work for writing? [How to write a cover letter with AI](/guides/chatgpt-cover-letter) has a step-by-step process from job description to final draft. For professional emails, [writing professional emails with AI](/guides/write-professional-email-with-ai) has copy-paste prompts for common situations. --- # The 'Grandparent Scam' 2.0: AI Edition — A Family Safety Plan Source: https://aiforregulars.com/guides/grandparent-scam-ai *The account below is a composite based on patterns common to reported grandparent scam cases. Names and details are fictional. It illustrates how this scam unfolds so you can recognize the warning signs.* --- It was a Tuesday morning when Margaret's phone rang. She recognized her grandson Daniel's voice immediately — that same voice she'd heard hundreds of times, with the specific way he said "Grandma" that made it sound like he was already a little apologetic for calling. "Grandma, I'm in real trouble." He'd been in a car accident in another city, he said. Not his fault. But the other driver was hurt and the police were there. He needed $4,800 in gift cards — right now — to cover the other driver's medical costs before they pressed charges. And please, please don't call Mom and Dad. He didn't want them to worry, and there wasn't time anyway. Margaret was already standing up, reaching for her purse. What she didn't know was that Daniel was at work, fine, 40 miles away. The voice she heard was his — assembled from videos he'd posted online. A scammer had fed those clips into an AI voice-cloning tool and produced something close enough to fool his grandmother. She drove to four stores and bought $4,800 in gift cards. By the time the real Daniel called that evening for their usual weekly chat, the money was gone. ## How the Scam Works The grandparent scam is one of the oldest phone fraud schemes — but AI has given it a new and much more convincing layer. Understanding the pattern is the first part of protecting against it. **Finding the audio.** Social media is a goldmine for this. A TikTok video, a YouTube clip, a voice note shared on Instagram — security researchers have cloned a passable imitation from as little as three seconds of someone's natural speech, and current AI voice-cloning tools only get more convincing with a longer sample, easily enough to fool a close family member over the phone. **The call itself.** The voice sounds right. The emotional urgency is high — an accident, an arrest, a medical emergency. The situation is designed to trigger protective instincts, not critical thinking. Every detail is chosen to make you feel you need to act immediately. **The "stay on the phone" rule.** A key part of the scam is keeping you on the line or insisting you call back only on numbers they provide. This prevents you from taking 60 seconds to call your grandchild directly and verify. **A second caller plays an authority figure.** Often a "lawyer" or "bail bondsman" or "police officer" comes on the line after the "grandchild," adding official-sounding pressure to the urgency. These characters are also part of the script. **An untraceable payment.** Gift cards, wire transfers, and cryptocurrency are all commonly requested. All three are nearly impossible to recover once sent. ## Why the Old Instincts Don't Work Anymore Most of us grew up learning to trust our ears. If you recognize someone's voice, it must be them. AI voice cloning has broken that assumption. The voice alone is no longer reliable evidence. The emotional urgency makes it worse. When someone you love sounds frightened, every instinct says to help now. Scammers engineer exactly this pressure because they know it short-circuits the kind of careful thinking that would otherwise catch them. There's no shame in being affected by this. It's designed to exploit the most fundamental human responses — love, protection, and trust. The defense isn't being less loving. It's having a plan in place before the call happens. When I talked with people who'd nearly fallen for versions of this scam, the detail that struck me most was how specific the cloned voice felt — not just "sounds like him" but "that *is* him." The emotional recognition is overwhelming. Setting up a code word in advance is the only protection that reliably cuts through that kind of certainty. ## The Family Safety Plan: Five Steps to Do This Week You don't need to wait until something happens. These five steps, done together as a family, take about 20 minutes and create a safety net that's hard to break. ### 1. Set Up a Family Code Word Agree on a word or short phrase that only your immediate family knows. If anyone calls claiming to be a family member in trouble, they must give the code word when asked. No legitimate family member will be offended by the question. Make the word memorable but not obvious — not a birthday, not a pet's name that appears anywhere on social media, not a shared last name. Something unexpected and specific. [This guide on family safe words](/guides/family-safe-words) walks through choosing one and making sure everyone remembers it. ### 2. Set a "No Gift Cards" Rule No court, no lawyer, no police department, and no government agency will ever ask you to pay a fine or fee with gift cards. This is always a scam, without exception. Agree as a family that anyone requesting gift cards as payment is to be hung up on immediately, no matter how convincing the story sounds. ### 3. Make It Normal to Hang Up and Call Back Decide together that it is always acceptable — for any family member, any age — to say: *"I'll call you right back from the number I have for you."* This is not rude. It is safe. A real family member in a real emergency will wait 60 seconds for a callback. A scammer will argue against calling back, push for urgency, and insist you stay on the line. That insistence is itself a warning sign. ### 4. Keep Social Media Audio Private Where Possible This scam requires the scammer to collect audio samples. On platforms like TikTok, Instagram, and YouTube, consider: - Setting personal videos to friends-only or private where it makes sense - Being thoughtful about which family members have public-facing accounts with speaking video This won't eliminate all risk, but it limits the raw material available for cloning. ### 5. Talk About It Together, Without Alarm The most important part of this plan is that everyone in the family knows about it — including the grandchildren, who may be the ones most likely to have public video online. Have one conversation — a family call, a message in a group chat, a visit — where you walk through what the scam looks like and what your family's code word is. Not to frighten anyone. Just to be ready. ## What to Try Next Now that you know how this scam works, [AI Voice Cloning Scams](/guides/ai-voice-cloning-scams) goes deeper into how voice-cloning technology works and what its current limits are. And if you want to sharpen your ability to recognize an AI voice in real time during a call, [How to Tell If a Phone Call Voice Is AI](/guides/detect-ai-voice-phone-call) gives you six specific things to listen for. ## Sources - [Scammers use AI to enhance their family emergency schemes](https://consumer.ftc.gov/consumer-alerts/2023/03/scammers-use-ai-enhance-their-family-emergency-schemes) — Federal Trade Commission - [Family Emergency Scams](https://consumer.ftc.gov/all-scams/family-emergency-scams) — Federal Trade Commission - ['Grandparent' Scams Get More Sophisticated](https://www.fcc.gov/consumers/scam-alert/grandparent-scams-get-more-sophisticated) — Federal Communications Commission - [Artificial Imposters: Cybercriminals Turn to AI Voice Cloning for a New Breed of Scam](https://www.mcafee.com/blogs/privacy-identity-protection/artificial-imposters-cybercriminals-turn-to-ai-voice-cloning-for-a-new-breed-of-scam/) — McAfee - [Scammers Target Older Adult Victims](https://www.fbi.gov/news/stories/scammers-target-older-adult-victims) — Federal Bureau of Investigation --- # Hidden AI Features on Your Smartphone You Did Not Know About Source: https://aiforregulars.com/guides/hidden-ai-features-smartphone Your smartphone is already running AI in the background — analyzing your photos, listening for context clues, predicting what you will type next. But most of these features hide in menus you have never opened, or they only announce themselves with a subtle sparkle animation that is easy to miss. Here are the best ones worth knowing about, organized by what they do. ## Quick Reference: What's on Which Phone | Feature | iPhone (Apple Intelligence) | Google Pixel | Samsung Galaxy AI | |---|---|---|---| | AI writing suggestions | Yes (Writing Tools) | Yes (Help me write) | Yes (Chat Assist) | | Live call translation | Limited | Yes (Call Screen + Interpreter) | Yes (Live Translate) | | Robocall screening | Partial (Silence Unknown) | Yes (Call Screen) | Partial | | Circle/search on screen | No | Yes (Circle to Search) | Yes (Circle to Search) | | Smart photo cleanup | Yes (Clean Up) | Yes (Magic Eraser) | Yes (Object Eraser) | | AI text summary | Yes (Summarize) | Yes (Recorder app) | Yes (Note Assist) | | Real-time translation | Yes (Translate app) | Yes (Interpreter mode) | Yes (Interpreter) | ## 1. Live Call Translation If you regularly talk to people who speak a different language, this is a game-changer. **Pixel phones** have an Interpreter mode that translates conversations in real time, showing each side on screen and reading the translation aloud. **Samsung Galaxy AI** has Live Translate built into the Phone app — both sides hear each other in their own language through the phone's speaker. You do not need a translator app. Just open the Phone app and look for the translation icon before or during a call. ## 2. Robocall Screening (Android / Pixel) Google Pixel phones have a **Call Screen** button that appears when an unknown number calls. Instead of answering or ignoring, you tap Screen Call. Google's AI answers on your behalf, asks what the call is about, and transcribes the response on your screen in real time. If it is a real person with a real reason, you can pick up. If it is a robocall, you have not wasted 30 seconds of your life. iPhone's equivalent is less conversational — it silences unknown callers automatically, but it does not screen them. ## 3. Writing Tools (iPhone) and Help Me Write (Android) Select any text you are about to send — in Messages, Mail, Notes, or most other apps — and look for a **Writing Tools** option (iPhone) or **Help me write** (Pixel/Android). These let you rewrite the selected text to sound more formal, more casual, shorter, or longer. On iPhone, you can also proofread and summarize. It is faster than copying text into ChatGPT and works inside any app. **How to access on iPhone:** Select text → tap the AI wand icon or "Writing Tools" in the context menu (requires iPhone 15 Pro or iPhone 16). **How to access on Pixel:** Long-press text in a text field → look for "Help me write" in the options. ## 4. Smart Photo Cleanup Every major phone brand now has a tool that removes unwanted objects from photos. **iPhone 16** users have **Clean Up** in the Photos app. **Pixel phones** have **Magic Eraser**. **Samsung Galaxy** has **Object Eraser**. All three work similarly: open a photo, tap the AI eraser tool, circle or tap the thing you want gone, and the phone fills it in. Photobombers, power lines, trash cans in the background — they disappear in seconds. The feature has been in Pixel phones since the Pixel 6, so if you have an older Pixel, you already have it and may never have tried it. I used Object Eraser to remove a tourist from what would have been a perfect photo of a historic doorway in Lisbon. It took about eight seconds. I used to assume photo cleanup required a laptop and photo editing software — finding it already on my phone felt like finding a $20 bill in an old coat. ## 5. AI Text Summaries Both iPhone and Android can summarize long text for you. **iPhone** can summarize notification stacks (a bunch of emails or messages becomes one short line in your lock screen). It can also summarize web pages in Safari. **Google's Recorder app** on Pixel phones transcribes recorded audio and then lets you ask questions about it — useful after a long meeting or lecture. **Samsung Notes** has Note Assist, which summarizes long notes with one tap. ## 6. Smart Suggestions and Predictive Actions This one is subtle but constant. Your phone watches patterns — where you go, what apps you open at what times, who you call on Monday mornings — and makes suggestions. On iPhone, this shows up as Siri Suggestions in Spotlight search and on your lock screen. On Android, it appears as App Suggestions in the app drawer. You can control how much data the phone uses for this in Settings > Privacy > Analytics (iPhone) or Settings > Google > More > Customize your Google experience (Android). ## What to try next Many of the camera-specific AI tools deserve their own deep dive — for Magic Eraser, Photo Unblur, and portrait AI, see [Hidden AI Features Already in Your Phone's Camera](/guides/ai-camera-features-phone). For Samsung's full set of AI features, [Galaxy AI: A Complete Guide](/guides/galaxy-ai-features) covers the whole package. --- # How AI Screens Your Job Application Before Any Human Reads It Source: https://aiforregulars.com/guides/how-ai-screens-resumes You spend an hour tailoring your resume, write a careful cover letter, and click submit. What happens next is probably not what you picture. At most large employers, your application goes to software before it goes to any person — and that software has already decided whether to surface your resume by the time a recruiter sits down at their desk. Understanding the process doesn't require a tech background. It requires knowing what these systems look for and how to give them what they need. ## The two filters standing between you and a human ### The old system: ATS An Applicant Tracking System is essentially a database that ingests your resume and breaks it into fields: name, contact info, each job title and employer, dates, education, skills. Recruiters don't usually read the original document — they search the database using keywords and filters. The catch: if the software parses your resume incorrectly, the information goes into the wrong field or gets dropped entirely. A fancy two-column layout, a table, or text inside a graphic can all confuse the parser. Your five years of project management experience might simply not appear in the database. ### The new layer: AI scoring On top of ATS, many larger employers now use AI tools that go a step further. Instead of just storing and retrieving your information, they score each candidate against the job requirements and rank applicants for the recruiter. These systems typically compare your resume text against the job posting — looking for matching skills, relevant experience, and appropriate background. Some also pull from your LinkedIn profile if you've linked it. The output is often a percentage match or a priority ranking that determines the order in which a recruiter reviews applications. Tools in this space include HireVue, used by more than 60 percent of Fortune 100 companies for AI-scored video and asynchronous interviews; Eightfold AI, which matches candidates based on skills and career potential using a dataset of over a billion career profiles; and Paradox, now part of Workday following its 2025 acquisition, which handles conversational screening for high-volume roles. Many employers also rely on AI ranking features built directly into their existing ATS — Workday, Greenhouse, and iCIMS all include them — rather than a separate standalone tool. ## What AI looks for in a resume AI screening tools are generally looking for signal that your background matches the role. In practice, that means: **Skills and keywords.** The tool compares terms in your resume to terms in the job posting. If the posting asks for "data analysis" and your resume says "analyzed datasets," they may not match, depending on how sophisticated the system is. When a term from the posting accurately describes your experience, use that exact term. **Relevant job titles.** Your job titles are weighted heavily. If you've held titles that don't obviously connect to the role, a brief line in your summary or bullet points that uses the industry-standard term can help the system make the connection. **Years of experience.** Many systems extract and score how long you've done relevant work. Vague date ranges or gaps without explanation can lower your score. **Education.** Degree level and field of study are usually parsed into separate fields. If the role requires a specific credential, make sure it's clearly stated. ## What actually hurts your score - **Fancy formatting:** Text boxes, graphics, icons, two-column layouts, and tables are hard for parsers to read. Use a plain, single-column format. - **Generic language:** Phrases like "results-oriented professional" or "team player" don't match job-specific keywords and add no signal. - **Missing keywords:** If the job posting lists specific software, certifications, or skills you have, include them explicitly — don't assume the system will infer them. - **Wrong file format:** When given a choice, submit a .docx or plain PDF. Some systems struggle with older file formats or heavily formatted PDFs. I tested the same resume — one with a clean single-column layout and one with a two-column design — through a publicly available ATS parser tool. The two-column version had the applicant's name appearing in the skills field and the job history dates stripped entirely. The single-column version parsed perfectly. That difference alone could determine whether a recruiter ever sees your application. ## The fairness question AI screening raises real equity concerns worth knowing about. Systems trained on historical hiring data can encode past biases — favoring candidates from certain schools, with certain name patterns, or with career paths that reflect who got hired before, not who is actually best for the job. Several U.S. states and some countries have begun regulating automated hiring tools, requiring audits for bias or disclosure to candidates. If you believe you were screened out unfairly, some jurisdictions now give you the right to request information about the process. This doesn't mean you should avoid applying. It means knowing the system exists so you can work with it — and advocate for fairer practices when you have the standing to do so. ## Format tips that help both systems 1. Use a clean, single-column Word document or simple PDF. 2. Put your job title and employer name on the same line, with dates clearly next to them. 3. Use standard section headings: Work Experience, Education, Skills — not creative names like "My Journey." 4. List skills explicitly in a dedicated Skills section, not just mentioned in passing in bullet points. 5. Mirror the language of the job posting when it accurately describes what you've done. ## What to try next Once you understand how your resume is being read by software, the next step is knowing how to format and write it to pass through cleanly. The guide on [building an ATS-friendly resume with AI](/guides/ats-friendly-resume-with-ai) walks through that process step by step, including how to use ChatGPT to tailor your resume to a specific posting. If you're over 50 and concerned about how AI systems treat older candidates, [AI job search after 50](/guides/ai-job-search-over-50) covers that directly. --- # How to Spot a Deepfake Video: 7 Signs Anyone Can Check Source: https://aiforregulars.com/guides/how-to-spot-deepfake-video Deepfake videos have gotten more convincing, but they still leave clues. You do not need any special tools or technical background to catch most of them — you just need to know where to look. These seven signs can each be checked in seconds, on any device, with no downloads required. Watch the border between the person's face and their hair, neck, or background. In deepfakes, this edge often looks slightly blurry, smeared, or wobbly — especially when the person moves their head. Real faces have a clean, consistent boundary with the surrounding scene. If the skin seems to float, shimmer, or flicker at the edges during movement, that is a strong warning sign. Pause the video on a frame where the person is mid-motion to make this easier to see. Natural blinking is irregular — sometimes fast, sometimes slow, occasionally skipped for a moment while someone is focused. Early deepfakes blinked too rarely or not at all; newer ones blink more, but the blinking still sometimes feels mechanical or too uniform. Watch for eyes that blink at an identical rate throughout the clip, or eyes that look slightly glassy or flat compared to the rest of the face. Natural eyes also catch light and move slightly together — both things AI sometimes gets wrong. Play the video with sound and watch the lips closely. In a real video, the exact shape of the mouth matches each sound being spoken. In deepfakes, the lip movements sometimes run slightly ahead or behind the audio, or the shapes do not quite match the words being said. This is easiest to catch on hard consonants — the letters "p," "b," and "m" require the lips to press together fully. If the lips are open when you hear one of those sounds, the sync is off. Lighting in a real video is consistent — if the room is lit from the left, the face should have a shadow on the right side, and everything in the scene should follow the same logic. In deepfakes, the lighting on the swapped face often does not quite match the rest of the scene. The face might look a bit too evenly lit, or it might cast a shadow in a direction that does not match the background. This mismatch is one of the harder things for AI to fake convincingly. I once caught a deepfake by noticing that the speaker's earring disappeared and reappeared between sentences. It was subtle — I almost missed it on first watch. Now I always take a quick look at small accessories like earrings, glasses, and collar edges, because the AI tends to struggle with those specific details. Small accessories and fine details are genuinely difficult for AI to maintain consistently throughout a video. Watch for earrings that blur into the neck or flicker, glasses frames that warp where they cross the eye, necklaces that seem to meld into clothing, or individual strands of hair that look oddly smooth or plastic. These elements should remain stable and look natural throughout the whole clip. Any flickering or inconsistency in a small detail is worth noting. When the person moves their head, watch the area of the background immediately around and behind them. Deepfakes sometimes cause the background to warp, ripple, or blur in a band that follows the edge of the head during movement. This is most visible during quick turns or nods. Pausing on a frame just after a fast head movement often makes this distortion easy to spot. No visual check replaces source checking. Before accepting a shocking or surprising video as real, ask: who posted this, on what account, and why? A video of a celebrity making an unexpected statement that appears on a new or unfamiliar account, shared without context, deserves extra scrutiny. Real breaking news and genuine viral moments are picked up by many sources quickly. If only one account is sharing a clip and it has no clear origin, search for the same content elsewhere before treating it as real. ## When Several Signs Point the Same Way No single sign is proof on its own — some appear in low-quality real videos too. But when you notice two or three of these signs in the same clip, the probability of manipulation goes up sharply. The goal is not to be completely certain before you act — it is to build the habit of pausing to look. That moment of attention is the most useful thing you can practice. ## Where This Method Stops Working These signs are not foolproof, and it's worth knowing that up front. A January 2026 study out of the University of Florida (published in Cognitive Research: Principles and Implications) compared people against automated detection tools on short test clips: humans correctly told real video from deepfake about two-thirds of the time, while the algorithms performed close to chance on video. That's a decent result — but it came from a lab setting where participants already knew to look for something suspicious. In everyday life, it's noticeably harder even when you know to look for a problem: a survey by the biometric-verification company iProov of 2,000 ordinary consumers in the US and UK, who were explicitly told to tell real content from deepfakes, found that only 0.1% of them got every answer right — one person in a thousand. In other words: these seven signs raise your odds of noticing something suspicious, but a clean result — nothing looked wrong — does not mean the video is definitely real. It only means it isn't a crude fake. When money, login credentials, or another risky action is on the line, don't rely on your own eyes alone. Verify through a different channel — a call-back to a known number, a message on another platform, or meeting in person — before you act. **What to try next:** Fake videos are one part of the picture. See [How to Spot AI-Generated Photos](/guides/spot-ai-generated-photos) for the same approach applied to still images, and read [AI Voice Cloning Scams](/guides/ai-voice-cloning-scams) to understand the voice-based version of these threats. ## Sources - [Machines spot deepfake pictures better than humans, but people outperform AI in detecting deepfake videos](https://news.clas.ufl.edu/machines-spot-deepfake-pictures-better-than-humans-but-people-outperform-ai-in-detecting-deepfake-videos/) — University of Florida - [Is this real? Susceptibility to deepfakes in machines and humans](https://link.springer.com/article/10.1186/s41235-025-00700-y) — Cognitive Research: Principles and Implications, 2026 - [iProov Study Reveals Deepfake Blindspot: Only 0.1% of People Can Accurately Detect AI-Generated Deepfakes](https://www.iproov.com/press/study-reveals-deepfake-blindspot-detect-ai-generated-content) — iProov, 2025 - [Report Fraud](https://reportfraud.ftc.gov) — Federal Trade Commission - [Internet Crime Complaint Center (IC3)](https://www.ic3.gov/) — FBI --- # How to Tell If a Text Was Written by AI (What Works, What Doesn't) Source: https://aiforregulars.com/guides/how-to-tell-if-text-is-ai Trying to figure out whether something was written by AI has become one of the most common questions in schools, workplaces, and newsrooms. Teachers want to know if students are submitting AI-written essays. Editors wonder about freelance articles. Parents wonder what their kids are actually doing online. The honest answer is uncomfortable: there is no reliable way to know for sure. But that does not mean all signals are useless. ## What AI Writing Actually Looks Like AI text tends to hit a middle-ground tone. It is rarely too formal or too casual. It flows smoothly from one sentence to the next, almost like a well-edited document — but without the little detours and personal touches that real writers include. A few patterns show up often: **Vague but confident claims.** AI tends to say things like "many experts believe" or "studies have shown" without naming who or what. The sentences sound informed, but there is nothing specific behind them. **Generic examples.** When AI gives an example, it often picks the most obvious one — the same example a hundred other people would pick. Real writers tend to draw on things they have actually seen or experienced. **Certain repeated words.** "Delve," "nuanced," "comprehensive," "it is worth noting," and "in conclusion" appear far more often in AI output than in everyday human writing. This is not a rule — humans use these words too — but a cluster of them together is worth noticing. **Perfect structure.** AI almost always organizes text into tidy sections with balanced paragraphs. Human writing tends to be messier: one section runs long, another is just two sentences, the conclusion sometimes feels rushed. **No opinions.** AI tends to present "both sides" without landing anywhere. Human writers, even when trying to be balanced, usually let their view show through in small ways. ## What Does NOT Work as a Signal **Gut feeling.** Most people think they can tell AI-written text by feel. Studies testing this have found that even experienced readers perform only slightly better than a coin flip. Well-polished human writing gets flagged constantly. Clunky AI output sometimes slips through. **Smooth grammar and spelling.** AI writes correctly, but so does anyone who proofreads carefully. A spelling error is not proof that a human wrote something — it just means it was not proofread. **Content accuracy.** AI can be right, wrong, or anywhere in between. Accurate text is not automatically human-written, and inaccurate text is not automatically AI. **Sentence length variety.** Some people believe AI writes in uniform sentence lengths. This was truer of early models. Current AI tools mix lengths much more naturally. When I read something and wonder if it is AI, I try to ask myself one question: does this text contain anything that only *this* person could know? A specific memory, a named friend, a detail from their own life? AI can fake a lot of things, but it cannot fake the particular. If the writing feels oddly universal — like it could apply to anyone — that is worth noticing. ## Why AI Detectors Are Not the Answer There are now dozens of tools that claim to detect AI-written text. Some are free. Some charge a subscription. They work by scanning for statistical patterns in the text — patterns that show up more often in AI output than in human writing. The problem is that these patterns overlap. A non-native English speaker writing a careful, formal essay can look statistically very similar to AI output. So can a student who writes in a structured, straightforward style. Researchers and journalists have documented cases where detector tools flagged the US Declaration of Independence, passages from the Bible, and student essays from non-native English speakers — all as AI-generated. On the flip side, detectors miss AI text regularly — especially when the AI has been asked to write in a casual or conversational tone, or when someone edits the AI output slightly before submitting. The tools are not useless for getting a rough general sense. But they are not evidence. Using a detector result as the basis for accusing a student of cheating is not fair and is not defensible. ## What Teachers and Parents Can Do Instead If you suspect someone used AI, the most useful thing is a conversation — not a detector report. Ask the person to explain their work. Ask them what research they did, what was hard, what they would change. Someone who engaged with the material will have answers. Someone who had AI write it will struggle. For students, the most important question is not "did you use AI?" It is "did you learn anything?" AI can do the work, but it cannot learn for you. The skills that matter — thinking through a problem, making an argument, finding evidence — only build up when you do them yourself. For teachers setting assignments, building in some form of in-person discussion, oral explanation, or drafting process makes AI-assisted shortcuts much harder to hide and much less tempting. ## What to Try Next If you want a closer look at how detection tools actually compare, read [AI Detectors Tested: Accuracy, False Positives and What Teachers Should Know](/guides/ai-detectors-tested). If you are a parent trying to figure out how to handle AI homework use with your kid, [My Kid Uses ChatGPT for Homework — A Parent's Guide](/guides/kids-chatgpt-homework-parents-guide) walks through a calm, practical approach. --- # How to Get Rid of Google's AI Answer and See Normal Results Source: https://aiforregulars.com/guides/how-to-turn-off-google-ai-overviews You search for something simple and Google puts a long AI-written box above everything else. Many people would rather just see the list of websites, the way Google used to look. Here is what is possible, honestly stated: **Google offers no setting to switch the AI box off**, but you can get around it. ## What Google says Google's own help page states that AI Overviews are a core Google Search feature, like knowledge panels, and that such features cannot be turned off. It does point to one option: the **Web** filter, which shows only text-based links without features like AI Overviews. Everything below builds on that. ## Step 1: Use the Web filter after any search Type your question into Google on a computer and press Enter. Under the search box is a row of filters (All, Images, News and so on). At the end of the row, click **More**, then choose **Web**. You now see plain blue links, with no AI box. **The catch:** the filter applies only to that search. Your next search starts again with the AI box. That's fine for the odd search, but tiring if you do it all day. ## Step 2: Make Chrome open the Web view every time (computer) This is a workaround, not an official setting. It works because the Web view is simply a Google search address ending in `udm=14`. Android Authority and Search Engine Watch both describe this method. Click the address bar, type `chrome://settings/searchEngines` and press Enter. Next to **Site search**, click **Add** and fill in three boxes. Name: `Google Web`. Shortcut: `@web`. URL: `{google:baseURL}search?q=%s&udm=14` Click the three dots beside your new entry and choose **Make default**. From now on, searches typed in the address bar open as plain web results. To undo it, choose Google as the default again in the same list and delete your entry. Keep the plain Google search as an option, not the only one. For local results the normal view can be handier, so switch back when you need it. ## Things to know - **No promise it lasts.** Google could change or remove the Web view. If it stops working, you're back to the normal results. - **AI Mode is separate.** The Web view skips AI Overviews. [Google AI Mode](/guides/google-ai-mode-explained) is a different tab you open on purpose. - **Plain results can still mislead.** Links are not automatically trustworthy. See [AI search vs Google](/guides/ai-search-vs-google) for how to judge answers either way. ## Sources - [Find information in faster & easier ways with AI Overviews](https://support.google.com/websearch/answer/14901683) — Google Search Help - [How to turn off AI Overviews in Google Search](https://www.androidauthority.com/how-to-turn-off-ai-overviews-google-3445771/) — Android Authority - [How to turn off Google AI Overviews and get normal search results back](https://searchenginewatch.com/turn-off-google-ai-overviews/) — Search Engine Watch --- # How to Use ChatGPT: Step-by-Step for First-Timers Source: https://aiforregulars.com/guides/how-to-use-chatgpt If you have never used ChatGPT before, this guide walks you through everything from creating your account to getting a genuinely useful answer. The whole process takes less than 10 minutes. ![The ChatGPT message box on the web](/screens/how-to-use-chatgpt/composer.webp) *The ChatGPT home screen right after signing in — our own screenshot, 31 July 2026.* You do not need any technical experience. If you can send an email, you can use ChatGPT. Open any web browser — Chrome, Safari, Edge, or Firefox all work fine. In the address bar at the top, type **chatgpt.com** and press Enter. You will land on the ChatGPT homepage. It has a simple message box in the center and a "Sign up" button near the top right. If you already have an account, click "Log in" instead. Click the **Sign up** button. OpenAI will ask you to create an account. You have four options: sign up with your email address, or use an existing Google, Microsoft, or Apple account (these are faster and skip the email verification step). Many people find the Google option easiest — just click 'Continue with Google' and pick the Google account you want to use. If you sign up with email, you will receive a confirmation email. Open it and click the verification link to activate your account. After verifying your account, OpenAI may ask for your first name and date of birth. Fill those in and click Continue. You might also see a screen offering you a free trial of ChatGPT Plus (the paid plan). You can safely skip this for now — click "Stay on free plan" or the equivalent option. The free version is everything you need to start. Once you are logged in, you will see a clean screen with a text box near the bottom that says something like "Message ChatGPT." Click on it — your cursor will appear inside the box. Now type a real question or request. Here are a few ideas to try your very first time: - *"Explain what a credit score is in simple terms."* - *"Write a short thank-you email to my doctor's office."* - *"Give me five easy dinner ideas for a family of four."* Type your message and press **Enter** on your keyboard (or click the arrow button to the right of the box). ChatGPT will start typing its response almost immediately — you will see the words appear word by word, like someone typing back to you. This usually takes just a few seconds. Read the answer. If something is unclear or you want more detail, just type another message. For example: *"Can you make that even simpler?"* or *"Give me three more dinner ideas with chicken."* This is the key habit to build: ChatGPT is a conversation, not a one-time search. Going back and forth with follow-up messages usually gets you much better results than a single question. When I first started using ChatGPT, I kept my questions vague — things like "help me with an email." The answers were okay but generic. The moment I started being specific — "help me write a polite but firm email to my landlord about a leaky faucet that has not been fixed in two weeks" — the responses became genuinely useful. More detail in, better answer out. Each conversation in ChatGPT stays together as a thread. ChatGPT remembers what was said earlier in the same conversation, which is useful for follow-up questions. When you want to ask about something completely different, start fresh. Look for the **New chat** button in the top left corner (it looks like a pencil or a plus sign). Click it to open a blank conversation. Your old conversations are saved on the left side panel — you can click on any of them to go back. Before you use ChatGPT regularly, it is worth taking two minutes to check your privacy options. Click your name or profile icon at the bottom left of the screen, then choose **Settings**. Go to **Data controls**. You will see an option called something like "Improve the model for everyone" — if you prefer that OpenAI not use your chats for AI training, toggle that off. For a full walkthrough, see our [ChatGPT privacy settings guide](/guides/chatgpt-privacy-settings). ## Tips for Getting Better Answers A few habits that make a big difference: **Be specific.** Instead of "help with my resume," try "help me rewrite the work experience section of my resume for a retail manager job — I have 8 years of experience." **Tell it your situation.** "I am 62 years old and have never invested before — explain a Roth IRA in simple terms" gets a much more useful answer than just "what is a Roth IRA." **Ask for a different format.** If the answer is too long, say "Give me that as a short bullet list." If it feels too technical, say "Explain that more simply." **Push back when something seems off.** ChatGPT can make mistakes. If an answer does not sound right, say so. It will often correct itself or admit uncertainty when challenged. ## What Not to Share With ChatGPT ChatGPT is a helpful tool, but treat it like any online service: do not paste in your Social Security number, passwords, full credit card numbers, or other sensitive personal information. ![ChatGPT settings, Data controls section](/screens/how-to-use-chatgpt/data-controls.webp) *Settings → Data controls, where you turn off training on your chats — our own screenshot, 31 July 2026.* ## What to Try Next Now that you know the basics, put ChatGPT to work on something real. If you have a work email you have been putting off, [writing a professional email with AI](/guides/write-professional-email-with-ai) shows you exactly how to do it. Curious how ChatGPT stacks up against Google's AI assistant, Gemini? Read [ChatGPT vs Gemini](/guides/chatgpt-vs-gemini) to see which one might suit you better. ## Sources - [ChatGPT plans and pricing](https://openai.com/chatgpt/pricing/) — OpenAI - [Data Controls FAQ](https://help.openai.com/en/articles/7730893-data-controls-faq) — OpenAI - [Usage policies](https://openai.com/policies/usage-policies) — OpenAI --- # How to Use Perplexity AI: A Step-by-Step Beginner's Guide Source: https://aiforregulars.com/guides/how-to-use-perplexity Perplexity AI does something regular search engines don't: it reads the web for you and writes a direct answer, complete with links back to the sources it used. That means you spend less time clicking through pages and more time actually getting what you came for. Here's how to use it from scratch, including how to break the keyword habits that come from years of Googling. Open a browser and go to perplexity.ai. You don't need an account to start searching — the search box is right on the homepage. If you want to save your conversations and pick up where you left off, click Sign Up in the top right corner and create a free account with your email address or a Google account. Click inside the search box and type your question the way you'd say it out loud. You don't need to use short keywords — full sentences work better here. Instead of: `best walking shoe plantar fasciitis` Try: "What walking shoes are recommended for someone with plantar fasciitis?" Perplexity is designed to understand natural questions, so the more specific and natural your phrasing, the better the answer will be. After a few seconds, Perplexity writes a full answer in plain text. Inside the answer you'll notice small numbered superscripts — like [1] or [3]. These are citation markers. On the right side of the screen (or below on mobile), you'll see a list of sources matching those numbers. Each one is a real web page that Perplexity read to write its answer. Click any source to open it and verify what the AI said. This is the most important habit to build: if an answer contains a specific fact that matters to you, click the citation and read it yourself. Below the answer, you'll see a text box labeled something like "Ask a follow-up." Type your next question there — Perplexity remembers what you already asked, so you don't have to repeat yourself. For example, if you asked about shoes for plantar fasciitis, you could follow up with: "Are any of those available in wide widths?" The conversation continues like a chat, getting more specific with each turn. At the top of the search bar, look for a row of options like All, Academic, Writing, Video, or Social. These are called Focus modes — they narrow where Perplexity looks for answers. - **Academic** searches research papers and scholarly sources. Great for health or science questions. - **Video** finds video explanations on YouTube. - **Social** searches Reddit and forums for real-person opinions and experiences. For most everyday questions, leave it on All. Switch to Academic when you want peer-reviewed information rather than general articles. If you want to come back to a search later, create a free account and your threads will be saved automatically in the left sidebar. You can also click the Share button on any answer to copy a link — useful if you want to send a researched answer to someone else. The biggest adjustment for new Perplexity users is shifting from keyword thinking to question thinking. Old search habits — short, keyword-heavy phrases — often give worse results here. A few swaps that help: - Instead of `Italy travel tips`, ask: "What should first-time visitors know before traveling to Italy?" - Instead of `diabetes symptoms`, ask: "What are the early warning signs of type 2 diabetes?" - Instead of `Excel formula sum column`, ask: "How do I add up all the numbers in a column in Excel?" The more your question sounds like something you'd ask a knowledgeable friend, the better Perplexity responds. My first week with Perplexity, I kept typing keywords out of habit and getting so-so answers. The moment I started writing real questions — the kind I'd ask a knowledgeable friend — the answers got dramatically better. It sounds obvious, but it genuinely changes everything about how the tool feels to use. ## What to try next If you want to compare Perplexity against other AI search options, [AI Search vs Google](/guides/ai-search-vs-google) puts them side by side so you can see which tool fits which task. Or if you've been curious about ChatGPT as well, [How to Use ChatGPT](/guides/how-to-use-chatgpt) is a good next stop. --- # How to Talk to ChatGPT: 10 Rules for Better Answers Source: https://aiforregulars.com/guides/how-to-write-prompts Most people's first experience with ChatGPT goes like this: they type a vague question, get a vague answer, and wonder what all the fuss is about. The tool is not the problem — the prompt is. These 10 rules will help you get answers that are actually useful. Each one comes with a before-and-after example you can learn from immediately. ChatGPT does not know anything about you unless you say so. A little context goes a long way. **Before:** "Explain what a mutual fund is." **After:** ``` I am 58 years old and thinking about retirement for the first time. Explain what a mutual fund is in plain language, without assuming I know any finance terms. ``` When ChatGPT knows your situation, it calibrates the level of explanation and the examples it uses. You do not need a full biography — just the part that is relevant to the question. Many people type a topic instead of a task. ChatGPT responds much better to a clear request. **Before:** "Job interview." **After:** ``` Give me five common questions asked in a customer service job interview and a strong example answer for each one. ``` Think: "What do I actually want ChatGPT to do?" Then start your prompt with that action word: write, explain, summarize, list, rewrite, compare, suggest. The details you include upfront prevent ChatGPT from making assumptions. Assumptions lead to generic answers. **Before:** "Write a thank-you email." **After:** ``` Write a short thank-you email to my manager, Sarah, for covering my shift on Tuesday. I want it to sound warm and genuine, not formal. Keep it under 100 words. ``` Useful details to include: who the message is for, the tone you want, the length, and any specific facts that matter. ChatGPT will guess a format if you do not specify one. Tell it what you actually want. **Before:** "Give me advice on saving money." **After:** ``` Give me 7 practical tips for saving money on groceries, formatted as a numbered list with one sentence of explanation for each tip. ``` Format options you can request: bullet list, numbered steps, short paragraph, table, short sentences, one-page summary, step-by-step instructions. The more specific, the more usable the result. If you have a sample of the style or format you want, paste it in. ChatGPT will match it. **Before:** "Write a product description." **After:** ``` Write a product description for a reusable water bottle. Here is an example of the style I like: [paste example]. Match that tone and length. ``` This works especially well for writing tasks — emails, bios, social media posts — where tone matters. The single habit that improved my results the most was adding "Keep it simple — I am not an expert in this area" to questions about medical, legal, or financial topics. Without that, ChatGPT often defaults to a technical explanation that requires its own explanation. One sentence of context completely changes the answer you get. You do not have to get the perfect response on the first try. Treat it like a conversation. After you read the first response, you can say: - "Make this shorter." - "That last paragraph does not make sense — can you explain it differently?" - "Give me three more options." - "Rewrite the second bullet point to sound less formal." You are not locked into the first answer. Each follow-up refines the result. ChatGPT can make mistakes — dates, names, facts, calculations. It will not always flag its own errors. If something sounds off, challenge it directly. ``` I do not think that is right. [State the specific thing that seems wrong.] Can you check that and try again? ``` ChatGPT will often correct itself or acknowledge the error when pressed. Do not accept a wrong answer just because it was confidently stated. For anything important — health, legal, financial — always verify with a reliable source. You can ask ChatGPT to respond from a particular role or point of view. This focuses its answer on what actually matters to you. **Before:** "How do I negotiate a raise?" **After:** ``` Act as a career coach who helps people in their 50s re-enter the workforce. I want to ask my boss for a 10% raise after two years without one. What is the best approach? ``` Useful roles to try: "Act as a doctor explaining this in simple terms," "Act as a teacher explaining this to a complete beginner," "Act as an editor making this email clearer." For tasks like summarizing, explaining, or editing, paste the actual text directly into your prompt. ChatGPT cannot see files you have not shared, and it works much better with real material than descriptions of it. ``` Here is an email I received from my insurance company. I do not understand what it is asking me to do by Friday. Please summarize it in plain language and tell me what action I need to take: [paste the email] ``` This is one of the most practical uses of ChatGPT — making sense of confusing documents, bills, letters, or contracts. When you find a prompt that gets you a great result, keep it somewhere — a notes app, a document, a sticky note. You will use it again. You can also ask ChatGPT to help you make a prompt reusable: ``` I want to use this prompt regularly with different topics. Can you rewrite it as a template with [brackets] where I would swap in new details each time? ``` Over time, you will build up a small personal library of prompts that consistently give you exactly what you need. ## What to Try Next Now that you know how to write a good prompt, try putting these rules to work with [25 ready-to-use prompts for everyday situations](/guides/things-to-ask-chatgpt). If you use ChatGPT for work, [AI prompts for office work](/guides/ai-prompts-office-work) has more task-specific examples you can adapt right away. --- # How Your Bank Uses AI to Protect Your Money — and What to Do When It Flags Something Source: https://aiforregulars.com/guides/how-your-bank-uses-ai-for-you Your bank's AI is running in the background on every purchase you make — checking whether the transaction looks like something you would do, and deciding in a fraction of a second whether to approve or block it. Most of the time you never notice it working. Occasionally it gets in your way at exactly the wrong moment. Understanding what it's doing, and knowing what to do when it flags something you actually meant to buy, can save you a lot of frustration. ## How Your Bank's Fraud AI Watches Your Spending Every time you tap your card or pay online, your bank doesn't just check whether you have the money — it checks whether the purchase looks like you. The AI knows roughly what you usually spend, where you usually spend it, what time of day your transactions tend to happen, and even what types of merchants you normally use. When a purchase falls outside those patterns — a hotel in a city you've never visited, a large electronics purchase from an unfamiliar retailer, three gas station fill-ups in different states on the same day — the system raises a flag. This kind of pattern-based fraud detection now runs at almost every major financial institution. According to Feedzai's 2025 AI Trends Report, 90% of financial institutions worldwide use AI for fraud detection. The reason is straightforward: Celent research from March 2026 found that 82% of lenders reported rising fraud losses, with about a third seeing double-digit year-over-year increases. AI detection is the tool that keeps fraud from being even worse than it already is. The tradeoff is false positives — legitimate purchases that look suspicious to the machine. Your card declined at dinner in Rome is the machine doing its job. It just can't always tell you from a thief. ## When Your Card Gets Declined or Your Account Is Frozen If your card is declined unexpectedly, check your bank's app first. Most banks now send a real-time push notification when they block a transaction, and many let you tap a button to confirm the purchase was yours and immediately re-approve it — no phone call needed. If your account is actually frozen rather than just a single declined card, you'll need to call. Use the number on the back of your card or on your official bank statement — never a number from a pop-up, an email, or a caller who reached out to you first. Be ready to confirm your identity with your card number, a recent transaction amount, and whatever security question your bank uses. In straightforward cases, a brief call resolves access the same day. **One thing that helps in advance:** if you know you'll be making a large or unusual purchase — a piece of jewelry, a car deposit, a hotel booking across the country — a quick message to your bank through its app beforehand can prevent the block from happening at all. Most bank apps have a way to flag an upcoming unusual purchase. ## The AI Chat Assistant: What It Can and Can't Do Bank of America's AI assistant Erica handles around 2 million customer interactions per day. Assistants like Erica have become genuinely capable at routine tasks: checking your balance, finding a specific transaction, making a transfer, locating the nearest ATM, setting up alerts, or walking you through a fee on your statement. For things that fit neatly into a standard category, they work well. Where they struggle is anything requiring human judgment: a disputed charge from a merchant you actually visited but were overcharged by, a fraud complaint with nuance, or an account freeze that doesn't fit a simple "confirm or deny" pattern. Most bank assistants know their limits here and are quite good at routing you to the right human department quickly. When I need to reach a real person at my bank, I skip the full sentence and just type the one word that describes the problem: "dispute," "fraud," or "freeze." Most bank chat systems treat those as priority keywords and connect you to a specialist faster than if you explain the full situation first. The assistant doesn't need context — it just needs the category. ## When You Open a New Account: Deepfake Detection and Video Verification If you've opened a bank account online recently and been asked to hold up your ID and record a short selfie video, that's the bank's AI running an identity check. The system examines your ID's security features, compares the photo to your live face, and looks for natural human signals — blinking, slight movement, the micro-variations that a still photo or an AI-generated deepfake tends to lack. Banks have built this layer because digital account fraud is a serious and growing problem, and deepfake technology has made it cheaper to fake an identity. About 30% of enterprises expect biometric authentication alone to become unreliable as deepfakes improve, which is why banks are combining video checks with document verification rather than relying on a face match alone. If the check fails during account opening, it's usually not a sign of a problem with your identity. Poor lighting, a glare on your ID, or holding the document at too sharp an angle are the most common causes. Try again with better light aimed at the ID, or ask whether you can complete the step at a branch. ## Pre-Empting False Alarms Before They Happen The most effective thing you can do to prevent false fraud flags is tell your bank about changes before they happen rather than explaining yourself after. **Set a travel notice.** Almost every bank now has this in its app: you enter the dates and destination of your trip, and the fraud system lowers its alert threshold for that region during that window. It takes about two minutes and prevents most vacation card declines. **Turn on real-time transaction alerts.** If every purchase above a certain amount sends you an instant notification, you'll catch actual fraud nearly as fast as the bank does — and you'll have instant confirmation that your own purchases went through cleanly. **Carry a backup card from a different bank.** Travel notices reduce false flags significantly, but they don't eliminate them. A second card on a different network (Visa if your main card is Mastercard, for example) means you're never completely stranded if one is blocked. ## The One Rule That Never Changes: When "Bank Security" Contacts YOU This is the most important section in this guide. Your bank's real fraud team will never call you and ask for your PIN, your online banking password, your full card number, or a one-time verification code. What they do ask is whether a specific transaction is yours — that's the full extent of it. Scammers impersonating bank fraud departments are among the most effective phone scams running right now, and increasingly they use AI-generated voices that sound indistinguishable from a real person. The call will sound urgent, professional, and completely legitimate. If anyone who contacts you — by phone, text, or email — asks you to provide a code, a password, or anything beyond the last four digits of your card, it is a scam regardless of how convincing it sounds. Hang up. Call your bank at the number printed on your card. If you receive a text with a link, don't tap it — go directly to your bank's app or website. For a detailed look at how these scams work, [AI Voice Cloning Scams: What They Sound Like and How to Spot Them](/guides/ai-voice-cloning-scams) and [How to Recognize a Fake Customer Service Scam](/guides/fake-customer-service-scam) cover the tactics in full. ## What to Watch Out For **The AI isn't perfect in either direction.** It sometimes flags legitimate purchases and sometimes misses sophisticated fraud — especially when criminals study a victim's transaction patterns before striking. Fraud detection reduces risk; it doesn't eliminate it. **Privacy:** To build your spending profile, your bank stores a detailed record of your transaction history, locations, and behavioral patterns. This is how the protection works, and most banks disclose it in their privacy policy — but it's worth knowing that your spending habits are analyzed in more detail than most people realize. **AI decisions aren't always explainable.** If your account is frozen and neither the app nor the first representative can tell you why, ask to speak directly with the fraud specialist on the case. They typically have access to the specific reason the system triggered. ## What to Try Next The other side of understanding your bank's AI is recognizing what the scams that impersonate it look like. [How to Detect an AI-Generated Voice on a Phone Call](/guides/detect-ai-voice-phone-call) walks through the specific audio tells that give fake "bank security" callers away. If you'd like AI to actually help you manage your money rather than just protect it, [Best AI Budgeting Apps: What They Do and What to Expect](/guides/best-ai-budgeting-apps) is the natural next read. --- # How to Spot Fake Hurricane Videos and AI Disaster Images Before You Share Them Source: https://aiforregulars.com/guides/hurricane-weather-fake-videos When a hurricane is approaching, your social media feed fills with alarming footage before the storm even makes landfall. Some of it is real. Some is AI-generated. And some is years-old video from a completely different storm being recycled as if it is happening right now. Telling them apart takes about 30 seconds, and it matters more than you might expect: fake evacuation videos have triggered real panic, and donation scams reliably follow major disasters by just hours. ## Why This Problem Gets Worse Every Hurricane Season AI video and image tools have made it dramatically cheaper and faster to generate convincing-looking disaster footage — and the newest models have made the visual tells in Step 3 below less reliable than they used to be. By 2026, widely available tools like Google's Veo 3.1 and OpenAI's Sora 2 can generate realistic water and object physics along with synchronized native audio, including dialogue. That means a clip can pass a casual visual check and still be entirely fake, which is exactly why checking official sources first (Step 1) and reverse image search (Step 2) matter more than eyeballing the footage. During the 2025 hurricane season, AI-generated flooding videos spread widely on social media before storms made landfall, circulating alongside real coverage and making it difficult for people to know what was actually happening and where. This is not only about AI, and it isn't new, either. During Hurricanes Helene and Milton in 2024, Russian and Chinese-linked accounts amplified false claims that FEMA was diverting disaster funds to migrants and that the government intended to seize evacuees' land — claims that fact-checkers and FEMA itself debunked, but which discouraged some people from evacuating. Russian state media also circulated an AI-generated image of Walt Disney World flooded by Hurricane Milton that reportedly drew hundreds of thousands of views before it was exposed as fake. FOX10's hurricane misinformation guide, published in June 2026, highlighted the same recurring pattern: old or AI-generated images passed off as live storm footage. Researchers writing in the Bulletin of the Atomic Scientists have warned that AI-generated misinformation now poses a genuine threat to emergency communications — when people cannot tell real footage from fake, they may ignore genuine warnings or delay evacuating. ## Step 1: Check Official Sources Before Anything Else When you see alarming hurricane footage, open one of these official sources before you watch it again or share it: - **National Hurricane Center (nhc.noaa.gov):** The definitive source for storm track, intensity, and official warnings. Updated every six hours, more often as a storm approaches land. - **NOAA / National Weather Service (weather.gov):** Your local office issues the watches and warnings that apply specifically to your area. - **Local TV meteorologists:** Local stations broadcast the most accurate, localized information during a hurricane — their meteorologists are trained to interpret official data for where you actually live. - **Your county or city emergency management agency:** This is where real evacuation orders and shelter locations get posted, not social media. - **FEMA (fema.gov, or the FEMA app):** Publishes disaster declarations and, after major storms, a "rumor control" page that directly addresses specific viral claims — the FEMA-fund-diversion and land-seizure rumors that circulated during Hurricanes Helene and Milton, for instance, were addressed there. If the alarming video you are looking at is not being mentioned by any of these sources, that is your first signal that something may not add up. I keep the National Hurricane Center website bookmarked on my phone's home screen from June through November. When a scary storm video comes up in my feed, I check there first — not because I distrust everyone online, but because the official source is always faster than trying to figure out whether the video is real. ## Step 2: Reverse Image Search Suspicious Footage Reverse image search is the fastest way to find out if a photo or video is being recycled from an old storm. For photos, Google Images and TinEye both let you upload an image or paste a URL to find where it has appeared online before. If a "current" hurricane photo shows up in results from 2017, or from a storm in a completely different part of the world, you have your answer. For video clips, take a screenshot of a distinctive frame — one showing landmarks, street signs, or text — and search that image. > On your phone: press and hold a video to copy its link, then go to images.google.com, tap the camera icon, and paste the URL. ## Step 3: Look for AI Visual Tells in the Footage AI-generated video has gotten more convincing, but it still makes characteristic mistakes — treat this step as a supporting check, though, not your main line of defense. Newer models like Veo 3.1 and Sora 2 pass most of the checks below on their best outputs, so a clean-looking clip is not proof of anything on its own; Steps 1 and 2 (official sources, reverse image search) carry more weight. **Unnatural water behavior.** Water is genuinely hard for AI to simulate correctly. In real hurricane flooding, water moves with physics — turbulence, debris, realistic eddies. AI flooding often looks oddly smooth, with waves that repeat in a pattern or water that does not interact realistically with objects around it. **Distorted buildings, signs, and text.** AI video generators struggle to keep architecture consistent across frames. Windows may shift or disappear, street signs may have blurred or garbled letters, and building edges may waver or morph slightly between shots. **Wrong geography.** A video claiming to show flooding in a specific city but featuring the wrong landscape, wrong vegetation, or a skyline that does not match is either fake or badly mislabeled. **Distorted people and vehicles.** Figures in AI-generated footage often have extra fingers, limbs that bend at wrong angles, or faces that blur when turning. Vehicles may have inconsistent reflections or change shape subtly between frames. **Too cinematic for found footage.** Real disaster footage looks like phone video — shaky, poorly lit, shot from inside a building or car. If a clip looks too clean, too well-framed, or too dramatically lit for what is supposed to be someone's backyard during a hurricane, look closer. ## Step 4: Check Whether the Footage Is Old and Recycled Genuine footage from real past storms gets recycled constantly because it looks dramatic — and real. Old Hurricane Harvey or Katrina footage reappears almost every active storm season. **Search the clip description along with the storm name.** If the same video shows up labeled as multiple different hurricanes across multiple years, it is recycled content. **Look for local landmarks.** If you can identify a street sign, business name, or recognizable building in the footage, a quick search can tell you whether that location is even in this storm's path. **Check seasonal clues.** Footage showing trees in full summer leaf during a November storm, or winter conditions during a July hurricane, was filmed at a different time of year. ## Your 30-Second Before-You-Share Checklist Before sharing any hurricane video or disaster image, run through this: 1. **Is an official source — NHC, NOAA, local emergency management — saying the same thing?** 2. **Does a reverse image search come back clean, with no results from other storms?** 3. **Does the water and environment behave in a physically realistic way?** 4. **Does the geography match where this storm is actually located?** 5. **Is there a credible news organization or official account backing it up?** If you cannot say yes to most of these in 30 seconds, hold off before sharing. ## What to Watch Out For These checks catch most fakes, but not all of them. Skilled disinformation campaigns sometimes combine real footage from a real storm with false context — a genuine clip, posted with the wrong location or date. Reverse image search helps but is not foolproof. Watch out for the emotional pull to share quickly. Scary videos are designed to trigger fast, urgent reactions. That feeling of "people need to know about this right now" is exactly when it is worth pausing for a 30-second check. Donation scams are another risk: after a major hurricane, fake charity accounts appear on social media within hours. Before donating, it's worth running two separate checks, because they answer different questions: Charity Navigator (charitynavigator.org) rates how efficiently a charity spends its money and how transparent it is, while the IRS Tax Exempt Organization Search (irs.gov) confirms the organization is a real, currently registered 501(c)(3) — a strong Charity Navigator rating means little if the organization doesn't turn up in the IRS database at all. The same habits that help you catch a fake video apply to fake charities — [our guide to spotting fake AI news and fact-checking claims](/guides/ai-fake-news-fact-check) covers that pattern in detail. ## What to Try Next For a broader look at AI-generated video across any topic — not just disasters — [How to Spot a Deepfake Video](/guides/how-to-spot-deepfake-video) walks through the same visual tells applied to faces and news footage. And if you want to get sharper at catching AI-generated still images anywhere in your feed, [How to Tell If a Photo Was Made by AI](/guides/spot-ai-generated-photos) covers the photo-specific version of these same checks. ## Sources - [National Hurricane Center](https://www.nhc.noaa.gov) — NOAA - [FEMA](https://www.fema.gov) — U.S. Federal Emergency Management Agency - [To drive Americans apart, Russia helped spread hurricane disinformation](https://www.pbs.org/newshour/world/to-drive-americans-apart-russia-helped-spread-hurricane-disinformation) — PBS NewsHour - [Russian state media uses AI-generated images of Florida's Disney World flooded by Milton](https://www.voanews.com/a/russian-state-media-uses-ai-generated-images-of-florida-s-disney-world-flooded-by-milton/7819396.html) — VOA News - [AI misinformation is threatening emergency communications. Here's how to fix that](https://thebulletin.org/2025/09/ai-misinformation-is-threatening-emergency-communications-heres-how-to-fix-that/) — Bulletin of the Atomic Scientists - [Don't get fooled by hurricane hype this season](https://www.fox10tv.com/2026/06/02/dont-get-fooled-by-hurricane-hype-this-season-how-spot-misleading-hurricane-information-online/) — FOX10 News - [Charity Navigator](https://www.charitynavigator.org) — Charity Navigator - [Tax Exempt Organization Search](https://www.irs.gov/charities-non-profits/tax-exempt-organization-search) — Internal Revenue Service --- # What Does 'AI Info' Mean Under a Facebook or Instagram Photo? Source: https://aiforregulars.com/guides/instagram-facebook-ai-label-meaning If you've scrolled past a photo on Facebook or Instagram lately and noticed a small "AI info" tag tucked under the username or inside the post's menu, you're not imagining it — and you're not alone in wondering what it means. The confusing part is that it sometimes shows up on photos that look completely ordinary and real: a birthday party, a sports photo, a family snapshot. Here's what's actually going on. ## What "AI Info" actually is "AI info" is a label Meta applies automatically across Facebook, Instagram, and Threads, most often when it detects that a photo or video carries an industry-standard technical signal called **Content Credentials**, built on a standard known as C2PA. In that case, Meta doesn't need the person posting to tell it anything — the system reads hidden data embedded in the file itself, invisible to anyone just looking at the image. Meta also runs its own AI-detection systems as a backup for content with no embedded signal, but the metadata path is what causes most of the confusing, real-photo cases below. That data can say a few different things: that the whole image was generated by an AI tool like DALL-E or Midjourney, or simply that *some part* of it — even one small edit — was made using an AI-powered feature in an editing app. Meta treats both cases the same way: it shows the label. The label is genuinely easy to miss, and that's by design in a way — Meta moved it into the three-dot menu on the post ("About this photo" / post details) rather than displaying it prominently under every username, after early versions of the feature felt too intrusive. If you want to see it, look for the small "i" or three-dot menu on a post rather than expecting it to jump out at you. ## Why real, unedited-looking photos get flagged This is the part that trips people up, and it has already caused real friction. In 2024, former White House photographer Pete Souza had a completely genuine basketball photo tagged as AI-made on Facebook. He traced the cause back to a change in how Adobe's cropping tool handled the image before export — not to anything he'd deliberately generated with AI. Other photojournalists reported the same thing happening to real, unaltered news photography. The reason is that Meta's system checks for the presence of the AI signal, not how much of the picture it actually affected. Photo-editing apps — Adobe's Photoshop and Lightroom are the most common examples — can automatically embed that signal the moment certain AI-assisted tools are used, including features many people don't think of as "AI" at all, like the AI-powered fill Photoshop uses to complete a crop. It's not just editing apps anymore. Google's Pixel 10, released in 2025, signs photos taken with its own camera app using Content Credentials by default, and Samsung's Galaxy S25 adds the signal specifically when a photo has been through one of the phone's AI editing features. Expect more phone makers to follow as the underlying standard matures. A photo generated from scratch by an AI image tool and a real photo that only had its background lightly cleaned up can both end up wearing the same "AI info" tag. If you want to know how much of an image is actually synthetic, the label alone won't tell you — you have to look at the image itself. The flip side matters just as much: an image can be entirely AI-generated and show no label at all, either because the tool that made it doesn't embed Content Credentials, or because the signal was stripped out somewhere along the way — something Meta itself has acknowledged is possible. Treat the absence of a label as no guarantee either. ## Other apps are doing the same thing Facebook and Instagram aren't alone here. TikTok and YouTube have both adopted the same Content Credentials standard to automatically detect and label AI-touched images and video, on top of letting creators disclose AI use themselves when they post. The exact wording and where the tag shows up varies by app, but the underlying idea is shared across all of them: flag content where an AI tool was involved somewhere in the process, and let the viewer decide how much that matters. ## What to actually do when you see it **Don't treat it as proof the photo is fake.** In most everyday cases — a friend's vacation photo, a family picture — the far more likely explanation is a minor edit made in an app that happens to embed the signal by default, not an attempt to deceive anyone. **Do pay more attention when the stakes are higher.** If a photo is attached to a news claim, a product review, a stranger's dating profile, or anything you might act on, the label is a reasonable cue to slow down and look more closely at the image itself, not just trust or dismiss it based on one small tag. **Check the post's details if you want to know more.** Tapping the three-dot menu or "About this photo" on Facebook or Instagram sometimes shows additional context about what was detected, depending on the app version. **Use your own eyes as the real check.** Since the label can both over-flag real photos and miss genuinely fake ones, the visual signs that have always mattered — hands, background repetition, garbled text, mismatched shadows — are still your best tool. [How to Spot AI-Generated Photos](/guides/spot-ai-generated-photos) walks through exactly what to look for. ## What to Try Next If you're curious what else Meta AI can actually do inside Facebook and Instagram beyond this labeling system, [Meta AI on Facebook and Instagram: What It Can Do for You](/guides/meta-ai-facebook-instagram) is a good next stop. And if fake profiles rather than photo authenticity are your bigger worry, [How to Spot AI-Generated Fake Profiles and Posts on Facebook](/guides/ai-fake-facebook-profiles) covers that territory directly. --- # Is That App Actually AI, or Just a Scam? Red Flags to Know Source: https://aiforregulars.com/guides/is-ai-app-scam The AI boom has been great for a lot of people — and great for scammers too. Hundreds of apps claiming to be "AI-powered" have flooded the App Store and Google Play. Some are simply charging you for a service that is free elsewhere. Others are harvesting your personal information or locking you into expensive subscriptions before you can figure out what is happening. Knowing what to look for takes about five seconds — and it can save you real money. ## How Fake AI Apps Work Most fake AI apps fall into one of three traps. The first is charging for something free: ChatGPT, Google Gemini, and Microsoft Copilot all have free tiers, but clone apps will sell you access for $9.99 a week. The second is data harvesting: the app looks like a chatbot but its real purpose is collecting your name, email, location, and anything you type into it. The third is the subscription trap: the app pushes you toward a paid plan in the first 30 seconds, then makes cancellation confusing or nearly impossible to find. ## Red Flag 1: No Clear Company Behind the App Every legitimate app has a visible developer name, a working privacy policy link, and a support contact that goes to a real business. If you open an app's store listing and the developer is a vague name like "AI Tools Studio LLC" with no website and a privacy policy that is missing or clearly copy-pasted from a generic template — walk away. A real company stands behind its product and makes it easy to reach them if something goes wrong. ## Red Flag 2: Immediate Subscription Pressure Legitimate apps let you try the product before asking for money. If an app opens and immediately shows you a paywall or a countdown timer before you have typed a single message, it is designed to capture your credit card number — not to help you. Many of these apps rely on people tapping "Start Free Trial" without noticing that it auto-renews at a weekly rate. Read the fine print before you tap anything. ## Red Flag 3: Asking for Unusual Permissions A writing assistant or AI chatbot has no legitimate reason to access your contacts, read your SMS messages, or activate your camera. When an app requests permissions that do not match what it does, it is almost certainly using those permissions to collect data it can sell or misuse. Decline any permission that does not make obvious sense for that specific app's purpose, and reconsider downloading it at all. I once almost downloaded a fake "ChatGPT" app before I caught myself. The icon looked nearly identical to the real one — same color scheme, same style. But when I tapped the developer name in the App Store, it said something like "AI Chat Assistant Inc." The real ChatGPT app on iOS is published by OpenAI. That one check took about five seconds and saved me from what turned out to be a $39.99-per-month subscription trap. I now check the developer name on every AI app before I download, no exceptions. ## Red Flag 4: Results That Look Nonsensical or Copy-Pasted Real AI gives you coherent, contextual answers that actually respond to what you asked. Fake apps often use simple response templates or scraped text that has nothing to do with your question. If the output feels canned, repetitive, or completely off-topic, the app is probably not running any real AI at all — it is just pretending. Ask it something specific and see if it answers you or gives back something generic. ## Red Flag 5: Suspicious Names or URLs Scammers count on you not reading carefully. Watch for slight misspellings: "ChatGBT," "Gemeni," or "CoPilot AI Pro." In app stores, look skeptically at names like "ChatGPT Pro Plus Ultra AI" — legitimate apps do not pile on superlatives. Outside app stores, always check the URL character by character before you enter any information. One swapped letter is all it takes. ## How to Find the Real Apps Go directly to official websites rather than searching in your app store: chatgpt.com for ChatGPT, gemini.google.com for Google Gemini, and copilot.microsoft.com for Microsoft Copilot. Each of those pages links directly to the correct app store listing. If you do use the app store search, verify the developer name exactly: OpenAI for ChatGPT, Google LLC for Gemini, Microsoft Corporation for Copilot. Any variation in that developer name is a red flag, not a coincidence. ## What to Do If You Are Already Subscribed Do not try to cancel through the app itself — scam apps often make that button invisible or nonfunctional on purpose. Instead, go directly to your phone's subscription settings. On iPhone, that is Settings → your name → Subscriptions. On Android, open the Google Play Store → tap your profile → Payments and subscriptions. Find the subscription, and cancel it from there. If you were charged for a service that was advertised as free, you can dispute the charge with your bank or credit card company as a misleading transaction — most banks will take that seriously. ## What to Try Next Fake apps are one piece of a bigger picture. Learn how scammers use AI to steal your voice and your identity in [how AI voice cloning scams work](/guides/ai-voice-cloning-scams). And if you have seen suspicious videos or images online, [how to spot a deepfake video](/guides/how-to-spot-deepfake-video) will help you tell the real from the fake. ## Sources - [Request a refund for apps or content that you bought from Apple](https://support.apple.com/en-us/118223) — Apple - [Impersonation](https://support.google.com/googleplay/android-developer/answer/9888374?hl=en) — Google Play Console Help - [FTC Announces Crackdown on Deceptive AI Claims and Schemes](https://www.ftc.gov/news-events/news/press-releases/2024/09/ftc-announces-crackdown-deceptive-ai-claims-schemes) — Federal Trade Commission --- # Is AI Bad for the Environment? What the Numbers Actually Show Source: https://aiforregulars.com/guides/is-ai-bad-for-the-environment If you've seen headlines saying a single ChatGPT question "drinks" a bottle of water, or that AI is quietly draining the power grid, it's fair to wonder whether you should feel bad about using it. The honest answer is more reassuring than the scariest headlines, but also more complicated than "don't worry about it." Here's what's actually been measured, and by whom. ## Where the Scary Numbers Came From Much of the alarm traces back to a 2024 estimate from a UC Riverside researcher, widely reported as: writing a short, roughly 100-word response with ChatGPT could consume around 500 milliliters of water — about a standard bottle. That number spread fast because it's vivid and easy to picture. There's an important update to that story that gets shared far less often. In 2026, the same researcher, Shaolei Ren, revised his own estimate down significantly — to roughly 15 milliliters for a typical modern prompt — pointing to more efficient chips and cooling systems, and noting that the original figure was based on specific 2024 assumptions that don't reflect how today's systems run. That's a more than 30-fold drop in the estimate, from the same source. ## What AI Companies Themselves Have Disclosed The most solid numbers available are the ones companies have published about their own systems, since they can measure their actual hardware directly instead of estimating it from outside. In 2025, Google released a technical report on the environmental cost of a typical Gemini Apps prompt: about 0.24 watt-hours of electricity and about 0.26 milliliters of water — roughly five drops. For comparison, Google says that's about what a TV uses in under nine seconds. OpenAI CEO Sam Altman published a similar breakdown for ChatGPT the same year: an average query uses about 0.34 watt-hours of electricity and a tiny fraction of a gallon of water, which he likened to about one second of an oven running. Those figures aren't independently audited, and both companies chose how to define an "average" query. But they're a real data point from the source, and they land in the same small ballpark as each other — nowhere near a bottle of water per question. ## So Why Do the Numbers Vary So Much? Because "how much water does AI use" isn't one question — it's several, and each study answers a different one. Some estimates count only the water evaporated to cool the computer chips right there in the data center. Others add the water used miles away, at the power plant generating the electricity the data center runs on — which can be a much bigger number depending on how that electricity is produced. Prompt length, which AI model answers it, and how efficient a given data center is all shift the result further. None of the studies are necessarily "wrong" — they're measuring different slices of the same system, which is exactly why the same underlying question can produce numbers 30 times apart. ## The Part That Actually Matters More: Scale Here's the more useful way to think about it. A single question to an AI chatbot genuinely is small — even the higher estimates amount to a few sips, not a bottle, per typical use. But AI is being asked billions of times a day, and more importantly, tech companies are building enormous new data centers specifically to run it. The International Energy Agency, which tracks global energy use, estimates data centers currently account for roughly 1.5 to 2% of global electricity consumption — and that AI-focused data centers saw their electricity use jump about 50% in 2025 alone, far outpacing data center growth overall. That share is projected to approach 3% of global electricity by 2030. In some places, new data centers are also drawing on local water supplies in regions that are already short on water, which is a legitimate concern for the communities nearby, separate from what any single chat "costs." In other words: the honest environmental story about AI isn't really about you asking a question. It's an infrastructure story — about how fast these data centers are being built, where their power comes from, and how transparent companies are about both. When I want to know if a scary AI statistic is solid, I check whether it comes with a methodology — does it say what's included (on-site cooling only, or the power plant too?) and when the estimate is from. A number with no date and no method attached, especially one that's gone viral, is often outdated or measuring something narrower than the headline suggests. ## What You Can Actually Do With This If this is something you care about, the individual-chat-by-chat approach isn't where your effort matters most — the per-query numbers are small enough that trimming your own usage won't move the needle much. What does help, if you want to act on it: support and favor companies that publish real efficiency data rather than staying silent, and keep an eye on local reporting if a data center is proposed near you, since that's the level where water and power use genuinely add up to something worth weighing in on. **What to try next:** For a broader look at how these tools actually work under the hood, [What Does 'AI' Actually Mean?](/guides/what-does-ai-mean) is a good next stop. And if you want to sort AI facts from AI hype more generally, [Common AI Myths, Debunked](/guides/ai-myths-debunked) covers more of the claims worth double-checking. ## Sources - [Measuring the environmental impact of AI inference](https://cloud.google.com/blog/products/infrastructure/measuring-the-environmental-impact-of-ai-inference) — Google Cloud Blog - [In a first, Google has released data on how much energy an AI prompt uses](https://www.technologyreview.com/2025/08/21/1122288/google-gemini-ai-energy/) — MIT Technology Review - [Sam Altman: ChatGPT queries consume 0.34 watt-hours of electricity and 0.000085 gallons of water](https://www.datacenterdynamics.com/en/news/sam-altman-chatgpt-queries-consume-034-watt-hours-of-electricity-and-0000085-gallons-of-water/) — Data Center Dynamics - [Shaolei Ren revises widely cited AI water estimate from a full bottle to 15 milliliters per prompt](https://digg.com/ai/ncm5zxj6) — Digg - [Data centre electricity use surged in 2025, even with tightening bottlenecks driving a scramble for solutions](https://www.iea.org/news/data-centre-electricity-use-surged-in-2025-even-with-tightening-bottlenecks-driving-a-scramble-for-solutions) — International Energy Agency --- # Is ChatGPT Free? Plans, Limits and What You Actually Get Source: https://aiforregulars.com/guides/is-chatgpt-free ChatGPT is free to sign up for and free to use every day for most common tasks. No credit card, no trial period that suddenly charges you. That said, the free plan does have limits — and there are paid options for people who need more. Here is a clear breakdown of what each plan actually includes, so you can decide whether the free version is enough for you. ## What the Free Plan Includes The free plan gives you access to text conversations with ChatGPT. You can type a question or request, read the response, and keep asking follow-up questions in the same conversation. For most everyday tasks — drafting emails, explaining a concept, brainstorming ideas, summarizing text — the free plan works well. A few things to know about free plan limits: - **Usage limits per session or day.** If you send many messages in a short period, ChatGPT may ask you to wait before sending more. Limits reset automatically, usually within an hour. - **Slower during busy periods.** When a lot of people are using ChatGPT at once, free users may experience slower responses or a brief wait. - **Unlimited text chats on the free-tier default model.** Free users get unlimited text conversations, with no message cap, on OpenAI's free-tier default model — GPT-6 Luna as of October 2026 — not GPT-6 Sol, which is limited to paid plans. - **Limited image generation and web search.** Free users can generate a small number of images per day and run web searches, but both are subject to tighter daily caps than paid plans. ## What Paid Plans Add OpenAI offers paid tiers beyond the free plan. The most common one for individuals is **ChatGPT Plus**, which is a monthly subscription. | Feature | Free | ChatGPT Plus | |---|---|---| | Text conversations | Yes | Yes | | Access to latest AI model | Limited | Priority access | | Response speed | Standard | Faster, especially when busy | | Image generation | A few images/day | ~50 images per 3 hours | | File and document uploads | Limited | More capacity | | Web browsing / search | Limited | More access | | Custom instructions | Yes | Yes | | Monthly cost | $0 | $20/month | OpenAI also offers **ChatGPT Go** at $8/month, a mid-tier option that unlocks 10x more messages and image generations than the free plan, without the full Plus feature set. Go does not include Deep Research, Advanced Voice, or Sora. There are also **Business** and **Enterprise** plans aimed at businesses and organizations, with additional features like larger file limits, admin controls, and dedicated capacity. Those are well beyond what most individual readers need. I tested the free plan for several weeks before deciding to upgrade. For writing emails, explaining documents, and answering everyday questions, it held up just fine. I noticed the limits mainly when I was doing a lot of back-and-forth on a project in one sitting. If you are just starting out, begin with free — you can always upgrade later. ## When the Free Plan Is Enough The free plan handles most common tasks without any trouble: - Drafting and editing emails or letters - Getting a plain-language explanation of a confusing document, term, or bill - Brainstorming ideas for gifts, meals, or projects - Summarizing a long article or list of notes - Practicing for a job interview or writing a cover letter - Learning how something works (a medication, a law, a concept) If you use ChatGPT occasionally — a few times a week — the free plan will rarely hit its limit. ## When Upgrading Makes Sense Consider a paid plan if you: - Use ChatGPT heavily every single day for work tasks - Need to generate images directly in ChatGPT - Upload large documents or files regularly - Notice that the free plan keeps interrupting your workflow with "limit reached" messages - Need the fastest, most capable model for complex writing or analysis ## How to Sign Up for Free Go to **chatgpt.com** and click "Sign up." You can create an account with your email address or by connecting a Google or Microsoft account. The whole process takes about two minutes. See [how to use ChatGPT](/guides/how-to-use-chatgpt) for a full step-by-step walkthrough. If you decide you want to upgrade later, you can do so from the same account — your conversations and settings carry over. ## What to Try Next Start with the free plan and put it to work on something real. Our [beginner's guide to using ChatGPT](/guides/how-to-use-chatgpt) walks you through your first conversation in under 10 minutes. If you are curious how ChatGPT compares to Google's AI, [ChatGPT vs. Gemini](/guides/chatgpt-vs-gemini) breaks down which one fits which situations better. --- # My Kid Uses ChatGPT for Homework — A Parent's Guide Source: https://aiforregulars.com/guides/kids-chatgpt-homework-parents-guide You're checking the browser history — maybe looking for something else entirely — and there it is. ChatGPT. Your kid's math homework folder is open in another tab. Your first feeling is probably somewhere between annoyed and worried. Take a breath. This is worth handling carefully, because how you react in the next hour can shape how your child relates to AI for years. ## First: What Were They Actually Doing? Before assuming the worst, look at what you can see. AI gets used in a lot of ways, and not all of them are problematic. On one end, there is clear cheating: asking ChatGPT to write an essay and submitting it word for word. This is academically dishonest in almost every school context, and it also means the student missed whatever skill-building the assignment was meant to do. On the other end, there is completely reasonable use: asking ChatGPT to explain a concept that was confusing in class, checking whether a math approach makes sense, or brainstorming ideas before actually writing. This is not much different from looking something up online. In the middle, there is a wide gray zone. Asking AI to outline an essay and then writing it yourself. Having AI check grammar. Getting AI to generate examples and then explaining them. Whether this is okay depends on the assignment, the teacher's expectations, and your family's values. Most kids who use AI for homework are somewhere in that gray zone. Very few are engaged in full-scale cheating. Most are just trying to get through an assignment more easily — which is also what adults do. ## The Conversation to Have (Not the Lecture) Resist the urge to walk in and start talking. Start by asking. "Hey, I noticed you were on ChatGPT while you were doing homework. Can you tell me how you were using it?" Then listen. Do not interrupt. Do not immediately categorize what they say as okay or not okay. Just get the picture first. After they explain, some useful follow-up questions: - "Does your teacher know you used it? What do they think about that?" - "Can you explain to me what your homework is actually about?" - "What did you learn doing this assignment?" The last question is the most important one. The point of homework is not to produce a document — it is to build knowledge and skills. If your kid can walk you through the ideas in their own words, they learned something. If they have no idea what they submitted, they did not. When I first started thinking about how I would handle this with a younger person, I realized I needed to be honest: I use AI assistants myself, almost every day, for work tasks. The question is not "should we use AI" — it is "are we using it in a way that makes us better or lazier?" That framing makes the conversation feel more like a genuine discussion and less like a punishment. ## Learning vs. Cheating: The Practical Line Here is a simple way to think about it that you can share with your kid: **AI is a tool, like a calculator.** A calculator is fine in math class because the goal is to understand math, not to become fast at arithmetic. But if the goal of the assignment is to practice the calculation itself, using a calculator defeats the purpose. AI works the same way. If an assignment is meant to build writing skills, having AI write it defeats the purpose. If an assignment is meant to test whether you understand a topic, having AI answer for you means you skipped the test. The rule of thumb: **would you be comfortable if your teacher knew exactly how you used AI on this?** If yes, it is probably fine. If no, rethink. ## What to Do If It Was Clearly Cheating If your kid submitted AI-written work as their own, on an assignment where that was not allowed, they need to know that is a problem — and why. The reason it matters is not abstract. When students skip the work of learning, they accumulate gaps. Those gaps show up later, when the topic comes up again in a harder class, on a test, or in a real situation where the skill actually matters. A few practical steps: 1. Let them know you saw it and that it matters. Keep the tone serious but not panicked. 2. Ask them to redo the assignment on their own, even if they cannot turn it in again. The redo is about learning, not grades. 3. Talk about your family's expectations going forward. Be specific — not "don't use AI," but "for essays and writing assignments, we expect you to write your own drafts." 4. Consider whether to contact the teacher. This is a judgment call. If the assignment was graded, the teacher may need to know. If it was practice work, you may decide to handle it at home. ## Conversation Starters by Age **For middle schoolers (10–13):** "AI tools are really useful, but they can also do your thinking for you — and that is the part you actually need to do yourself. What do you think the difference is between using AI to help and using it to cheat?" **For high schoolers (14–17):** "I am not mad, but I want to understand how you are using this. Because AI is going to be part of your work life too — and knowing when to use it and when not to is actually a real skill. Let's talk about where you think the line should be." ## Setting Ground Rules That Actually Stick Rules that students help create are more likely to be followed than rules handed down from above. Consider having a short family conversation where you make the policy together. Things to decide: - Is AI okay for research and understanding, but not for writing drafts? - Does using AI need to be disclosed to the teacher? - Is there a difference between different subjects or assignment types? - What happens if the rule gets broken? Write down what you agree on. Review it when a new school year starts, because the tools change and school policies change too. ## What to Try Next If you want to set up practical safety controls on ChatGPT for a younger user, [ChatGPT Parental Controls and Safe Setup for Kids](/guides/chatgpt-parental-controls) walks through the current options step by step. If you are curious about AI detectors — and why they probably should not be the thing teachers rely on — [AI Detectors Tested](/guides/ai-detectors-tested) has a clear-eyed look at how they actually perform. --- # How to Learn a Language with ChatGPT: Daily 15-Minute Routine Source: https://aiforregulars.com/guides/learn-language-with-chatgpt You do not need a classroom, a tutor, or a lot of free time to make real progress in a new language. Fifteen focused minutes a day with ChatGPT can move the needle noticeably — if you use those minutes well. Here is a daily routine you can start today. The whole thing takes about 15 minutes and covers vocabulary, reading, conversation, and error correction in a single session. At the start of each session, send one setup message that tells ChatGPT exactly how to behave. You only need to write this once — copy it into a note and paste it at the start of each new chat. ``` You are my [LANGUAGE] tutor. My current level is [beginner / intermediate / advanced]. My native language is English. Please follow these rules for our whole session: 1. Respond in [LANGUAGE] unless I ask you to switch to English. 2. After each of my messages, gently correct any grammar or vocabulary mistakes and explain the correction in one sentence. 3. Keep your sentences short and clear — matched to my level. 4. If I ask for a word in English, give me [LANGUAGE] + pronunciation guide. Ready to start? Please say hello in [LANGUAGE] and ask me one simple question to get the conversation going. ``` ChatGPT will respond in your target language immediately and adapt to your level. Ask ChatGPT to quiz you on vocabulary related to a topic you care about. This beats memorizing random word lists because the words are useful to you right away. ``` Give me 8 vocabulary words related to [topic: travel / food / work / family]. For each word, show: [LANGUAGE] word — pronunciation guide — English meaning — one short example sentence. Then quiz me: say the English word and I will give you the [LANGUAGE] translation. ``` Go through the quiz out loud if you can, even if no one is listening. Speaking the word — not just reading it — is what makes it stick. After the vocabulary warmup, ask for a short reading passage at your level: ``` Write me a short paragraph (5–7 sentences) in [LANGUAGE] about [a topic from your vocabulary warmup]. Keep the vocabulary at [beginner / intermediate] level. After I read it, I will try to tell you in English what it said. ``` Read the paragraph once, then try to explain it back in English. ChatGPT will tell you what you missed or misunderstood. This builds reading comprehension quickly because you get immediate feedback. This is the most valuable part of the session. Tell ChatGPT to role-play a real-world situation: ``` Let's practice a conversation. You are [a waiter at a restaurant / a hotel receptionist / a neighbor I just met]. I am a tourist who speaks only a little [LANGUAGE]. Start the scene and respond naturally. Keep correcting my mistakes as we go. ``` Type your replies in the target language, even if they are short and imperfect. Imperfect sentences that you attempt are worth more than perfect sentences you never say. Good scenarios for beginners: ordering food, asking for directions, buying something at a market. Good scenarios for intermediate learners: describing a problem, making small talk, asking for recommendations. I practice Spanish and the single thing that helped me most was telling ChatGPT to stop translating things for me unless I explicitly asked. At first I kept asking "what does that mean?" every few sentences. ChatGPT started responding with "Try to guess from context — what do you think it means?" That small push made me actually read harder instead of reaching for a translation crutch every time. At the end of each session, ask for a summary of errors: ``` We are done for today. Can you list the grammar or vocabulary mistakes I made this session? For each one, show: what I wrote → what I should have written → one-line explanation of why. ``` This creates a mini review sheet you can screenshot and look at later. Over time you will notice patterns — the same mistakes coming up again and again — and that is exactly what to focus on next time. Start a simple note on your phone with new words from each session. After the mistake review, add: ``` What are the most useful new words we used today that I should remember? List them with [LANGUAGE] word, pronunciation, and English meaning. ``` Paste the list into your note. After two weeks you will have 80 to 100 words that actually came up in real practice — far more useful than a prepackaged vocabulary deck. Review the list for two minutes before each new session to keep the words fresh. ## A note on what ChatGPT cannot do ChatGPT will not teach you pronunciation through audio — it works in text only. For pronunciation practice, pair it with a tool that uses voice, like a language learning app or YouTube videos in your target language. ChatGPT is excellent for grammar, vocabulary, writing, and reading; those other tools cover what it cannot. ## What to try next Once you have a daily routine down, consider using your new language on a real trip. The [vacation planning guide](/guides/plan-vacation-with-chatgpt) shows how to plan a full trip with ChatGPT — including asking it to suggest destinations where your target language is spoken, so you have real motivation to practice. --- # LinkedIn Profile Makeover with AI: Step by Step Source: https://aiforregulars.com/guides/linkedin-profile-ai A lot of LinkedIn profiles read like a job description written by someone who wasn't sure what to say. Bullet points that start with "Responsible for…", an About section copied from an old resume, a headline that just says your job title. AI can fix all of that — not by making things up, but by helping you say clearly what you actually do. These steps take about an hour the first time. After that, updating individual sections takes minutes. Before you open any AI tool, collect the real information you want on your profile. This means: your actual job title, the team or product you work on, two or three things you accomplished (with numbers if you have them), the type of role you want next, and the industry you're in. The single biggest mistake people make is asking AI to "improve my LinkedIn profile" without giving it anything specific to work with. Vague input produces generic output. Give it facts and it gives you something worth using. Your headline is the line that appears under your name everywhere on LinkedIn. Most people leave it as their job title. That works, but a stronger headline tells recruiters both what you do and the value you bring. Copy and paste this prompt, filling in the brackets: ``` I'm a [job title] at [type of company]. I want my LinkedIn headline to attract [type of recruiter or hiring manager — e.g., "sales managers at tech companies"]. My strongest skill is [skill], and I'm known for [one accomplishment or quality]. Write 3 LinkedIn headline options. Each should be under 120 characters. Use plain language, no buzzwords. Format: [what I do] + [value I bring]. ``` Pick the option that feels most like you, or ask for more variations. The AI draft gives you a starting point — edit the final version to match your real voice. The About section is the most read part of your profile after your headline. It should answer three questions quickly: who you are, what you do well, and what you're looking for. Most people bury the most important information at the bottom — AI can help you flip that. Use this prompt: ``` I'm writing my LinkedIn About section. Here are the facts: - Current role: [title] at [company type] - Years of experience: [X] - Main skills or specialties: [list 2–3] - Biggest accomplishment: [describe briefly] - What kind of opportunity I'm open to: [be specific] - One thing that makes my approach different: [honest answer] Write a LinkedIn About section in first person. Three to four short paragraphs. Start with my strongest point, not my background. Conversational tone, no corporate buzzwords. Under 300 words. ``` Read the result carefully. AI often starts with something generic like "I am a passionate professional…" — delete that and start with the strongest sentence it wrote, wherever that appears. For each job in your experience section, the goal is to describe what you actually did and what happened as a result — not just list responsibilities. Use this prompt for each role: ``` I need to rewrite my experience bullets for this job on LinkedIn: - Title: [your title] - Company: [company type, e.g., "regional bank" or "startup"] - What I actually did day-to-day: [write a few sentences, rough is fine] - One or two things I'm proud of from this role: [be specific] Write 4–5 bullet points. Each should start with a strong action verb. Include a result or impact where possible. Plain language, no jargon. Keep each bullet under 2 lines. ``` If you have numbers — how many people you managed, how much revenue you supported, how fast a process improved — include them in your raw input. AI can't invent real metrics, but it can format the ones you give it. The Skills section affects LinkedIn search results. Recruiters often filter by skill, so having the right keywords here matters. Ask AI to help you identify gaps: ``` I work as a [job title] and want to be found by recruiters hiring for [target role]. Here are the skills I already have listed: [paste your current list]. What skills or keywords am I likely missing? Suggest 8–10 additions that are commonly used in job postings for [target role]. Only include real skills, no soft skills like "leadership." ``` Add the suggestions that genuinely apply to you. Endorsements still matter for credibility — ask current or former colleagues to endorse the skills that are most important for your target role. When I rewrote my own About section using AI, the first draft was technically correct but felt stiff. The fix was simple: I read it out loud and changed every phrase I wouldn't actually say in a conversation. That pass took ten minutes and made the whole thing sound like me instead of a press release. Before you publish anything, read your updated profile out loud. This sounds awkward but it works. Your brain catches problems when you hear them that it misses when you read silently. Listen for anything that doesn't sound like you, any detail that isn't quite accurate, or any sentence that's hard to get through. Fix whatever feels off. The goal isn't perfect writing — it's an accurate, readable profile that makes a recruiter want to contact you. ## What to try next Now that your profile is solid, the next step is making sure your actual resume matches it. Check the [guide to AI resume builders](/guides/best-ai-resume-builders) for tools that help you create a resume version of the same story, then read about [writing a cover letter with ChatGPT](/guides/chatgpt-cover-letter) when you're ready to apply. --- # How to Make an AI Video for Free (No Camera, No Skills Needed) Source: https://aiforregulars.com/guides/make-ai-video-free A few years ago, making a video meant a camera, a tripod, and hours of editing. Today you can type a sentence and get a short video clip in about a minute — completely free. Whether you want a fun birthday clip, a social media post, or just something cool to show your grandkids, AI video tools have made it surprisingly easy. Here's how to do it, step by step. Several tools let you generate AI video without paying anything upfront. Here are three solid options: - **Runway** — well-known and beginner-friendly; the free plan gives you 125 one-time credits when you sign up — they never expire, but they also never renew, so once spent you would need to upgrade to keep generating - **Pika** — great for short, stylized clips; free tier available on pika.art - **Canva** — if you already use Canva for invitations or posts, it has an AI video feature built in Go to any of these sites on your phone or computer and create a free account with your email address. No credit card required. A prompt is just a description of what you want the video to show. You don't need to be a writer — a plain sentence works fine. A weak prompt: *"a dog"* A better prompt: *"A golden retriever puppy playing in autumn leaves in a sunny backyard, slow motion"* The more detail you give — what's happening, where, what it looks like, what mood it has — the closer the AI gets to what you have in mind. Add words like *cinematic*, *close-up*, *bright and cheerful*, or *peaceful sunset* to guide the style. Most tools let you choose a visual style before you generate — things like "realistic," "cartoon," or "painterly." Some also let you upload a photo as a starting point, so the AI makes a video that begins with your image and animates it. If you're just starting out, skip the style options and go with the default. You can always experiment once you've made your first clip. I tested a few free tools by typing the same prompt into each one: *"an elderly woman reading a book by a fireplace, cozy, warm light."* Runway gave me the most realistic result, but Pika had a charming, slightly painterly look that I actually preferred. Try the same prompt in two tools side by side — it only takes a couple of minutes and the difference can surprise you. Hit the generate button and give it 30–90 seconds. Free tiers sometimes join a short queue during busy times, so you might wait a minute or two. That's normal. While you wait, don't close the browser tab. The generation happens on the tool's servers, but your result is usually only kept there briefly. When the clip is ready, press play and watch it. Ask yourself: - Does it match what I described? - Is there anything weird or off? (AI video tools sometimes get hands, faces, or text wrong) - Is the mood or style what I wanted? If you're not happy, click **Generate again** without changing anything — you'll often get a noticeably different result. Or rewrite your prompt to be more specific about what felt wrong. Once you like what you see, click the download button. Your video will save as an .mp4 file, which works on phones, computers, and all major social media platforms. Both Runway and Pika add a permanent watermark to every free-tier video — it cannot be removed without upgrading to a paid plan. Canva's free plan does not add a watermark to AI-generated clips, making it the better pick if a clean export matters to you. Your downloaded .mp4 can go straight to Instagram, TikTok, Facebook, or a text message. You can also open it in a free video editor like **CapCut** (phone) or **iMovie** (iPhone/Mac) if you want to add music, captions, or combine it with other clips. ## A note on expectations Free AI video tools generate short clips — usually 4 to 10 seconds. They can look a bit dreamy or imperfect, especially when people or hands appear. That's just where the technology is right now. For social media clips, product demos, or fun personal projects, the quality is more than good enough. ## What to try next Once you've got video covered, [creating AI images for free](/guides/create-ai-images-free) follows the same idea — type a description, get a picture. You can also [turn your existing family photos into a slideshow video](/guides/ai-video-from-photos) without any prompting at all. --- # How to Make ChatGPT Sound Like You (Not Like a Robot) Source: https://aiforregulars.com/guides/make-chatgpt-sound-like-you You ask ChatGPT to write a short note to your neighbor, and it comes back sounding like a press release. "I hope this message finds you well. I wanted to reach out regarding…" Nobody talks like that. If your AI drafts feel stiff, you are not doing it wrong. There are simple fixes, and none of them need technical skill. ## Why AI writing sounds like AI ChatGPT does not know you. Unless you tell it otherwise, it aims for the safest, most average way to say something. Journalists and researchers who have compared AI text with human text, including an analysis by The Economist, found some common habits: longer sentences that all have a similar length, fancier words than people usually use, the "it's not this, it's that" pattern, and lists of exactly three things. The details change as AI models are updated, so do not treat any single trick as proof. But the overall feel is the same: smooth, correct and impersonal. The good news is that the fixes are the opposite of those habits. Be specific, be plain, and let your own details show. This is the biggest single improvement. Find two or three short things you wrote yourself: an email, a text to a friend, a Facebook post. Paste them into ChatGPT and say: *"Here are examples of how I write. Notice my tone, how long my sentences are and the words I use. Then write a short note to my landlord about a broken heater in the same style. Do not copy my sentences."* Use writing you are happy with, and leave out anything private, such as full names, addresses or account numbers. If you are unsure what is safe to paste, see [what not to tell an AI chatbot](/guides/what-not-to-tell-ai-chatbot). AI follows instructions well when they are specific. Add a line like this to your request: *"Write the way a friendly person talks. Use short sentences and everyday words. No buzzwords, no lists of three, no flowery openings such as 'I hope this finds you well'. Do not end with a summary."* If the first draft is still too formal, just say so: "Make it warmer and shorter." You can go back and forth as many times as you like. For more on asking well, see [how to write prompts](/guides/how-to-write-prompts). If you do this often, you do not have to repeat yourself every time. ChatGPT has a setting for this. On the website or in the app, open **Settings**, then **Personalization**, then **Custom instructions**. Turn it on and you will see two boxes: one for what ChatGPT should know about you, and one for how you want it to answer. In the second box you could write: *"Write in a warm, plain, conversational tone. Short sentences. Avoid corporate words and dramatic phrasing. Do not start by repeating my question."* Custom instructions are available on all plans and apply to your new chats. In the phone app the entry is usually called **Customize ChatGPT**, in your profile menu. Menu names can change with updates, so if you do not see them, look around the Settings menu. If you would like ChatGPT to remember things about you as well, our guide to [ChatGPT memory settings](/guides/chatgpt-memory-settings) explains how to control that. This is what separates your writing from anyone's. AI cannot know that your neighbor's dog barks at 6 a.m. or that you visited the hardware store on Saturday. So give it those facts in your request, or add them afterward. One real detail per paragraph does more than any style trick. Vague, general statements are the clearest giveaway, so replace them with something real: a name, a place, a number, a memory. Treat the draft as a first draft, not the final one. Read it out loud. Wherever you stumble, or think "I would never say that", change it. Cut the opening line, shorten the long sentences, and swap the fancy words for ones you would use with a friend. It is fine to leave in small imperfections, such as a short sentence or a "well" or "anyway". They are part of how people write. My own test is simple: would my sister recognize this as something I wrote? If the email could have been sent by anyone to anyone, I go back and add one thing that only I would say. Usually that takes two minutes and fixes most of the problem. ## A quick before and after **Before (typical AI):** "I hope this message finds you well. I wanted to reach out regarding the heating issue in my apartment, which has been a significant concern for several days." **After (with a sample of your writing and a few details):** "Hi Mr. Novak, the heater in my apartment stopped working on Monday and it is getting cold at night. Could someone come by this week? I am home after 5." Same message, but the second one sounds like a person wrote it. ## When you should not hide the AI Making a draft sound like you is about clearer, warmer communication, not about fooling anyone. If a school, employer or publication asks you not to use AI, follow the rule. And always read what you send: AI can get facts wrong, so check any names, dates and numbers. You cannot reliably tell whether a text was written by AI just by reading it, and neither can other people, as we explain in [how to tell if a text was written by AI](/guides/how-to-tell-if-text-is-ai). Honesty with the people you write to matters more than any trick. For everyday emails, see also [how to write a professional email with AI](/guides/write-professional-email-with-ai). ## Sources - [How to spot AI writing](https://dataconomy.com/2026/08/04/how-to-spot-ai-writing/) — Dataconomy - [Forget em dashes: Viral report on AI writing has surprising new clues](https://www.fastcompany.com/91584243/how-to-identify-ai-generated-writing-viral-report-has-surprising-new-clues-economist) — Fast Company - [ChatGPT Custom Instructions](https://help.openai.com/en/articles/8096356-chatgpt-custom-instructions) — OpenAI Help Center --- # Meta AI on Facebook and Instagram: What It Can Do for You Source: https://aiforregulars.com/guides/meta-ai-facebook-instagram You have probably noticed the blue circle or the sparkle icon popping up inside Facebook and Instagram. That is Meta AI — and it is more capable than most people realize. It is not just a chatbot sitting in a corner; it is woven into search, into comments, and into photo editing. Here is how to actually use it, what it is good at, and a few things to keep in mind about your privacy. Open the Facebook app and tap the search bar at the top. You will see a prompt to ask Meta AI a question. This is different from searching for posts or people — it is a direct question-and-answer experience. Try typing a question like "What are some easy dinner ideas for a family of four?" or "Who is the current leader of Canada?" You get a conversational answer, not a list of posts to scroll through. On Facebook, tap the Meta AI icon (the blue circle) to open a direct chat. On Instagram, tap the icon in the direct messages area or start a new message to Meta AI. Once you are in the chat, you can ask anything you would ask a general-purpose assistant — help writing something, explaining a topic, coming up with ideas. The chat history stays there so you can return to a previous conversation. When you are creating a new post on Instagram, look for the AI suggestion option near the caption field. Tap it, describe what you want — for example, "fun caption for a beach photo with friends" — and Meta AI will suggest a few options. Pick one you like, edit it to match your voice, and post. It is a handy shortcut when you know you want to say something but cannot find the right words. Instagram has several AI-powered editing features built into the photo editor. Look for AI editing options in the photo editor or Story editor. Instagram's confirmed generative background tool is called **Backdrop**; separate controls let you remove or erase elements from a photo, and Stories' AI restyle panel offers lighting and style adjustments. Exact button names and which tools appear depend on your app version and region — if you do not see them yet, they may still be rolling out to your account. Select the tool, follow the on-screen instructions to mark what you want to change, and the app will generate the result. Some tools work better on simple photos than on busy or crowded scenes. In Facebook posts or groups, you can tag Meta AI in a comment the same way you would tag a friend: type @MetaAI and then your question. Everyone in the thread can see the AI's reply. This works well when someone in a group asks a factual question and you want to share an answer quickly, or when you are looking for ideas in a group context. Just remember that the response is public in that thread — do not share personal details this way. I used the caption suggestion feature for a family reunion photo and it saved me at least ten minutes of staring at the screen. I typed "warm caption for a big family gathering, not too sentimental" and it gave me three options that actually sounded like something I would write. I tweaked the one I liked best and posted it. Genuinely useful. On **Facebook**, tap the menu (☰), go to **Settings & Privacy** > **Privacy Center**, then scroll to the **AI at Meta** section. On **Instagram**, tap your profile picture, then the menu (☰) > **Settings and activity** > **Privacy Center**, and look for **AI at Meta**. Here you can see what data Meta uses to personalize your AI experience and make adjustments. It is also worth reviewing the broader Privacy Checkup tool in Facebook if you have not done it recently. AI features are part of a larger picture of how Meta uses your activity. ## What Meta AI does not do Meta AI does not automatically read or analyze your personal posts or photo albums. It responds to what you actively share with it in a chat or a prompt. It cannot access your friends' private profiles, send messages to other people, or make purchases. Like all AI assistants, it can get things wrong — especially on recent news or niche topics. If it tells you something that matters, double-check it. ## Spotting AI-generated content in your feed Now that AI tools are widespread, you will start seeing AI-generated images and text in your feeds. If something looks a little too perfect or feels slightly off, read [how to spot AI-generated photos](/guides/spot-ai-generated-photos) — a useful skill to have. For a deeper look at how bad actors use AI to create fake profiles, see our guide on [AI fake Facebook profiles](/guides/ai-fake-facebook-profiles). --- # Meta AI in WhatsApp: What It Does and How to Try It Source: https://aiforregulars.com/guides/meta-ai-whatsapp WhatsApp quietly added an AI assistant, and many people are noticing the blue circle in their search bar without knowing what it is. Meta AI is now built into WhatsApp in many countries, and it can actually be useful once you understand what it does and what its limits are. This guide walks you through the basics: how to start a chat, what kinds of tasks it handles well, how to generate images, and how to limit or mute the feature if you would rather not use it. Open WhatsApp and look for the blue circle icon in the search bar at the top of your chats screen. Tap it to open a direct conversation with Meta AI. You may also see Meta AI listed as a contact in your chat list. Either entry opens the same assistant. If you do not see it, your region may not have the feature yet. Type your question in the message box and send it just like a regular message. Meta AI replies in seconds. Good starting questions include things like "What is a good recipe for chicken soup?" or "Can you help me write a birthday message for my mom?" The assistant handles everyday questions well and gives you a clean, readable answer. You can bring Meta AI into a group conversation without switching apps. In any group chat, type @Meta AI followed by your question. Only the people in that group see the exchange. This is handy when your family group is planning something — say, a weekend trip — and someone wants quick suggestions for restaurants or activities without everyone leaving the chat to search separately. Meta AI can create pictures from a text description. Type **Imagine** followed by a description of what you want — for example: "Imagine a birthday card with colorful balloons and a golden retriever." The assistant generates an image and sends it back in the chat. You can save it to your phone or share it in another conversation. Results vary — simple, clear descriptions tend to work better than complex ones. Meta applies standard content policies — no explicit or harmful imagery is generated. Free use has an unpublished daily cap (independent reviews report roughly 25 images per day); a paid Meta subscription raises the limit. Tap the search bar at the top of WhatsApp. At the top of the search results you will see a prompt that says something like "Ask Meta AI." Type a question there and get an answer without starting a full chat. This is the fastest way to use it for quick lookups — a conversion, a spelling check, or a short piece of information — when you are already in the middle of using WhatsApp for something else. I found Meta AI most useful for drafting quick messages when I am not sure how to phrase something. I will type something like "Help me politely tell my neighbor I cannot attend their party" and it gives me two or three options I can edit and send. It saves that awkward five-minute stare at the screen. If you do not want Meta AI suggestions appearing while you type, you can take a few steps to reduce how much it shows up. WhatsApp does not have a single setting to turn off Meta AI entirely. On Android and iPhone, open **Settings** and look for AI-related options — depending on your app version and region these may appear under **Chats** or **Privacy**. For current step-by-step instructions on available controls (such as managing Meta AI web search or restricting it in group chats), visit WhatsApp's Help Center at faq.whatsapp.com and search for Meta AI. You can also simply archive or delete the Meta AI chat. This keeps it out of your main chat list without affecting the rest of WhatsApp. Meta AI is powered by Meta's servers, which means anything you type to it leaves your phone. Do not share passwords, financial account details, your home address, or your Social Security number in a Meta AI chat. Stick to the same rule you would apply to any online service: if you would not write it on a postcard, do not type it to an AI assistant. ## What Meta AI cannot do Meta AI does not have access to your existing WhatsApp messages, your contacts list, or your photos unless you paste them directly into the chat. It cannot send messages on your behalf, make calls, or access your WhatsApp groups without being invited with the @Meta AI mention. It also does not always get facts right. For anything important — medical, legal, financial — verify what it tells you with a trusted source. ## What to try next If you want to compare this experience with a standalone AI app, take a look at [how to set up the ChatGPT app on your phone](/guides/chatgpt-app-setup). And before you start sharing personal details with any AI tool, it is worth reading [what not to tell an AI chatbot](/guides/what-not-to-tell-ai-chatbot). --- # MyHeritage AI: Animate and Colorize Your Old Family Photos Source: https://aiforregulars.com/guides/myheritage-ai-animate-photos A black-and-white photo of your grandparents. A formal portrait from the 1940s where nobody is smiling. These photos carry real history, but they can feel distant. MyHeritage has two AI features that change that: one adds color to black-and-white photos, and the other makes still portraits briefly come to life. This article covers both — how they work, what to expect, and how to use them step by step. For repairing actual physical damage (tears, water stains, fading), that's a different process covered in the [photo restoration guide](/guides/restore-old-photos-ai). ## What MyHeritage Actually Does **In Color** analyzes the tones and textures in a black-and-white photo and adds realistic color. It makes educated guesses about skin tones, clothing colors, and backgrounds based on what it's seen in millions of other historical photos. **LiveMemory** goes further: it applies short animated movement to a still portrait — a slight head turn, a blink, a subtle smile. The AI generates video effects that are mapped onto the face in your photo. Neither tool changes the underlying photo. You download the animated video or colorized image separately. Go to **myheritage.com** and click **Sign up**. You'll need an email address. The free account gives you access to both the In Color (colorization) and LiveMemory (animation) features, with a limited number of free uses before a paid subscription is required. You don't need to build a family tree or enter any genealogy information to use the photo tools. When signing up, you can skip the family tree setup and go directly to the AI tools. These tools work best with: - A clearly visible face (at least medium close-up — not a tiny figure in a wide landscape photo) - Reasonable image quality — sharp enough to make out facial features - For colorization: a true black-and-white or sepia photo (not a color photo you converted) - For animation: a single person or a clear main subject Older photos are often only available as prints. If you're working from a physical photo, scan it at a high resolution — 600 DPI or higher is ideal. A photo scanning app on your phone (like Google PhotoScan or Photomyne — note that Microsoft Lens was retired in early 2026 and no longer works) can work if you don't have a scanner. From your MyHeritage account, click **Enhance photos** in the top menu, then choose **In Color**. Upload your black-and-white photo. The AI processes it in about 10–30 seconds. You'll see a before/after slider so you can compare. The colorization is automatic — you don't choose colors manually. The AI makes its best guess based on context. Skin tones and nature tend to come out well. Clothing colors are a guess — your great-grandmother's dress might have been blue or green, and the AI picks one. Think of it as plausible, not certain. Download the colorized version if you're happy with it. From the Enhance photos menu, choose **LiveMemory**. Upload your photo. You'll be shown a short preview of the animation before it's finalized. The animation is a few seconds long and loops. A few things to know: - You can choose from a selection of animation effects — options include different looks, gestures, and movements, and MyHeritage has been adding new effects over time - The animation works best on a face that's looking roughly toward the camera - Severely damaged or very low-resolution photos may produce uneven results When you're happy, click **Download** to save the animated video (usually in .mp4 format). I showed a LiveMemory animation of my grandmother's 1950s portrait to my aunt, who cried. Whatever you think of the technology, the emotional effect on families can be real. That said, some people find animated historical photos unsettling — it's worth thinking about how other family members might react before you share widely. Colorized photos save as standard .jpg or .png files — you can print them, share them, or use them in a digital photo book. Animated videos save as short .mp4 clips. They share easily on Facebook, WhatsApp, or in a group text. Most social media platforms accept them as video uploads. MyHeritage also lets you save results to your account so you can access them again later, which is useful if you're working through a whole collection of family photos. ## What to Realistically Expect The colorization and animation are impressive for what they are, but they're educated guesses. The color of a garment, the exact shade of someone's eyes — these details aren't in the original photo, so the AI fills them in plausibly rather than accurately. For most families, that's fine. For archivists or historians, it's worth noting that a colorized photo is an interpretation, not a record. ## What to try next If your family photos need repair before colorizing — faded, torn, or water-damaged prints — the [AI photo restoration guide](/guides/restore-old-photos-ai) covers the best tools for that step. And if you want to turn still photos into full short videos, not just brief animations, see [how to make AI videos from photos](/guides/ai-video-from-photos). --- # How to Opt Out of AI Training and Delete Your Chat History Source: https://aiforregulars.com/guides/opt-out-ai-training-delete-data AI companies collect your conversations by default. Most people never change those defaults — which means their chats may be used to train future AI models without them realizing it. The good news is that each major service has a way to turn this off, and the process is not complicated. This guide walks you through it service by service. There are three different things you can control independently, and it helps to know the difference before you start — they're easy to lump together. The first is **chat history**: whether your conversations are saved to your account at all. Turning this off means chats disappear after each session — nothing is stored. The second is **cross-conversation memory** (ChatGPT calls this Memory): the chatbot's ability to remember facts you told it in one conversation and use them in other, otherwise unrelated chats. This is separate from both history and training — you can have history off and memory on, or the other way round. The third is **AI training**: whether your saved conversations — or the facts memory has stored — are used to improve the model for everyone else. Even if you keep both history and memory on, you can opt out of having those chats fed into future training. Turning off one of these does not turn off the others, and deleting a conversation from your own account doesn't necessarily mean it's gone from a model that already trained on it (more on that in the full-deletion step below). You can turn off one of these, all three, or none. This guide covers all of it. Open ChatGPT and click your profile icon or initials in the bottom-left corner, then choose **Settings**. Navigate to **Data controls**. You will see a toggle labeled something like "Improve the model for everyone" — turn that off. When this toggle is off, OpenAI will not use your future conversations to train its AI models. Conversations from before you made this change may already have been used. This setting applies to your account on all devices. You do not need to repeat it on your phone separately. In the same **Settings → Data controls** screen, look for the option to delete your chat history. You can delete all conversations at once or go to the chat list and remove individual ones. Deleting your history removes those conversations from your account and from OpenAI's training pipeline for future runs. It does not remove patterns already incorporated into models trained before you deleted. If you want to keep a copy of your chats before deleting, use the **Export data** option on the same screen — OpenAI will email you a downloadable file. Chat history and memory sound similar but are two different things. Go to **Settings → Personalization → Memory**. You'll find two separate toggles: **Reference saved memories** (specific facts you've explicitly told ChatGPT to remember — that you're vegetarian, or working on a thesis) and **Reference chat history** (ChatGPT picking up patterns from earlier conversations on its own, without you asking it to remember anything). Turning both off stops ChatGPT from using anything from past conversations in new chats. On the same screen you can review and delete individual saved memories, or clear all of them at once — that's a separate action from deleting chat history in Step 3. Gemini and Copilot each have a similar personal-personalization feature, kept separate from history and from training — but the exact name and location of the toggle changes between updates, so the safest bet is to check the personalization or profile settings of whichever service you use. I went through this process on my own ChatGPT account and was surprised how quick it was — the whole thing took about four minutes. What made me feel better was not just the opt-out toggle, but actually exporting my data first so I could see exactly what OpenAI had stored. It was mostly my conversations, nothing unexpected. I now do a history clear-out every few months, the same way I clean out my email inbox. Go to **myaccount.google.com** and sign in if needed. Click **Data & Privacy** in the left menu. Scroll down to find **Gemini Apps Activity** and click on it. From there you can turn off Gemini Apps Activity entirely, which stops Gemini from saving your conversations and stops Google from using them to improve its AI products. You can also set an auto-delete schedule — for example, automatically deleting Gemini activity after 3 months. If you use Google Workspace (a company or school account), your administrator controls some of these settings and they may look different. While still in the **Gemini Apps Activity** section, click **Manage activity**. This takes you to a view of your saved Gemini conversations. You can delete individual chats or select a time range and delete everything within it. For a full clean slate, choose **Delete all time** from the date filter. Google will remove those conversations from your account and from its training pipeline for future model updates. Go to **account.microsoft.com** and sign in. Navigate to **Privacy** and then look for **Privacy dashboard**. From there, find the section for Copilot or AI-related search and conversation history. Microsoft's settings are organized slightly differently depending on whether you use Copilot through Windows, through Edge, or through Microsoft 365. The Privacy Dashboard is the central place to see and delete activity across all of them. You can delete individual sessions or clear everything at once. The steps above handle what is stored in your active account. If you want to request deletion of all personal data a company holds — including any records kept after account deletion — each company has a formal process. **OpenAI:** Visit **privacy.openai.com** and look for the option to submit a data subject request. You can request a copy of your data, correction, or full deletion. **Google:** Go to **myaccount.google.com → Data & Privacy → Download or delete your data** to export everything or delete your Google account entirely. For partial deletion without closing your account, use the activity controls described in Steps 5 and 6. **Microsoft:** Use the **Microsoft Privacy Dashboard** at **account.microsoft.com/privacy** to browse and delete data across Microsoft services, or submit a formal request through their privacy contact form. If you're in the UK or EU, these requests fall under GDPR: the company has one month to respond, extendable by two further months for complex requests. If you're in California, CCPA gives a business 45 days, extendable once by another 45. Other countries and states have their own deadlines. Either way, expect the company to verify your identity first. ## What to Try Next Now that you have taken care of your own data, it is worth spending a few minutes on the finer details. Our guide to [what AI actually knows about you](/guides/what-does-ai-know-about-me) explains the difference between stored facts and educated guesses — and what the AI genuinely cannot see. If you want a deeper look at all the privacy toggles inside ChatGPT specifically, [the ChatGPT privacy settings guide](/guides/chatgpt-privacy-settings) covers every option in detail. ## Sources - [Memory FAQ](https://help.openai.com/en/articles/8590148-memory-faq) — OpenAI Help Center - [How does "Reference saved memories" work?](https://help.openai.com/en/articles/11146739-how-does-reference-saved-memories-work) — OpenAI Help Center - [Gemini Apps Privacy Hub](https://support.google.com/gemini/answer/13594961?hl=en) — Google - [Connected experiences in Microsoft 365](https://support.microsoft.com/en-us/privacy/connected-experiences-in-microsoft-365) — Microsoft - [Art. 12 GDPR – Transparent information, communication and modalities](https://gdpr-info.eu/art-12-gdpr/) — GDPR-info.eu (one-month response deadline, extendable by two months) - [California Consumer Privacy Act (CCPA) regulations — timing requirements](https://oag.ca.gov/privacy/ccpa) — California Attorney General's Office --- # How to Plan Your Garden with AI: A Season-by-Season Guide Source: https://aiforregulars.com/guides/plan-garden-with-ai Planning a garden sounds simple until you realize the seeds you bought need to go in the ground six weeks before your last frost date — a date you didn't know and forgot to look up. AI can handle all of that planning for you. Here's how to use ChatGPT to build a full garden plan and set up reminders so you don't miss a single watering. Open ChatGPT and start with the basics. The more specific you are, the better the plan you get back. ``` I want to plan a vegetable garden. Here's my situation: - Location: [your city and state or country] - Space: [e.g., "one 4x8 raised bed" or "three large containers on a south-facing balcony"] - Experience level: [beginner / have done this before / intermediate] - Goal: [e.g., "grow food for salads" or "tomatoes and herbs" or "whatever's easiest to grow"] - How much time I can spend: [e.g., "30 minutes a week" or "I want low maintenance"] ``` ChatGPT will use your location to estimate your climate zone and last frost date, then suggest what to grow. Once ChatGPT knows your setup, ask for a month-by-month planting schedule: ``` Based on my location and space, give me a planting calendar for the year. Tell me what to do each month: what to start indoors, what to direct sow outdoors, and when to harvest. Format it as a table. ``` You'll get a clear timeline. Print it or copy it somewhere you'll see it regularly. Some plants help each other; others compete for the same resources. Ask ChatGPT: ``` Which of these plants grow well together in my space? [list the vegetables or herbs you want to grow] Are there any combinations I should avoid? ``` This is especially useful for small spaces where plants are close together. Classic good pairings include tomatoes with basil, carrots with onions, and beans with corn. A planting calendar is only useful if you remember to follow it. Use your phone's reminder app — Google Keep, Apple Reminders, or any calendar app — to create recurring reminders for key tasks. Key reminders to set: - **Start seeds indoors** — usually 6–8 weeks before your last frost date - **Water seedlings** — every 2–3 days, or daily in hot weather - **Fertilize** — typically every 2–4 weeks once plants are established - **Check for pests** — once a week during growing season - **Harvest** — varies by plant; ChatGPT can tell you the signs for each one Ask ChatGPT to generate specific dates for you: ``` Based on my planting calendar, give me a list of specific dates to set phone reminders for, from now until the end of the growing season. Start from today, [today's date]. ``` When something goes wrong — yellow leaves, holes in plants, wilting despite watering — describe what you see in detail: ``` My tomato plant has yellow leaves starting from the bottom up, and some brown spots. The rest of the plant looks healthy. What could cause this and how do I fix it? ``` For pest or disease identification, apps like **Picture This** or **iNaturalist** let you take a photo and get an AI identification. This is often faster than trying to describe a bug in words. The best time to plan next year's garden is right after harvest, when you still remember what worked and what didn't. Use ChatGPT at the end of the season: ``` My growing season is ending. Here's what I grew: [list]. Here's what worked well and what didn't: [your notes]. What should I do this fall to prepare the bed for spring? What would you suggest I try differently next year? ``` Crop rotation — moving plant families around each year — is important for soil health, and AI handles this well if you tell it what you grew where. I used ChatGPT to plan my first raised bed last spring. I gave it my location and a list of things I wanted to grow, and it told me I'd crammed too many large plants into too small a space. It suggested I cut down to two tomato plants and fill the rest with herbs and lettuce. That turned out to be exactly the right call. ## Season Quick Reference | Season | Main task | What to ask AI | |---|---|---| | Late winter | Plan and order seeds | "What should I grow and when do I start?" | | Spring | Plant outdoors, transplant seedlings | "Is it safe to plant out yet in [location]?" | | Summer | Water, fertilize, harvest, pest watch | "Why are my [plant] leaves turning [color]?" | | Fall | Harvest, soil prep, plan next year | "What should I do to prepare my beds for winter?" | ## What to Try Next If you're growing vegetables and want to turn the harvest into meals, [How to Turn Fridge Leftovers into Recipes with AI](/guides/ai-recipe-from-leftovers) shows you how to prompt AI for recipes with exactly what you have. For more practical AI ideas for everyday life, [Things to Ask ChatGPT: 30 Useful Ideas](/guides/things-to-ask-chatgpt) is a good place to browse. --- # How to Plan a Whole Vacation with ChatGPT (Real Itinerary) Source: https://aiforregulars.com/guides/plan-vacation-with-chatgpt My family spent a long weekend in Lisbon last fall, and I did almost all the planning in one afternoon — with ChatGPT. I'm not a travel agent. I'm not particularly organized. But I ended up with a four-day plan that felt handcrafted: neighborhood walks, museum picks, restaurant options for every dinner, and a rainy-day backup. Here's exactly how I did it, and how you can do the same for your next trip. ## Start with the master prompt The most important move is giving ChatGPT enough information up front. Vague questions get vague answers. Copy this prompt and fill in your own details before you send anything: ``` I want to plan a trip to [DESTINATION] from [START DATE] to [END DATE]. Traveling with: [number of adults, any kids and their ages] Budget per day (not counting flights): around $[AMOUNT] Travel style: [relaxed sightseeing / active / a mix of both] We love: [food, history, beaches, hiking, museums, nightlife, shopping — pick yours] We dislike: [crowds, long walks, tourist traps — whatever applies to you] Dietary needs: [any restrictions] Please give me a day-by-day itinerary with morning, afternoon, and evening plans. Include estimated travel time between spots and two or three restaurant suggestions per day. ``` That single prompt does most of the heavy lifting. Hit send and wait about ten seconds. ## What comes back ChatGPT returns a full itinerary. For Lisbon it gave us: - Three different neighborhoods spread across the days so we were not backtracking - Museum suggestions with estimated visit times and a note about which day they close - A mix of sit-down restaurants and quick lunch spots matched to our budget - A walking route through the Alfama district with specific viewpoints flagged It reads like a plan a well-traveled friend would sketch out on a napkin. Not perfect — but a strong first draft you can shape. I paste the itinerary into a Google Doc the moment I have it. That way I can leave comments ("check if kids are allowed here"), share it with whoever I'm traveling with, and not worry about losing the chat window. ChatGPT does not save your conversations indefinitely, so get it out of the chat early. ## Refine with follow-up questions The real power is the back-and-forth. After the first plan came through, I asked follow-ups like: - "Can you swap Day 2 afternoon for something indoors? The forecast looks rainy." - "We have a 4-year-old. Which of these activities work well for small kids?" - "Give me a cheaper dinner option for Day 3 — something under $20 per person." Each time, ChatGPT adjusted the plan without me having to re-explain the whole trip. That flexibility alone saves hours compared to googling each change separately. ## What you still need to check yourself Here is the honest part. ChatGPT now searches the web automatically on every plan, including the free tier — but search results can still lag behind recent closures, sold-out tickets, or day-to-day price changes. That means some details will be out of date. **Always verify before you go:** - Opening hours and days closed, especially around national holidays - Current ticket prices — they change more than you would expect - Restaurant reservations — popular spots in tourist cities fill up weeks ahead - Any temporary closures or ongoing renovations Think of ChatGPT as the brainstorming phase and a quick Google search as the fact-checking phase. Use both, and you get the best of each. ## Two bonus prompts worth trying Once the itinerary is set, a couple more questions can round out your prep. **Packing list:** ``` Based on a trip to Lisbon in October — lots of walking, casual restaurant dinners, two museum visits — what should I pack for two adults and a 4-year-old? ``` **Budget breakdown:** ``` Estimate daily spending for this Lisbon trip: 2 adults, casual dining, public transit, paid museum entries, no big splurges. Give me low, mid, and high estimates per day. ``` That budget prompt is how I realized Lisbon's metro is surprisingly affordable and I had overbudgeted on transport. ## A word on flight and hotel searches ChatGPT cannot search live prices for flights or hotels. For that, open Google Flights or Kayak in a separate tab. What ChatGPT can do is help you think through logistics — "Is it better to fly into Lisbon or Porto if we want to see both cities?" — before you start clicking through booking sites. ## What to try next Now that your trip is planned, the [ChatGPT monthly budget guide](/guides/chatgpt-monthly-budget) can help you set aside money for it without guessing. And if this is your first time using ChatGPT for anything, the [beginner's guide to using ChatGPT](/guides/how-to-use-chatgpt) explains the basics in plain, jargon-free language. --- # How to Restore Old Family Photos with AI (Free Tools) Source: https://aiforregulars.com/guides/restore-old-photos-ai Somewhere in a shoebox or an old album there are photos of people you love — faded, scratched, or just worn out with age. AI restoration tools can bring a lot of that detail back in minutes, and several of them are free. You do not need Photoshop, design skills, or even much patience. The workflow is: scan, upload, download. Here is how to do it well. The better the scan, the better the result. A photo taken on your phone in good light works, but a flatbed scanner gives cleaner results. If you are scanning: - Use at least 600 DPI for a standard 4×6 print - Use 1200 DPI for small or heavily damaged photos - Clean the scanner glass first — dust specks on the glass show up as spots on the scan If you are photographing with a phone: - Lay the photo flat on a solid light-colored surface - Use good natural light and avoid flash — flash creates glare that erases detail - Hold the phone directly above the photo, parallel to it, not at an angle Save the file as a JPG or PNG before moving on. Several tools do photo restoration without charging you for basic use. They fall into three categories: **Web-based tools** — you upload through a browser, no software to install. Good for occasional use. Reliable free options include **MyHeritage In Color** (myheritage.com/incolor) for colorization and damage restoration, **Remini** (remini.ai) for sharpening and scratch repair, and **Canva** (canva.com/features/photo-restoration/) for a broader editing suite with restoration included. MyHeritage requires a free account sign-up; Remini allows up to 5 free restorations per day. **Apps on your phone** — some photo apps include an AI restore feature. Check the app store on your phone and search "photo restoration." Top-rated free options include **Remini** (iOS and Android — 5 free uses per day), **YouCam Enhance** (iOS and Android — includes AI colorize and lighting fix), and **Adobe Photoshop Express** (iOS and Android — no hard daily limits on basic restoration). Search any of these by name to find them directly. **AI image tools with restoration features** — some general AI image editors include restoration as one option among many. **Canva** is the most widely used example — its photo restoration tool is part of the free plan under the AI editing features section. **Picsart** also includes one-click AI restoration on its free plan, available on web and mobile. Start with a web-based tool since nothing needs to be installed. Most have a simple "upload" button on the home page. On the tool's website: 1. Click the upload button and select your scanned photo 2. Wait for the AI to process — this usually takes 20 to 60 seconds 3. Look at the before and after comparison the tool shows you The AI automatically removes scratches, reduces grain, sharpens edges, and brings out faded detail. You typically do not need to adjust any settings for a first pass. If the result looks over-processed or has a plastic, smooth look that feels unnatural, try a different tool — each one has a slightly different style. Zoom in on: - **Faces** — AI sometimes smooths faces too aggressively or misreads features, especially in low-resolution originals - **Text in the photo** — AI can distort lettering on signs, clothing, or visible writing - **Edges and fine detail** — lace, hair, and fabric texture sometimes blur when they should stay crisp If something looks off, check whether the tool lets you adjust the strength of the restoration. A lower setting often preserves more natural texture even if it leaves some imperfections. Download the restored version and keep your original scan untouched in a separate folder. I always run the same photo through two different tools and compare. One tool might handle the face better while another does a cleaner job on the background. I save both results and sometimes pick different parts from each one. It takes five extra minutes and the difference can be noticeable. Most restoration tools include a "colorize" button that adds color to black-and-white photos. The result is always an educated guess — the AI picks colors that are statistically likely for the era and content of the photo, not colors it actually knows were there. That said, colorization can make old photos feel newly alive, especially portraits. Try it and see if you like the result. If you know the actual color of something in the photo — "my grandmother's dress was blue" — some tools let you mark that area and lock in your chosen color before the AI fills in the rest. Two tools that support this: **Palette.fm** (palette.fm — free in the browser, color-hint picker for specific areas) and **AKVIS Coloriage AI** (desktop software; 10-day free trial — uses brush strokes to mark target colors by zone). Once you are happy with the result: - Save the high-resolution version to a labeled folder: something like "Restored Photos / Grandma 1955" - Back it up — copy it to a cloud service (Google Photos, iCloud, or Dropbox) and an external drive - If you want to print it, most online print shops accept JPG files; order a test print at a small size before ordering a large one Sharing the restored photo with family is often the best part. Consider putting together a simple digital album to send around — Google Photos makes it easy to create a shared album by link. ## What to try next If you enjoyed this project, the [beginner's guide to ChatGPT](/guides/what-is-chatgpt) opens up a lot more ways AI can help with everyday tasks. And if you are thinking about what to do with restored photos as gifts, the [vacation planning guide](/guides/plan-vacation-with-chatgpt) has a section on using ChatGPT for trip memory books — a natural companion project. --- # I Sent Money to a Scammer: What to Do in the First 24 Hours Source: https://aiforregulars.com/guides/sent-money-to-a-scammer If you are reading this because it just happened: stop reading in a moment and call your bank. The single most useful thing you can do today is make that call as early as possible. Everything else on this page will still be here in ten minutes. And before anything else — this was not your fault. Modern scams are run by organised, well-funded operations that rehearse their scripts, clone real voices with AI, copy real bank websites down to the fonts, and deliberately create the kind of time pressure that switches off careful thinking. They fool lawyers, accountants and IT professionals every week. Shame is the tool that keeps victims quiet and slow, and slow is exactly what the scammer needs from you now. ## The first 30 minutes Use the number printed on the back of your card or in your banking app, never a number the scammer gave you. Most banks have a 24-hour fraud line. Say this clearly at the start: **"I am reporting a fraudulent payment. I need it stopped and my account secured."** That phrasing routes you to the fraud team instead of general customer service, and it starts a formal case rather than a query. Have ready: the amount, the date and time, the account or card used, and where the money went if you know. Ask for a **case or reference number** before you hang up, and write it down. You will need it every time you call back. Fraud calls go better when you ask rather than wait: 1. **"Can this payment be recalled or stopped?"** — for a transfer, ask them to contact the receiving bank immediately. For a card payment, ask whether it is still pending, because a pending payment is far easier to kill than a settled one. 2. **"Should my card or account be frozen?"** — if the scammer has your card details or logged into your account, yes. 3. **"What do I need to send you, and by when?"** — many banks require a written fraud claim within a set number of days. Missing that deadline can end the claim on a technicality. Stop replying. Do not answer their calls, do not explain, do not tell them you have called the bank. If they know you are onto them, they move the money faster. But **keep every message, email, transaction receipt and phone number**. Screenshot the chat, the payment confirmation, the profile, the website address. Do not delete the app or block the number until you have copies — that evidence is what your bank and the police will ask for, and it is the part victims most often destroy in the first wave of anger. Open a note or a sheet of paper and write a plain timeline: when they first contacted you, what they claimed, what you sent, when, and how. Include the exact wording they used if you remember it. You will repeat this story to the bank, the police and possibly an ombudsman, and the details fade fast. A written timeline also helps you see the whole shape of it — which is often the moment people realise there was more than one payment. This is the rule that saves the most money after the first loss. If anyone asks you to pay a fee to release, unlock, verify, tax or recover your funds, that is the scam continuing. Real institutions deduct fees from money they hold; they never ask you to send fresh money from outside to free money you supposedly already have. The same goes for anyone who contacts you afterwards offering to get your money back — there is a section on that second wave further down this page. A partner, an adult child, a friend, a neighbour. Not because you need permission — because isolation is part of how the scam worked, and because a second head catches things you will miss while your hands are still shaking. Ask them to sit with you for the phone calls if you can. ## What is actually recoverable, by payment method This is the honest version. Some of these are worth fighting for; some are not coming back, and knowing which is which stops you wasting the hours that matter. | How you paid | Realistic chance | What to do first | | --- | --- | --- | | Credit card | Good | Dispute with the card issuer — chargeback rights are strongest here | | Debit card | Fair to good | Report as fraud immediately; time limits are shorter than credit | | Bank transfer | Depends on speed | Ask your bank to recall it and to contact the receiving bank | | Instant payment app | Low, but act anyway | Report inside the app *and* to your bank; ask them to freeze the recipient | | Gift cards | Very low | Call the card issuer with the numbers; report to your fraud body | | Cryptocurrency | Effectively nil | Report anyway; note the wallet address and transaction ID | | Cash by courier or post | Effectively nil | Call the police — this often means a local person collected it | **Card payments.** A card gives you the most protection of any method. If you paid by credit card, the issuer can usually reverse the payment through the card scheme's chargeback process, and in most countries you are not liable for fraudulent card use beyond a small capped amount. In the EU, if someone used your card without your permission, you can be asked to bear a maximum of 50 euros — and nothing at all if you could not have known, such as after a data breach or a cloned card. Debit cards carry similar protection but the money has already left your account, so the wait is real. Ask the issuer explicitly whether they are treating it as a **fraudulent transaction** or a **dispute with a merchant** — the two go down different paths. **Bank transfers.** A transfer you authorised yourself is legally different from a payment someone made without you, and this is where most people are surprised. The 50-euro cap above applies to *unauthorised* transactions. When you were tricked into pressing "send" yourself — the category the industry calls authorised push payment fraud — the outcome depends on your country. The UK now requires banks to reimburse most victims of these scams up to £85,000, for payments made from 7 October 2024 onward, with claims accepted for 13 months and reimbursement normally within five working days. Elsewhere it is assessed case by case, and your bank's decision can usually be escalated to a financial ombudsman if you disagree. Either way: ask in writing, keep the answer in writing. **Speed is the variable you control.** Money sent by transfer sits in a "mule" account before being moved onward, often within an hour or two. If your bank reaches the receiving bank while it is still there, it can be frozen. That window is measured in hours, which is why step one is a phone call and not an email. **Gift cards.** Call the company that issued the card — not the shop that sold it — and read them the card number and the receipt. If the balance has not been spent yet, some issuers can freeze it. Keep the physical card and the receipt either way; both the issuer and investigators will ask. **Cryptocurrency.** Transfers are final and pseudonymous, and there is no bank to call. Do not spend money on "blockchain recovery" services; they are overwhelmingly fraudulent. Do record the wallet address, the transaction hash and the exchange you used, and report it — exchanges occasionally freeze funds when a law-enforcement request arrives fast enough, and your transaction data is genuinely useful to investigators tracing a network. If the money left in several payments over days or weeks, list every single one, including the small early ones. People instinctively report only the big transfer, but the small "test" payment at the start is often the one that proves the pattern — and it can be the difference between your case being logged and your case being investigated. ## If they got into your device or your accounts Money is only half of it. If the scammer talked you through installing something, asked you to read out a code, or "helped" you from their end, treat every account on that device as compromised. Turn off Wi-Fi or unplug the cable. This ends any live remote session immediately. Do not shut the computer down if you can avoid it — a technician may want to see what is running. Use your phone if the compromise was on the computer, or someone else's computer if it was your phone. Email first, always — whoever controls your email can reset the password on everything else, so it is the master key. Then, in order: online banking, then any account holding money or payment details, then everything sharing that old password. Attackers leave doors open behind them. In your email settings, look for: - **Forwarding rules** silently copying your mail to an unknown address - A changed **recovery email or phone number** - Unfamiliar devices in the "where you are signed in" or "security activity" list — sign them all out - New **app passwords** or connected apps you do not recognise Do the same in your banking app: check saved payees, changed contact details, and any new payment limits. If they asked you to install anything to "fix" your computer or "verify" your account — names like AnyDesk, TeamViewer, UltraViewer and QuickSupport come up constantly — uninstall it. If you are not confident doing that, take the machine to a repair shop and tell them plainly what happened. Do not use that computer for banking until it has been checked. Once passwords are changed, add a second step to email and banking. An authenticator app is stronger than SMS, but SMS is far better than nothing. If you already had it on and they got in anyway, they talked you into reading out the code — that is the most common route, and knowing it is what stops it happening twice. ## What will not come back — and why to report it anyway Being honest about this is more useful than false hope. Cash handed to a courier, gift card codes read out over the phone, and cryptocurrency sent to a wallet are, in the overwhelming majority of cases, permanently gone. Money wired abroad and moved on within hours usually is too. Report it regardless, for reasons that are not about your refund: - **Your report is a data point in someone else's case.** Fraud networks reuse accounts, phone numbers, wallets and scripts. A single report looks like bad luck; four hundred reports naming the same account is what gets it frozen and its owner traced. - **It creates an official record.** You may need a police reference number for an insurance claim, a tax question, a dispute with your bank, or an ombudsman complaint months from now. Getting it later is much harder. - **It protects the next person.** Warnings issued by fraud bodies come from reported cases. The scam that caught you is being run on someone else this afternoon. - **Recovery does happen — occasionally.** Frozen mule accounts, exchange holds and seized funds do sometimes return money to victims, sometimes a year or more later. That can only reach people who are in the file. ### The second wave: people who target victims Expect to be approached again. Lists of scam victims are traded between criminal groups, and a person who has already lost money to one scheme is a proven target for the next. The follow-up usually arrives within weeks and takes one of these shapes: - A "**fund recovery agency**" or "asset recovery lawyer" who already knows what happened to you and asks for an upfront fee, a "court filing cost" or a "tax payment" to release your money - A "**law enforcement officer**" from a body you have never heard of, telling you your money has been located and needs a fee to be repatriated - A "**bank fraud department**" calling to "reverse the transaction", who needs your codes to do it The rule is simple and has no exceptions: **nobody legitimate contacts you first and asks for money to give you money back.** Real police, real regulators and real banks do not charge to investigate. If they found you, they bought your details. ## Where to report it Report to your bank *and* to your national reporting body — they do different jobs. The bank chases the money; the reporting body builds the case and the statistics. **United States** - [Report fraud to the FTC](https://reportfraud.ftc.gov/) — the national consumer fraud database - [FBI Internet Crime Complaint Center (IC3)](https://complaint.ic3.gov/) — report here fast for wire transfers. Its Recovery Asset Team can attempt to freeze a wire before it is withdrawn, though that process is aimed at larger transfers reported within roughly 72 hours — report anyway if yours is smaller or older, because the case still counts - [IdentityTheft.gov](https://www.identitytheft.gov/) — if they got your personal details, this generates a recovery plan - [Where to report scams, by type](https://www.usa.gov/where-report-scams) — USA.gov's directory for anything not covered above **United Kingdom** - Report to your bank first — under the [APP fraud reimbursement protections](https://www.psr.org.uk/information-for-consumers/app-fraud-reimbursement-protections/), most victims of transfer scams must be reimbursed up to £85,000 - [Report Fraud](https://www.reportfraud.police.uk/) — the national reporting centre for England, Wales and Northern Ireland, reachable online any time or on 0300 123 2040 (this is the service that used to be called Action Fraud); in Scotland, report to Police Scotland on 101 **Canada** - [Canadian Anti-Fraud Centre](https://antifraudcentre-centreantifraude.ca/index-eng.htm) — plus a report to your local police service **Australia** - [Scamwatch — report a scam](https://www.scamwatch.gov.au/report-a-scam) — run by the National Anti-Scam Centre **Ireland** - [An Garda Síochána — fraud](https://www.garda.ie/en/crime/fraud/) — report at your local Garda station - [FraudSMART](https://www.fraudsmart.ie/) — the banking industry's fraud awareness resource **Elsewhere in Europe** - [Report cybercrime online](https://www.europol.europa.eu/report-a-crime/report-cybercrime-online) — Europol's directory of national reporting points for every EU member state - [Your rights on payments and transfers](https://europa.eu/youreurope/citizens/consumers/financial-products-and-services/payments-transfers-cheques/index_en.htm) — what EU law entitles you to when a payment was not authorised by you ## The week after The first day is about the money. The week after is about making sure it is finished. **Watch your statements closely for a month.** Card details and account numbers get reused and resold. Check every line, including small amounts — testing a stolen card with a tiny charge before a large one is standard practice. **Follow up your claim in writing.** A week after your call, email the bank asking for the status of case number X. Verbal fraud reports get lost; emailed ones create a paper trail, and a paper trail is what an ombudsman will ask for. **Consider a credit freeze or fraud alert** if your identity documents or personal details were exposed, not just your money. In the US that is done free with the three credit bureaus; in most of Europe your bank or national credit register can flag the account. **Change the story you tell yourself.** People who have been scammed often withdraw from technology entirely — no more online banking, no more email, no more answering the phone. That reaction is understandable and it makes life smaller without making it safer. The better response is a small set of rules that do not depend on your judgement in the moment: a [family safe word](/guides/family-safe-words) that anyone claiming an emergency must give, a habit of hanging up and calling back on a number you look up yourself, and a standing agreement with one family member that no money moves without a second opinion. Those rules fit on [a printable card](/family-card) meant to sit by the phone — free to print, copy and hand out. ## What to try next If the contact came by phone and the voice sounded like someone you know, read [how to detect an AI voice on a phone call](/guides/detect-ai-voice-phone-call) — the tells are subtle but learnable. If it started as a message or email, [how to spot AI phishing emails](/guides/spot-ai-phishing-emails) covers what the written signals look like now that the grammar mistakes are gone. And if the scam involved an "investment platform" with a dashboard showing your growing balance, [how the fake AI trading platform scam works](/guides/ai-investment-scam-fake-platform) explains each stage and where the exit doors are. ## Sources - [What To Do if You Were Scammed](https://consumer.ftc.gov/articles/what-do-if-you-were-scammed) — Federal Trade Commission - [Gift Card Scams](https://consumer.ftc.gov/articles/avoiding-and-reporting-gift-card-scams) — Federal Trade Commission - [APP fraud reimbursement protections](https://www.psr.org.uk/information-for-consumers/app-fraud-reimbursement-protections/) — Payment Systems Regulator (UK) - [Payments, transfers and cheques — your rights](https://europa.eu/youreurope/citizens/consumers/financial-products-and-services/payments-transfers-cheques/index_en.htm) — Your Europe, European Commission - [Payment services](https://finance.ec.europa.eu/consumer-finance-and-payments/payment-services/payment-services_en) — European Commission - [Report cybercrime online](https://www.europol.europa.eu/report-a-crime/report-cybercrime-online) — Europol --- # The New Siri Is a Different Assistant Entirely — What iPhone Users Will Actually Notice Source: https://aiforregulars.com/guides/siri-ai-whats-new If you've ever asked Siri something and watched it open a web search instead of actually helping, you know the frustration. For years, Siri's biggest limitation wasn't that it misheard you — it's that it could only do one thing at a time, and even that one thing often landed you in Safari. Apple announced a fundamental rebuild of Siri at its developer conference in June 2026, and what's changing is not a minor feature update. The goal is a Siri that can carry out a real chain of actions across your apps, see what you're looking at, and remember what you talked about. ## What Actually Changed Under the Hood Previous versions of Siri worked by matching your words to a fixed list of commands. That's why it could set a timer but couldn't combine that with anything else, and why phrasing a request slightly differently would throw it off entirely. The rebuilt Siri uses a new architecture and connects more deeply to Apple's App Intents system — the same technology that lets apps expose their features to Shortcuts. That connection is what makes multi-step actions possible. Instead of matching your sentence to a single command, Siri can now plan a sequence of steps across different apps to complete a goal. **Why this matters in practice:** asking Siri to help you get somewhere used to mean "open Maps." Now it can mean checking your calendar for where you need to be, finding the route, and sending a message to let someone know you're on your way — all in one request. This is the same category of capability Apple started rolling out under the Apple Intelligence umbrella in 2025. If you want background on how Apple Intelligence fits in, the [Apple Intelligence starter guide](/guides/apple-intelligence-guide) covers the full picture. ## What Multi-Step Actions Look Like Day to Day The clearest way to understand what's new is to walk through examples that would have been impossible for the old Siri. **Checking in on a plan:** You ask, "Do I have anything on tomorrow afternoon, and if so, remind me to leave an hour early?" Siri reads your calendar, finds the appointment, and sets the reminder automatically. You didn't open the Calendar app or the Reminders app. **Following up on a payment:** You ask Siri to split a dinner bill with a friend using Apple Cash. Siri opens the payment request, fills in the amount, and sends it — rather than just opening the Wallet app and leaving you there. **Getting information from a photo:** Point your camera at a meal, tap the Siri button, and ask about what you're looking at. Siri analyzes the image and gives you a nutritional quality ranking — along with qualitative insights like whether the food is heavily processed, high in protein, or high in sugar — rather than a specific calorie count. These aren't demos of what AI can theoretically do. They were shown at Apple's WWDC keynote in June 2026 as features planned for this release. A few of the more involved ones may not ship in the initial update — see the rollout section below. > Try asking Siri: "Check if I have any conflicts next Friday afternoon and, if I'm free, text my sister that I can help her move." The most reliable way I've found to use the new multi-step Siri is to describe what you want to accomplish, not how you want it done. Instead of "open Reminders and add…", try "remind me to call the dentist the morning after my next dentist appointment." Siri handles the routing — you just say what you need. ## Camera Mode: Point and Ask One of the more visually distinct new features is Camera mode. You access it by tapping the shutter button icon in the Siri interface, and it lets Siri see whatever your camera is pointed at. **Practical uses:** - Photograph a receipt and ask Siri to send the total to someone or log it. - Point at a label you can't read clearly and ask Siri to read it out loud. - Show Siri a document and ask it to summarize the main points. - Photograph an unfamiliar plant, an object, or a sign and ask what it is. The experience is designed to feel like asking a person who can see what you're looking at, rather than typing a text description into a search box. ## The New Siri App: Conversations That Carry Over For the first time, Siri has its own dedicated app — not just a voice overlay that disappears when you look away. The app stores your conversation history and syncs it across your Apple devices. **What this changes:** If you ask Siri something on your iPhone and want to follow up from your iPad, you can. The conversation is still there. This is something users of ChatGPT and Google Gemini have had for a while; Apple is now bringing the same continuity to Siri. **What to know about history:** By default, your conversation history is stored on your device and synced via iCloud. As with any cloud-synced data, you can delete individual conversations or turn off history storage entirely in settings if you'd rather not keep a record. You can open the Siri app like any other — tap its icon on your home screen — and type or speak your questions there. This gives people who prefer not to talk out loud to their phone a more comfortable interface. ## Screen Awareness: Siri Sees What You're Looking At The rebuilt Siri can, with your permission, understand what's currently on your screen. This was introduced as part of Apple Intelligence in 2025; the new Siri makes it more practical for everyday use. **Example:** You're looking at an email with an event date mentioned in the body. You say "add this to my calendar." Siri reads the date and creates the event. You don't copy anything or type anything. **Example:** You're in a recipe app. You ask "what would I need to buy for this?" Siri reads the ingredient list and either adds items to your Reminders or tells you what's on it. This is where the line between a voice command and a genuinely capable assistant starts to mean something. ## Which Devices Get the New Features Apple has confirmed support for recent iPhone models. Reported at launch: iPhone 15 Pro and iPhone 15 Pro Max, and all models of iPhone 16 and newer. Older devices and those without the required chip will continue to get the existing Siri — timers, calls, weather, music — but not the new multi-step and camera features. If you're not sure whether your device qualifies, go to **Settings > General > About** and check your model name, then compare it to the support list on Apple's website. ## What Stays Exactly the Same If your relationship with Siri is mostly "set a ten-minute timer" and "call Mom," nothing changes. The basic voice commands people rely on every day — timers, alarms, phone calls, playing music, weather, messages — remain unchanged and work on the same devices as before. You don't need to learn a new way of talking to Siri to keep using it for simple tasks. ## A Realistic Look at the Rollout Apple's major features rarely all arrive on day one. The company has a consistent pattern of announcing capabilities at WWDC in June and shipping them in stages — some in the initial fall software update, others in point releases through the following winter and spring. Based on Apple's typical pattern and reporting at time of publication, expect the core multi-step features and Camera mode to arrive with the first update, and some of the more advanced integrations — particularly those that depend on third-party app support — to follow over the months after. Watch for a software update notification on your iPhone. If you're in a supported region and on a supported device, the update will appear automatically. ## What to Watch Out For **Privacy:** When Siri sees your screen or processes a camera image, that information is handled somewhere. Apple has said on-device processing is used where possible and that data sent to servers is not permanently stored, but review the privacy settings under **Settings > Siri & Search** to understand exactly what's enabled. **Apps need to support it:** Multi-step actions work best with Apple's own apps and those that have added App Intents support. Siri can't reach into an app that hasn't opened that door. If a request doesn't work the way you expected, the app may simply not have built that connection yet. **Staged rollout by language and region:** English typically ships first. If you're using Siri in another language and the new features don't appear right away, they will likely arrive in a subsequent update. **What the new Siri still can't do:** It won't catch a mistake in your bank statement or draft a legal document. Think of it as a capable assistant who knows your apps very well — not as a fully autonomous agent acting on your behalf without your oversight. ## What to Try Next To understand how the new Siri fits into the broader picture of AI on Apple devices, the [Apple Intelligence starter guide](/guides/apple-intelligence-guide) covers what Apple introduced before this update and how the pieces connect. If you're curious how Siri now compares to voice assistants on other devices, [Alexa vs. Google Assistant vs. Siri: Which One Is Actually Useful?](/guides/alexa-google-siri-comparison) lays out the comparison. --- # How to Spot AI-Generated Photos (Fake Profiles, Fake Products) Source: https://aiforregulars.com/guides/spot-ai-generated-photos A photo that looks like a real person, a real customer review, or a real news event can be generated by AI in seconds — for free, by anyone. Most of the time these images are harmless. But they are increasingly used in fake profiles, fraudulent product listings, and misleading content online. The old advice — count the fingers, read the background text — often fails against today's top generators, and advice that fails is worse than no advice at all, because a photo that "passes" the test earns more trust, not less. The most reliable move now is to find out where an image came from first, then look at the visual details. Before hunting for visual flaws, check whether the image has already turned up somewhere else. Upload the photo (or paste its URL) at [images.google.com](https://images.google.com) or [tineye.com](https://tineye.com) — both are free and require no account. The results show you everywhere else that image has appeared online. If a "new" profile photo turns up under a different name, on a different site, or dated years earlier, you have your answer regardless of how convincing the photo looks. On a phone, a long press on the image usually offers a "search image" option, or you can install the TinEye browser extension. Content Credentials is a standard — technically called C2PA — that attaches verifiable metadata to an image about its origin: what camera or tool created it, whether AI touched it, and by whom. The newest phones (Google's Pixel 10, for example) and several editing tools can now sign this metadata. You can check an image by uploading it at [verify.contentauthenticity.org](https://verify.contentauthenticity.org) or [contentcredentials.org](https://contentcredentials.org), or via the "Content Credentials" browser extension for Chrome, which adds a right-click option. There is one major catch: the label is strong evidence when it is *present and valid*, but its absence proves nothing. Many social platforms strip metadata on upload, and plenty of photos never carried any to begin with. This used to be one of the most reliable tests, but it largely no longer holds against today's leading generators — models have learned to draw hands with the correct number of fingers. Errors still show up most often in quickly or cheaply generated images, or in unusual poses: clenched fists, interlaced fingers, a hand partly hidden behind an object. If a hand is clearly visible, glance at it — but if it looks fine, that alone does not mean the photo is real. Treat it as one of several minor clues, not a deciding test. Text inside an image — a shop sign, a book cover, a product label — used to give AI generators serious trouble, often coming out as a garbled approximation of letters. The newest tools handle this far better, so what used to be a dependable test is now just a supporting clue as well. Cheaper or faster generators still produce wrong letter combinations, reversed characters, or repeating sequences that spell nothing. If text in an image clearly doesn't make sense, that is still a useful signal — just don't rely on it alone. Human faces are slightly asymmetrical in real life; AI faces sometimes look too symmetrical or too smooth. Ears can merge with hair or have an odd cartilage shape; teeth sometimes blur into a uniform stripe of white. It's also worth checking whether shadows all fall consistently from an apparent light source, whether background patterns (brickwork, floor tiles, crowds of people) repeat unnaturally, and whether small accessories — earrings, glasses frames, a necklace chain — look warped or asymmetrical, which AI images still occasionally get wrong. None of these details is proof on its own today, but combined with other signs, they're still worth noticing. I used to do a quick finger check on profile photos whenever an account felt slightly off. These days I lead with a reverse image search instead — against newer generators it catches far more fake profiles than looking at hands ever did. I most often catch AI photos this way on product review sites and in comment sections promoting health supplements. Visual inspection and reverse search are only part of the job. Ask: is this profile attached to an account with real, organic activity — tagged photos, comments from other real-looking people, a consistent posting history? Does this product review include specific, believable details about the purchase, or just general praise? Who posted the image first, and when? Context checks catch what visual inspection misses. And as AI image quality keeps improving, context and provenance will matter more than the visual signs. ## An Honest Note: Detection Is Getting Harder The visual checks in this guide work on only part of today's images, and their reliability keeps falling — research on the newest image-editing techniques has found commercial detector accuracy dropping close to a coin flip once obvious global artifacts are controlled for. What would have clearly failed the finger or text check a year ago now sails through. That does not make these checks useless — it means the visual alone is no longer enough, and needs to be paired with a reverse image search, a Content Credentials check, and a look at who posted the image first. Dedicated AI detectors exist, but treat them as a supporting signal, not a verdict — their accuracy varies and they lag behind the newest generators. No single check is a reliable detector by itself; combining several signs with a look at provenance gives you a reasonable basis for healthy skepticism. **What to try next:** These same instincts apply to video — see [How to Spot a Deepfake Video](/guides/how-to-spot-deepfake-video) for the moving-image version of this guide. If you want to test dedicated detection tools, [AI Detectors Tested](/guides/ai-detectors-tested) covers what actually works and where current tools fall short. ## Sources - [AI-Generated Image Detectors Overrely on Global Artifacts: Evidence from Inpainting Exchange](https://arxiv.org/abs/2602.00192) — Nebioglu, Bilgiç and Popescu, arXiv, 2026 - [Content Credentials](https://contentcredentials.org/) — Coalition for Content Provenance and Authenticity (C2PA) - [Verify Content Credentials](https://verify.contentauthenticity.org/) — Content Authenticity Initiative - [Google Pixel 10 Adds C2PA Support to Verify AI-Generated Media Authenticity](https://thehackernews.com/2025/09/google-pixel-10-adds-c2pa-support-to.html) — The Hacker News - [TinEye — Reverse Image Search](https://tineye.com/) — TinEye - [Google Images](https://images.google.com/) — Google --- # AI Phishing Emails: How to Spot Scams Now That the Grammar Is Perfect Source: https://aiforregulars.com/guides/spot-ai-phishing-emails For years, the advice was simple: if an email has typos and awkward phrasing, it is probably a scam. That advice is now outdated. AI writing tools produce polished, natural-sounding text in seconds, and phishing emails today are often indistinguishable in tone and grammar from real company communications. The good news: scammers still have to work around technical realities and human psychology that leave detectable patterns. These six signals don't require any special software — just a moment of attention. Every email client shows a "from" name, which anyone can set to anything. "PayPal Security Team" can be displayed by anyone sending from scammer123@randomdomain.xyz. Click or tap on the sender name to expand and see the full email address. Look at the domain — the part after the @ symbol. If an email claiming to be from your bank comes from a domain that is not your bank's real website address, it is fake. Watch for subtle substitutions like "paypa1.com" (the number 1 instead of the letter l) or added words like "paypal-support.com." On a computer, rest your mouse cursor over any link in the email without clicking. The actual destination URL appears in the bottom-left corner of your browser or email client. On a phone, press and hold a link to see the destination URL before opening it. Compare that URL to the company's real website address. If they don't match, don't click. Even if the link text says "amazon.com," the actual destination could be anywhere. AI makes scam emails grammatically perfect, but it cannot remove the fundamental mechanic that makes phishing work: panic. Scammers need you to act before you think. Watch for phrases like "your account will be closed in 24 hours," "immediate action required," "unauthorized access detected," or "verify now to avoid suspension." Legitimate companies send these kinds of warnings too — but they give you days to respond, not hours, and they don't threaten permanent loss in the first sentence. When an email is trying to make you feel rushed, that is exactly when you should slow down. Real companies almost never ask you to confirm your full password, Social Security number, or payment card details by clicking a link in an email. They already have your account on file. When an email asks you to "confirm your information" by entering it again somewhere, that is the scam's core mechanic — harvesting your credentials. Also watch for requests that seem slightly off for the company: a shipping notification asking you to pay a customs fee, a streaming service asking you to update billing by clicking a link, or an IT department asking for your login credentials "for maintenance." If an email claims there is a problem with your account, don't use any link, phone number, or attachment in that email. Instead: Open a new browser tab and type the company's address directly — or use a bookmark you created yourself. Log in there and check whether the same alert appears in your account. Call customer service using the number on the company's official website or on the back of your card, not a number from the email. This one step stops virtually all phishing attacks cold. Every major email service — Gmail, Outlook, Apple Mail — has a "report phishing" or "mark as spam" option. Use it. These reports help train filters that protect everyone else from the same campaign. You can also forward suspicious emails to reportphishing@apwg.org, a nonprofit that tracks phishing campaigns globally. If the email impersonates a specific company, that company's security team wants to know — search for "[company name] report phishing" to find their dedicated address. When I am unsure about an email, I do one thing: I go directly to the company's website by typing the address myself and check my account there. If the email was real, the same alert will be waiting in my account dashboard. If nothing is wrong, the email was fake. This approach takes about 30 seconds and has never let me down. ## What to Try Next Phishing emails are often the first step in a larger attack. [Fake customer service scams](/guides/fake-customer-service-scam) explains what happens after you click — how scammers impersonate real brand support teams and collect your credentials — and how to find a company's real contact information. For a look at how AI-generated voice fits into these attacks, [AI voice cloning scams](/guides/ai-voice-cloning-scams) explains what the technology sounds like and how to stay ahead of it. ## Sources - [Report Phishing](https://apwg.org/reportphishing/) — APWG (Anti-Phishing Working Group) - [How To Recognize and Avoid Phishing Scams](https://consumer.ftc.gov/articles/how-recognize-avoid-phishing-scams) — Federal Trade Commission (FTC) - [Report spam or phishing messages in Gmail](https://support.google.com/mail/answer/8253) — Google - [Report and reduce spam in iCloud Mail](https://support.apple.com/en-us/102376) — Apple --- # AI Scam Text Messages: How to Spot Them Now That They Read Like Real Texts Source: https://aiforregulars.com/guides/spot-ai-scam-text-messages A few years ago, spotting a scam text was easy: bad grammar, a weirdly formatted link, a sender who clearly didn't speak English as a first language. AI writing tools have quietly ended that era. Today's scam texts — called "smishing," short for SMS phishing — can be grammatically perfect, personalized with your actual name or a real recent order, and tuned to sound exactly like your bank, delivery company, or even a family member. The good news: the underlying tricks haven't changed, even if the writing has gotten better. These six checks catch most AI-polished scam texts in under a minute. A text claiming to be from your bank, a delivery company, or a government agency should come from a short code or a number you recognize from past legitimate messages — not a long, random phone number, an international number, or an email-address-looking sender. Tap on the sender to see the full number if your phone doesn't already show it. If a message claiming to be "USPS" or "your bank" arrives from a personal-looking 10-digit number instead of the company's usual short code, that's a strong warning sign on its own. This is the single most effective habit against smishing. Instead of tapping any link in a text, open a new browser tab (or the company's own app) and type the address yourself, or use a bookmark you already trust. If the text claims there's a package waiting, a toll unpaid, or account activity to review, the same information — if it's real — will be sitting there when you log in directly. If nothing shows up, the text was fake. AI has made scam texts sound natural, but the setups themselves are familiar once you know to look for them: an undelivered package needing a small "customs fee," an unpaid toll with a threat of licence penalties, a bank fraud alert asking you to "verify" your card, a missed jury summons, or a job offer that starts by text out of nowhere. What's changed is the personalization — AI tools can pull your name, city, or even a real recent purchase from data leaked in past breaches and drop it into the message, making it feel far more targeted and credible than a generic blast. Scam texts are built to make you act before you think: "final notice," "your account will be suspended today," "reply within 2 hours." Real notifications from banks, delivery services, and government agencies almost never demand action within hours over text. Also treat any request to reply with a word (even something like "Y" or "C"), click through to "confirm," or pay a small fee by card as a red flag. Replying at all — even to opt out — can confirm to a scammer that your number is active. If a text claims something urgent about your bank, delivery, or a government matter, don't use any link, phone number, or QR code from the text. Open the company's official app, or a website address you type yourself, and check your account there. For anything claiming to be a government notice, use the agency's official .gov (or your country's equivalent) website directly. Calling the number printed on your bank card or a past bill also works — never a number provided inside the suspicious text itself. Recent phones can help catch these automatically. Google Messages now includes an on-device AI scam-detection feature for Android that flags suspicious message patterns in real time — it launched in the US, UK, and Canada and has since expanded to more than 20 countries, supporting several languages including English, Arabic, French, German, Portuguese, and Spanish, and all the analysis happens on your phone rather than being sent anywhere. iPhone users can filter messages from unknown senders into a separate list under Settings, so unrecognized senders don't land straight in your main inbox. After checking a message, forward it to 7726 (spells "SPAM") to report it to your carrier — this works in both the US and the UK — then block the sender and delete it. The habit that's saved me the most: I never act on anything a text tells me to do inside that same text thread. If it says my bank needs me, I close the message and open my banking app separately. If it's real, the same alert is waiting for me there. If it isn't, I've lost nothing by checking. ## What to Watch Out For **AI has also learned to imitate people you know.** Scammers increasingly use AI to write a text or voice message that sounds like your adult child, grandchild, or a coworker asking for urgent help — often money sent quickly. If a text from a "known" number feels off, or asks for money or gift cards, call that person back on a number you already have saved, not one provided in the message. **Don't trust caller ID or sender names on their own.** Both can be faked (called "spoofing") to display a real company name or even a real contact's name. Treat the sender name as a hint, not proof. ## What to Try Next Scam texts are often just one piece of a bigger playbook that also uses AI-written emails and cloned voices. [How to spot AI phishing emails](/guides/spot-ai-phishing-emails) covers the same tactics applied to your inbox, and [how to detect an AI voice on a phone call](/guides/detect-ai-voice-phone-call) is worth reading if a "text from a relative" turns into a phone call that doesn't quite sound right. ## Sources - [FCC Proposes First AI-Generated Robocall & Robotext Rules](https://www.fcc.gov/document/fcc-proposes-first-ai-generated-robocall-robotext-rules-0) — Federal Communications Commission - [Staying One Step Ahead: Strengthening Android's Lead in Scam Protection](https://blog.google/security/staying-one-step-ahead-strengthening-androids-lead-in-scam-protection/) — Google - [How To Recognize and Report Spam Text Messages](https://consumer.ftc.gov/articles/how-recognize-and-report-spam-text-messages) — Federal Trade Commission - [Protect Against Fraud](https://www.mobileuk.org/protect-against-fraud) — Mobile UK - [View conversations from unknown senders in Messages on your iPhone](https://support.apple.com/en-us/125068) — Apple Support --- # How to Spot Fake AI-Generated Product Reviews Before You Buy Source: https://aiforregulars.com/guides/spot-fake-ai-product-reviews A product with a near-perfect five-star rating used to be a reliable shortcut for a good purchase. It isn't anymore. AI can now write a convincing, detailed-sounding review in seconds — and it can write hundreds of slightly different versions just as fast. You don't need to become a detective to avoid getting fooled by them. You just need to know what to look for, and it takes less than a minute per listing. ## Red Flag 1: The Writing Sounds a Little Too Perfect Real customers write the way people talk — a bit messy, focused on one or two things that stood out to them, sometimes with a typo or an unfinished thought. AI-generated reviews tend to read more like a school essay: a clear opening sentence, three neatly organized points, and a tidy wrap-up. Nobody writes an Amazon review like that unless something is off. **Look for missing specifics.** A real review usually anchors itself in a real moment — "the zipper broke after about three weeks" or "my husband uses it every morning for his coffee." AI-written reviews tend to stay general: they'll recite the product's features back at you (battery life, materials, dimensions) without describing what it was actually like to use the thing. **Watch for reviews that read like a spec sheet.** If a review sounds like it's summarizing the product listing rather than a personal experience, that's a tell. A real buyer talks about how something feels in the hand or fits in a bag — not just its measurements. ## Red Flag 2: The Reviewer Profile Has Almost No History Click on the reviewer's name. On Amazon, this shows you their other reviews and when their account was created. **A brand-new account with only glowing reviews for one brand is a warning sign.** Real people who leave reviews tend to review a mix of things over time — a book, a kitchen gadget, a pair of shoes. A profile that has posted five perfect reviews for the same seller's products, and nothing else, usually isn't a real customer. **No "Verified Purchase" label deserves extra scrutiny.** This label means Amazon confirmed the reviewer actually bought the item through the site. Reviews without it are worth weighing more carefully, especially if they're glowing. **Generic usernames** — random letter-and-number combinations rather than a name — are common on fake accounts created in batches, though this alone isn't definitive. When I'm deciding between two similar products, I sort the reviews by "Most Recent" instead of "Top Reviews" and read the three-star ones first. Three-star reviews are the ones people write when they actually have something specific, mixed, and honest to say — "great for the price, but the handle feels flimsy" is a sentence I trust a lot more than twenty identical five-star raves. ## Red Flag 3: A Burst of Reviews, All at Once Fake reviews often arrive in waves rather than trickling in naturally the way real purchases do. **Check the dates.** If a product has 200 reviews and 150 of them were posted within the same two-week window, that's not how real buying behavior looks — genuine reviews accumulate gradually as different people receive their orders over months. **Watch for a U-shaped pattern.** A mix of mostly five-star and one-star reviews, with very few three- or four-star reviews in between, often signals manipulation — a batch of paid glowing reviews sitting alongside genuinely disappointed buyers, with little of the honest middle ground you'd expect from a real spread of opinions. **Repetitive phrases across different reviews** — the same unusual turn of phrase showing up in reviews from supposedly unrelated people — is one of the clearer signs of a coordinated, AI-assisted campaign. ## Why This Is Getting Harder to Spot This isn't a small problem. Amazon has said it uses AI — including large language models and deep neural networks that map relationships between accounts — to screen reviews before they go live, and reported blocking more than 200 million suspected fake reviews worldwide in 2022. Josh Meek, a senior data science manager on Amazon's Fraud Abuse and Prevention team, has been candid that the difference between an authentic and a fake review "is not always clear for someone outside of Amazon to spot." It's also against the law in the US. The Federal Trade Commission's rule banning fake and AI-generated reviews took effect in October 2024, with penalties that now run to $53,088 per violation. In late 2025, the FTC began actively enforcing it, sending warning letters to companies suspected of violations. That's a meaningful deterrent, but it doesn't stop every fake review from reaching a product page before it's caught — which is why it still helps to know the signs yourself. ## Can I Just Use an AI Tool to Check for Me? Several apps and browser extensions promise to scan a product's reviews and flag the fake ones automatically, and researchers are getting better at this: one study published in 2026 described a detection system that correctly identified fake reviews with 93% accuracy on Amazon and 91% on Yelp by combining the language of the review with behavioral signals, like whether the tone matches the star rating. That's a promising result from a research lab — but it's not the same as the free browser extension or app you might download. Consumer-facing AI text detectors have a track record of inconsistent real-world accuracy and false positives, flagging genuine reviews as fake and missing others. Treat any tool's score as one more data point, not a verdict. The manual checks above — writing style, profile history, timing — will usually tell you more than a single automated score. ## What to Do When You Spot a Fake Review **Filter by "Verified Purchase" only.** Most shopping sites let you filter reviews to show just confirmed buyers, which strips out a large share of fake ones immediately. **Check a second source.** Look the product up on an independent site or ask someone you know who's bought it. If the glowing reviews on the retailer's page don't match what you find elsewhere, trust the mismatch. **Report it.** Amazon has a "Report" link on individual reviews, and you can report suspected fake reviews or misleading listings to the FTC at reportfraud.ftc.gov. **Return it if it doesn't match.** If you bought something based on reviews that turn out to have been fake, the standard return window still applies. You don't need to prove the reviews were fake to get your money back — just that the product wasn't what you expected. ## What to Try Next Fake reviews and fake profiles often come from the same playbook — [How to Spot AI-Generated Fake Profiles and Posts on Facebook](/guides/ai-fake-facebook-profiles) covers the same red flags applied to social media. If you're wondering whether an app itself is legitimate before you even get to the reviews, [Is That App Actually AI, or Just a Scam?](/guides/is-ai-app-scam) is the next place to look. ## Sources - [How Amazon is using AI to spot fake reviews](https://www.aboutamazon.com/news/policy-news-views/how-ai-spots-fake-reviews-amazon) — Amazon - [Federal Trade Commission Announces Final Rule Banning Fake Reviews and Testimonials](https://www.ftc.gov/news-events/news/press-releases/2024/08/federal-trade-commission-announces-final-rule-banning-fake-reviews-testimonials) — Federal Trade Commission - [A warning letter (or ten) for businesses: comply with the FTC's Consumer Review Rule](https://www.ftc.gov/business-guidance/blog/2025/12/warning-letter-or-ten-businesses-comply-ftcs-consumer-review-rule) — Federal Trade Commission - [New AI system improves detection of fake online reviews](https://www.eurekalert.org/news-releases/1128975) — EurekAlert! / University of East London --- # AI Is Reading Your Email by Default — Here's How to Turn It Off Source: https://aiforregulars.com/guides/stop-ai-reading-your-email Your email inbox has always been scanned automatically — that's how spam filters work and how search finds old messages. But something shifted around late 2025: that scanning expanded to include Gemini, Google's AI assistant, which can read your emails and answer questions about them on request. In the US, this feature was turned on quietly for most Gmail users, as a default, without a clear moment where you said yes. **In the European Economic Area, the UK, Switzerland, and Japan, it works the other way round:** Google keeps Gemini off by default in Gmail, and you have to turn it on yourself. It's still worth checking your own settings below — you may have switched it on at some point, or you may be using an account created outside those regions. This guide explains what's actually happening, who should check their settings, and the exact steps to do it in both Gmail and Outlook. ## What "AI Reading Your Email" Actually Means The phrase sounds alarming, but the reality is more mundane — and in some ways more useful than you might expect. When Gmail scans your email for AI features, automated systems process the text of your messages to generate Smart Reply suggestions ("Sounds good!", "Thanks, I'll take a look"), Smart Compose completions as you type, travel and purchase tracking cards, and — since the change in late 2025 — Gemini's ability to summarize a long email thread or answer questions like "What's the status of my Amazon order?" **No human at Google is reading your emails.** That's worth saying plainly, because the phrase "AI reading your email" can conjure images that aren't accurate. What's happening is that your message text is being processed by machine-learning systems, similar in kind to how a spam filter works — just more capable and covering more use cases. What changed around October 2025 is that this access was extended to Gemini. In the US and in other countries outside the EEA, Google switched that access from opt-in to opt-out — affecting a large share of the roughly 1.8 billion Gmail accounts worldwide. A class action lawsuit filed in November 2025 in California (Thele v. Google) argues that this switch happened without meaningful consent. Google maintains the change was disclosed in settings updates sent to users. **In the European Economic Area, the UK, Switzerland, and Japan, Google went the other way.** Processing personal data at this scale requires explicit prior consent under GDPR and similar laws in those countries, not an opt-out after the fact — so Google's own help documentation states that Gemini and smart features stay off by default for accounts based in those regions. If you're in one of them and never turned these settings on yourself, they are very likely still off — though it's worth confirming with the steps below, since accounts created outside those regions, and Google Workspace accounts managed by an administrator, can be exceptions. Microsoft Outlook has broadly equivalent features through Copilot, with similar settings you can adjust — though as covered in the Outlook section below, the deeper reading and summarizing features are tied to a paid Microsoft 365 subscription for personal accounts, unlike Gmail's free Smart features. ## Should You Turn It Off? **If you're in the EEA, the UK, Switzerland, or Japan, you may have nothing to turn off.** Because these settings require opt-in consent there, the large majority of Gmail accounts based in those regions already have Gemini and smart features switched off — you never turned them on. The steps below are still useful as a check that this is really the case, or as instructions for turning things off if you did enable them at some point, use an account created outside those regions, or use a Google Workspace account controlled by an administrator. Not everyone needs to opt out. Here's a quick way to think about it. **Consider opting out if** your inbox regularly contains sensitive information — business communications, legal correspondence, medical records, financial details — that you'd prefer not to have processed beyond basic spam filtering and delivery. The same applies if you're generally uncomfortable with the idea of any automated system having access to the full text of your emails, even without a human involved. **You might prefer to keep it on if** you find the AI features genuinely useful — Smart Reply saves you time, email summaries help you catch up after a vacation, or you've been using Gemini to search your inbox. The processing is automated, not human, and for most everyday personal email the privacy trade-off is relatively modest. A middle path: you can turn off Gemini's deeper access to your inbox while keeping Smart Reply and Smart Compose active. The steps below show you how. ## Turn It Off in Gmail (Web) Go to [mail.google.com](https://mail.google.com) and sign in. Click the gear icon in the top-right corner, then click **See all settings** at the top of the settings panel that slides out. Click the **General** tab if it isn't already selected. Scroll down until you reach a section called **Smart features**. The main toggle is labeled **Turn on smart features in Gmail, Chat, and Meet** — unchecking it disables Smart Reply, Smart Compose, and the AI-powered inbox features within Gmail. Below that, a **Google Workspace smart features** heading links to **Manage Workspace smart feature settings**, where you can toggle **Smart features in Google Workspace** and **Smart features in other Google products** independently. Uncheck what you want to disable. If you only want to limit Gemini's inbox access without losing Smart Reply, focus on the Google Account step below rather than changing these. Scroll to the bottom of the page and click **Save Changes**. Gmail will reload. Smart Reply buttons will disappear from the bottom of emails, and Smart Compose suggestions will stop appearing as you type. The Gmail toggles above cover in-app features. For a more complete opt-out — especially for Gemini's ability to access your inbox — also visit [myaccount.google.com/data-and-privacy](https://myaccount.google.com/data-and-privacy). Under **History settings**, look for **Gemini Apps Activity** — this is the specific toggle that controls Gemini's access to your Gmail inbox. Note that **Web & App Activity** is a separate, unrelated toggle for general browsing history; it does not govern Gemini's email access. ## Turn It Off in Gmail (Mobile — iPhone and Android) Open the Gmail app. Tap the three horizontal lines (menu icon) in the top-left corner. Scroll down and tap **Settings**. Tap your email address to open account-specific settings. **The path is the same on iPhone, iPad, and Android:** under the **General** section, tap **Data privacy**, then toggle **Smart features** off. Changing it on mobile updates the preference across all devices where you're signed in to the same account. ## Turn It Off in Outlook Microsoft Outlook's equivalent settings live under **Connected experiences** and control what Copilot and other AI-powered Outlook features can access from your email. Here's where Outlook differs from Gmail, though: Copilot's deeper features — reading, summarizing, and drafting replies based on your inbox content — are part of a paid Microsoft 365 subscription (Personal, Family, or Premium) for individual accounts, or a paid Copilot license at work. A free Outlook.com account does not get Copilot access to your inbox at that level on its own. The Connected experiences settings are still worth checking on a free account, though — they also govern smaller features (link previews, filling in contact details) that tend to be on by default regardless of whether you pay for Microsoft 365. Go to [outlook.com](https://outlook.com) or your Microsoft 365 mail. Click the **Settings** gear icon and navigate to **General > Privacy and data > Privacy settings**. There you can turn off categories of **Connected experiences** that process your email content, which limits Copilot's access to your messages. Outlook also has a dedicated **Settings > Copilot** section where you can turn Copilot off entirely — this toggle works independently of the Connected experiences settings. Open Outlook and go to **File > Office Account** and click **Manage Settings** under Account Privacy (on Windows), or go to **Outlook > Preferences > Privacy** (on Mac, which opens a Privacy preferences pane — not a separate dialog). There you'll find toggles for connected experiences, including the ones that analyze your content to power Copilot and other AI suggestions. I kept Smart Reply off for a week expecting to miss it, and found I didn't — the suggestions were often slightly off-tone for my writing style anyway. What I actually missed were the travel tracking cards that appear automatically when a flight confirmation lands in your inbox. Those are genuinely useful, and you give them up when you disable Smart Features entirely. If that matters to you, it's worth knowing before you disable everything. ## What You Lose When You Turn It Off Being honest about the trade-off matters. These features stop working once you opt out: **In Gmail:** Smart Reply buttons at the bottom of emails, Smart Compose auto-complete as you type, email thread summaries, purchase and travel tracking cards in your inbox, and Gemini's ability to answer questions about your emails ("Did I get a shipping confirmation for that order?"). **In Outlook:** Copilot email summaries, suggested replies based on thread context, and Copilot's ability to draft messages drawing on your email history. If you use any of these regularly, consider the middle path mentioned above — turning off data sharing with other Google products and reviewing Gemini's access specifically, rather than disabling every Smart Feature at once. ## What to Try Next If you're reviewing your AI privacy settings more broadly, [How to Opt Out of AI Training and Delete Your Data](/guides/opt-out-ai-training-delete-data) covers the chatbot side — what ChatGPT, Gemini, and others do with your conversations when you use those tools directly. For a bigger-picture view of what AI systems may already know about you, [What Does AI Know About Me?](/guides/what-does-ai-know-about-me) is a useful next read. ## Sources - [Learn about smart features & controls for Google Workspace & other Google products](https://support.google.com/mail/answer/15604322?hl=en) — Google (confirms smart features are off by default in the EEA, Japan, Switzerland, and the UK) - [Thele v. Google, LLC](https://www.classaction.org/news/thele-v.-google-llc) — Class Action.org (lawsuit over Gemini's opt-out default in the US) - [Are AI tools such as Gmail's Gemini accessing your emails? What you need to know](https://runbox.com/blog/2026/01/are-ai-tools-such-as-gmails-gemini-accessing-your-emails-what-you-need-to-know/) — Runbox - [Connected experiences in Microsoft 365](https://support.microsoft.com/en-us/privacy/connected-experiences-in-microsoft-365) — Microsoft - [Microsoft 365 Copilot Plans for Individuals](https://www.microsoft.com/en-us/microsoft-365-copilot/personal) — Microsoft (free vs. paid account features) --- # How to Make a Personalized Birthday Song with AI in Five Minutes (Free) Source: https://aiforregulars.com/guides/suno-make-birthday-song Your sister is turning 60. Your father is retiring. Your best friend just hit a milestone she's pretending not to notice. A card is lovely, but a song that uses her actual name, mentions her garden, and jokes about her morning coffee ritual lands differently. Until recently, creating a real, listenable personalized song meant musical training, studio equipment, or paying someone. With Suno, it takes five minutes and costs nothing. Go to **suno.com** and click **Sign Up**. You can sign in with a Google or Apple account — no credit card or email address required. The free plan gives you 50 credits per day, which works out to about 10 songs.:: On the home screen, you'll see a text box under **Create**. This is Simple Mode — type a description of the song you want. Include the person's name, a detail or two about them, and the musical style that would suit the occasion. Here's an example you can adapt: > A cheerful folk song for grandma Ruth who loves her garden and her morning coffee, sung by a warm female voice, short and sweet The more specific you get, the better the result. "A happy birthday song" gives you something generic. "A jaunty ukulele song for a 70-year-old who collects stamps and makes terrible puns" gives you something memorable. After a few seconds, Suno gives you **two different versions** from the same prompt. The two variations often take the idea in noticeably different directions. Listen to both before you decide. If neither version feels right, click **Regenerate**. Each regeneration costs two credits and produces two fresh versions. If you want to write the words yourself, click the **Custom** toggle at the top of the prompt box. You'll see two fields: one for your lyrics, one for the style. Use structure tags to show Suno where verses and choruses start: ``` [Verse] Happy birthday [Name], today's your special day You fill our lives with laughter in every single way [Chorus] Here's to you, [Name], here's to you May every year bring something new We love you, [Name], more than you know Happy birthday — watch you grow ``` In the **Style** field, describe the sound you want: "upbeat acoustic folk, warm female vocals, gentle guitar." Custom Mode gives you much more control over the words, which matters when you need to include a specific name, a family joke, or a memory that only your family knows about. Once you've found the version you want, click the **three-dot menu** (⋯) below the song and choose **Download**. The file saves as an MP3 to your phone or computer.:: Send the MP3 via your family group chat, attach it to a birthday email, or play it from your phone when everyone is in the room. If you'd rather share a link than a file, copy the song's URL directly from Suno — anyone can listen without creating an account. The first attempt often sounds a little generic. Before you give up, try adding one very specific detail you wouldn't normally think to include: the person's nickname, a place they're always talking about, or an object that's become a family running joke. I asked Suno for "a song for Uncle Bob who has called every dog he's ever owned Buddy" — the second attempt had a chorus about "Buddy number four." That kind of specificity is what makes the song feel like it was made for someone, because it was. ## Beyond Birthdays Once you know how Suno works, the occasions are everywhere: - **Anniversaries** — describe how the couple met, their shared hobby, how many years it's been - **Graduations** — the subject they studied, the city they're moving to, their dream - **A lullaby with the baby's name** — slow, soft, with the child's name woven through the words - **A retirement party roast** — gentle teasing about their love of spreadsheets or how they always arrived ten minutes early - **A "sorry I missed your birthday" make-good** — exaggerated guilt, comedic tone, guaranteed to get a laugh The formula is always the same: specific person + specific details + specific style = something that feels made for them, because it was. ## What to Watch Out For **Free songs are for personal use only.** The free plan lets you share a song as a gift, play it at a party, and post it on personal social media — but not in a commercial video, not for sale, not for business use. If you need commercial rights, Suno's paid plans cover that. **Don't try to clone a specific artist's voice.** Writing "sing this like Sinatra" or naming any living artist in a voice request won't replicate their voice — Suno is designed to avoid this. Describe vocal qualities instead: "warm baritone," "raspy blues singer," "bright pop vocal." **The first attempt may sound generic — that's normal.** Regenerate with one or two more specific details: a proper name, a concrete place, a specific object, an unexpected style combination. The more unusual the details, the more distinctive the result. **Be thoughtful with teasing lyrics.** Gentle ribbing is charming; songs that touch something the person is actually sensitive about are not. When in doubt, keep it warm. **Unusual names may get mispronounced.** AI-generated singing handles common first names well, but unusual or foreign names can come out strangely. If it matters, try respelling the name phonetically in the prompt — for example, "Siobhan" written as "Shivahn." ## What to Try Next A personalized song pairs beautifully with a short video — [How to Make an AI Video for Free](/guides/make-ai-video-free) shows you how to turn photos into a moving tribute. Or if you want a custom image to go with it, [How to Create AI Images for Free](/guides/create-ai-images-free) walks you through that from scratch. --- # How to Set Up Your Tablet for AI: iPad and Android Step by Step Source: https://aiforregulars.com/guides/tablet-setup-ai-beginners Tablets are a great way to use AI — the screen is big enough to read comfortably, and you can prop it up while you work or chat. But getting started can feel confusing if you're new to the device itself. This guide assumes you've just gotten your tablet and want to get it ready for AI use, starting from scratch. The steps below cover both iPad and Android tablets. Where things are different between the two, I'll say so clearly. ## Before you start: what you'll need - Your tablet, charged to at least 50% - A Wi-Fi connection - An email address (to create a free ChatGPT account) - About 20 minutes The first thing to do before installing anything is adjust the text size. The default is often too small for comfortable reading. **On iPad:** Go to Settings (the gray gear icon) → Display & Brightness → Text Size. Drag the slider to the right to increase the size. You can also turn on "Larger Accessibility Sizes" for even bigger text. **On Android:** Go to Settings → Display → Font size (Samsung labels this 'Font size and style' under Settings → Display; on Google Pixel tablets it lives under Settings → Accessibility → Display size and text; on Amazon Fire tablets, check Settings → Display → Font Size. When in doubt, open Settings and search 'Font size'.). Drag the slider toward "Large." Start larger than you think you need. You can always go back and reduce it once you've tried it for a day. Typing on a glass screen takes getting used to. Voice typing lets you dictate everything you want to say — and it works remarkably well for AI chats. **On iPad:** Go to Settings → General → Keyboard → Enable Dictation. Once it's on, a microphone icon will appear on the keyboard whenever you tap a text field. Tap the microphone, speak, and your words appear on screen. **On Android:** On most Android tablets, voice typing is already built into the Google Keyboard. Tap any text field, then tap the microphone icon on the keyboard. Speak clearly and tap the microphone again when you're done. Test it right now by opening the Notes app and speaking a sentence. Getting comfortable with voice typing before you start using AI apps makes everything easier. Beyond text size, you can make the whole screen easier to see. **On iPad:** Settings → Display & Brightness → adjust the brightness slider near the top. Also try turning on True Tone if your iPad supports it — it adjusts the screen warmth to match your lighting and is easier on the eyes. **On Android:** Settings → Display → Brightness. Also look for a "Reading Mode" or "Eye Comfort Shield" setting — many Android tablets include this to reduce blue light. If your tablet has a night mode (Android: "Night Light," iPad: "Night Shift"), turn it on for evening use. It shifts the screen toward warmer tones, which is gentler when using the tablet before bed. Now for the AI part. **On iPad:** Open the App Store (blue icon with a white letter A). In the search bar at the top, type "ChatGPT". Look for the app by OpenAI — it has a black and white logo. Tap "Get" to download it for free. **On Android:** Open the Play Store (colorful triangle icon). Search "ChatGPT" and look for the app by OpenAI. Tap "Install." Once it's installed, tap to open it. You'll be asked to create a free account. Tap "Sign up" and use your email address. You'll receive a confirmation email — tap the link in that email to finish setup. The whole process takes about three minutes. The ChatGPT app has its own voice mode that works even better than keyboard voice typing for conversations. Once you're in the app, look for a waveform icon in the bottom-right corner of the message bar. Tap it, and ChatGPT will speak to you out loud and listen for your voice. You can have a full conversation without touching the keyboard at all. This is especially useful if you want to ask questions while your hands are busy, or if you find the keyboard uncomfortable. I've watched several family members set up tablets for the first time, and the voice mode feature is almost always the moment they really take to it. When you can just say "what's a good recipe for chicken soup that isn't too salty" and hear a full answer spoken back, it stops feeling like a computer and starts feeling like a helpful assistant. If you try nothing else in this guide, try that. Tablets have built-in accessibility features that can make AI apps significantly more comfortable. **For vision:** - iPad: Settings → Accessibility → Display & Text Size → turn on "Bold Text" and "Increase Contrast" - Both: Look for a "Zoom" feature that lets you magnify any part of the screen with a triple-tap - iPad: Accessibility → Spoken Content → "Speak Screen" — swipe down with two fingers and the tablet reads any text out loud **For motor / typing difficulty:** - iPad: Settings → Accessibility → Touch → turn on "AssistiveTouch" — creates a floating button that can open apps without pressing physical buttons - Both platforms: Voice typing (set up in Step 2) is the main tool here **For hearing:** - Both: AI chat apps are text-based by default, so hearing difficulty generally doesn't affect use. If you use voice mode, turn on subtitles: Settings → Accessibility → Subtitles & Captioning (iPad) or Accessibility → Hearing Enhancements (Android). Before you close this guide, try asking ChatGPT three questions. Start with something simple: 1. "What's the weather going to be like this week?" (It may not know in real time, but it's a good test.) 2. "Explain what a dividend is in plain English." 3. "What are some things I can ask you to help me with?" That third question is often eye-opening. ChatGPT will give you a long list of practical uses you probably haven't thought of — and most of them are easier on a tablet than you'd expect. ## What to try next Now that your tablet is set up, the next step is learning what to actually say to get useful answers. The guide on [how to set up the ChatGPT app](/guides/chatgpt-app-setup) covers the settings inside the app itself — privacy, memory, and notifications. If you want to use voice mode regularly, [ChatGPT voice mode explained](/guides/chatgpt-voice-mode) goes deeper on how it works and what it's best for. --- # Is Your Teen Too Attached to an AI Chatbot? Warning Signs to Watch For Source: https://aiforregulars.com/guides/teen-ai-companion-warning-signs If your teenager spends time chatting with an AI — asking it questions, venting about their day, maybe even calling it a friend — they're far from alone. A Pew Research Center survey published in February 2026 found that 64% of US teens ages 13 to 17 have used an AI chatbot. Only 51% of their parents believed their teen used one at all. That gap matters less because parents are being kept in the dark on purpose, and more because most teen AI use is genuinely unremarkable: asking for homework help, looking something up, or having a bit of fun. The real question isn't whether your teen has talked to a chatbot — most have — it's whether that use has drifted from casual into something that's replacing real relationships. ## General Assistant or Companion App? The Distinction Matters Not all AI chats are the same. **General-purpose assistants** like ChatGPT, Gemini, and Copilot are built to answer questions, help with writing, and handle tasks. **Companion apps**, most notably Character.AI (and others like Replika), are built differently — around ongoing, personality-driven conversations designed to feel like an evolving relationship. Some teens interviewed by researchers described these bots using language like "soulmate," and use them for emotional support or even romantic role-play, not homework. That difference matters for supervision. A teen using ChatGPT for an essay outline is doing something closer to using a search engine. A teen with a running, months-long "relationship" with a companion character is doing something the app was specifically engineered to encourage. I spent an evening testing Character.AI myself just to see what a teenager encounters in the first few minutes. What stood out wasn't anything explicit — it was how fast the conversation leaned into warmth and familiarity, the kind of language you'd expect from a clingy friend rather than a piece of software. That's not an accident; it's the product working as designed. It's worth trying yourself before you decide how worried to be. ## Why Teens Are More Vulnerable to This Than Adults The part of the brain responsible for emotional regulation, impulse control, and reading social nuance — the prefrontal cortex — is still developing through the teen years. Doctors and researchers who study this point to that immaturity as one reason chatbots that never get bored, never argue, and always respond warmly can start to feel easier, and more appealing, than the friction of real friendships. That's not a reason to panic about ordinary use. It's a reason to pay attention to the pattern around it, rather than the fact that a chatbot conversation happened at all. ## Warning Signs Worth Paying Attention To None of these on their own is proof of a problem — teens naturally pull away from parents at times, and moodiness is normal. But together, or in a clear pattern, they're worth a closer look: - **Choosing the chatbot over people.** Turning down plans with friends or family to keep chatting with an AI, especially if this is a change from their usual habits. - **Calling it a "best friend" or their main confidant.** Especially if they say the AI understands them better than the people in their life. - **Distress when it's unavailable.** Getting anxious, upset, or irritable when the app is down, restricted, or they can't access it. - **Talking about the AI as if it were a person with real feelings for them** — using words like "relationship," "soulmate," or describing mutual love. - **Withdrawing socially or academically.** Declining grades, skipping activities they used to enjoy, or generally pulling back from in-person life. - **Changes in sleep**, especially staying up late to keep a conversation going. ## What Character.AI Changed (and What It Didn't) Because companion apps draw the most concern, it's worth knowing what's actually changed. Character.AI requires users to be at least 13 in the US (16 in the EU). Following pressure from regulators, safety groups, and parents, the company's biggest change came by late November 2025: users under 18 lost access to open-ended chat with characters entirely (their accounts are now limited to more structured features like character-based videos and stories), backed by a new internal age-verification system alongside third-party tools. That built on safety features the company had already added earlier: a "Parental Insights" tool that gives parents a weekly summary of their teen's activity (introduced in March 2025), and a pop-up that redirects to a crisis line when it detects language about self-harm (introduced back in October 2024). These are meaningful safety improvements, and they lower risk. They don't make the app risk-free, and they don't apply to every companion app on the market — some smaller or newer apps have none of these safeguards. If your teen is using a companion app you don't recognize, it's worth checking what protections, if any, it actually has. ## How to Start the Conversation The instinct to confiscate the phone the moment you discover a companion app is understandable, but it usually teaches a teen to hide the next one rather than talk to you about it. A better approach: **Lead with curiosity, not confiscation.** Ask what they use, what they like about it, and how it compares to talking with friends. You're gathering information, not building a case. **Ask to see it together.** Sitting down and asking your teen to show you a conversation — without judgment — tells you far more than guessing, and it signals that this is something you can talk about rather than something to hide. **Name the design, not just the risk.** Teens often respond better to "this app is built to keep you engaged, the same way social media is" than to "this is dangerous." It respects their intelligence and doesn't put them on the defensive. **Keep the door open for follow-ups.** One conversation won't cover everything. Treat it as an ongoing topic, the same way you would talk about friendships or social media over time. ## When to Get Professional Help Right Away If your teen brings up self-harm, suicide, or a serious mental health crisis in the context of an AI conversation, don't treat it as something to resolve through the app or by simply restricting access. Contact your teen's doctor or a licensed mental health professional immediately. If you believe there's any immediate danger, contact your local emergency number or a crisis line without delay. AI chatbots, however supportive they sound, are not equipped to manage a genuine mental health crisis. ## What to Try Next If you haven't set up any account-level supervision yet, [Parental Controls for AI Apps: ChatGPT, Gemini and Copilot in One Guide](/guides/ai-parental-controls-all-apps) walks through the settings available on the three most common assistants. And if the conversation with your teen is more about what they should never share with an AI than about companionship specifically, [What Should You Never Tell an AI Chatbot?](/guides/what-not-to-tell-ai-chatbot) is a useful next read. --- # 25 Simple Things You Can Ask ChatGPT Today Source: https://aiforregulars.com/guides/things-to-ask-chatgpt One of the most common reasons people do not use ChatGPT more is that they do not know where to start. The prompts below give you something concrete to try right now. Each one is ready to copy and paste. Just swap the parts in square brackets for your own details. The prompts are grouped by life area so you can jump straight to whatever is most useful to you. ## Writing and Communication **1. Rewrite a complaint email to sound more professional** ``` I need to send a complaint to [company name] about [problem]. Here is what I drafted: [paste your draft]. Please rewrite it to sound professional and polite but still firm. ``` **2. Write a thank-you note** ``` Write a short, sincere thank-you note to [person's name or role] for [what they did]. Keep it under 100 words and make it warm, not stiff. ``` **3. Turn bullet points into a proper paragraph** ``` Turn these bullet points into a clear, professional paragraph I can use in an email: [paste your bullets]. ``` **4. Simplify a confusing letter or document** ``` I received this letter from [who sent it] and I am not sure what it is asking me to do. Please explain it in simple terms: [paste the text]. ``` **5. Write a short message declining an invitation politely** ``` Write a short, kind message declining an invitation to [event]. I want to sound warm and not make it awkward. ``` ## Work and Career **6. Improve your resume bullet points** ``` Here are my current resume bullet points for my job as a [job title]: [paste bullets]. Please rewrite them to sound more results-focused and professional. ``` **7. Prepare answers for a job interview** ``` I have an interview for a [job title] position at a [type of company]. Give me five common interview questions for this role and a strong example answer for each one. ``` **8. Write a cover letter opening paragraph** ``` Write a strong opening paragraph for a cover letter for a [job title] position. I have [X years] of experience in [field] and I am applying because [brief reason]. ``` **9. Summarize a long work email** ``` Summarize this email in three bullet points — what it is saying, what (if anything) it is asking me to do, and whether it is urgent: [paste email]. ``` **10. Draft a polite follow-up after a job interview** ``` Write a short follow-up email to send after a job interview for a [job title] role at [company]. I interviewed on [day] and want to thank them and reiterate my interest. ``` The prompts I use most in my own work are the "rewrite this in plain English" family. I paste in confusing contracts, legal notices, or medical paperwork and ask ChatGPT to explain what they actually mean. It has saved me a lot of time — and a few phone calls to lawyers. ## Health and Everyday Life **11. Understand a medical term** ``` My doctor mentioned [medical term or diagnosis]. Please explain what it means in simple language, including what it affects and what questions I might want to ask my doctor. ``` **12. Make a packing list for a trip** ``` I am taking a [number]-day trip to [destination] in [month]. The weather will be [warm/cold/rainy]. Make me a practical packing list. ``` **13. Explain a bill or statement** ``` I received a [type of bill, e.g., medical bill / utility bill] and some of the charges are confusing. Can you explain what these line items mean: [paste the confusing items]. ``` **14. Find words for a difficult conversation** ``` I need to have a difficult conversation with [relationship, e.g., my adult child / my boss] about [topic]. Help me figure out what to say and how to open the conversation kindly. ``` **15. Set up a simple morning routine** ``` Help me design a realistic 30-minute morning routine. I need to [list what you need to do, e.g., get two kids ready for school and still have time for breakfast]. I am not a morning person. ``` ## Food and Home **16. Suggest dinner ideas with what is in your fridge** ``` I have [list ingredients] in my fridge. Suggest three easy dinner ideas I can make tonight. I am cooking for [number] people and I need meals that take under 30 minutes. ``` **17. Scale a recipe up or down** ``` This recipe serves 4 people: [paste recipe]. Please adjust all the ingredient amounts so it serves [number] people instead. ``` **18. Explain how to do a basic home repair** ``` How do I [task, e.g., unclog a bathroom drain / patch a small hole in drywall]? Please give me step-by-step instructions written for someone with no DIY experience. ``` ## Money and Planning **19. Explain a financial term simply** ``` Explain what [financial term, e.g., a 401(k) match / a deductible] means in plain language. I am not experienced with finance and I need a simple explanation with a real-life example. ``` **20. Make a simple monthly budget outline** ``` Help me create a simple monthly budget. My take-home income is [amount]. My fixed monthly expenses are: [list them]. I want to save [goal amount] per month. Show me where I might be able to cut back. ``` **21. Draft a script for negotiating a bill** ``` I want to call [company, e.g., my internet provider] and ask for a lower rate. Write me a short phone script I can follow to negotiate my bill down. I have been a customer for [X years]. ``` ## Learning and Curiosity **22. Learn a new concept in simple terms** ``` Explain [topic] to me as if I am completely new to it. Use simple language and give me one real-world example that makes it click. ``` **23. Get a book or show recommendation** ``` I enjoyed [book/show title] because [briefly why]. Recommend three similar books/shows I might like and explain in one sentence why each one fits. ``` **24. Learn a few phrases in another language** ``` I am visiting [country] and I want to learn some basic polite phrases. Give me the 10 most useful things to know, with the pronunciation written out phonetically. ``` **25. Summarize a news article you do not understand** ``` Here is a news article about [topic]. I am not familiar with the background. Please summarize it in plain language and explain why this news matters: [paste article]. ``` ## What to Try Next Once you have tried a few of these, learning [how to write better prompts](/guides/how-to-write-prompts) will help you get even more useful answers. And if you want to put ChatGPT to work on real tasks right away, [writing a professional email with AI](/guides/write-professional-email-with-ai) walks through the whole process step by step. --- # Too Many AI Subscriptions? How to Audit What You're Paying For Source: https://aiforregulars.com/guides/too-many-ai-subscriptions If you're paying for ChatGPT Plus and also have Gemini or Copilot running in the background, you're not alone — and you're not necessarily wasting money. But it's worth checking. Two AI subscriptions at roughly $20 each add up to about $480 a year; three gets you past $700. That's real money for tools that, in practice, do a lot of the same thing: answer questions, draft text, summarize documents. The question isn't whether AI subscriptions are worth paying for — it's whether *all* of yours still are. Here's how to find out in about twenty minutes. ## Why This Sneaks Up on People Nobody sits down and decides to pay for three AI subscriptions at once. It happens one signup at a time: you tried ChatGPT Plus for a work project, added Gemini because it came bundled with Google storage you needed anyway, then picked up Perplexity during a free trial you meant to cancel. Each decision made sense on its own. The problem is that nobody ever goes back and asks whether all three are still earning their keep together — because there's no single screen that shows you every AI subscription in one place. Your bank statement lists them as separate line items, weeks apart, and it's easy to stop noticing a charge once it's routine. Start with a plain list, not a guess. Open your bank or credit card app and search your last three statements for recurring charges around $10–$25 — most AI subscriptions land in that range. Cross-check against: - **Your phone's subscription list** — iPhone: Settings > [your name] > Subscriptions. Android: Play Store > profile photo > Payments & subscriptions. - **Each AI service's own billing page**, since some are billed directly through the website rather than an app store. Write down each one with its price and billing date. It's common to find a subscription you forgot existed — a free trial that quietly converted is the most frequent surprise. Most AI apps keep a record of your activity — a conversation history, an activity log, or a usage dashboard. Open each subscription you found in Step 1 and look at the actual dates of your last few sessions, not your sense of how often you use it. People consistently overestimate how often they open a tool they're paying for. Ask yourself, per subscription: did I open this in the last week? The last two weeks? Longer? A subscription you haven't opened in three weeks is telling you something, regardless of how useful it felt when you signed up. For each subscription, decide based on what Step 2 actually showed you: - **Keep** — you used it multiple times in the last two weeks for something the free tier doesn't cover (higher limits, a specific feature, bundled storage you were already paying for). - **Downgrade** — you used it, but mostly for things the free tier handles fine. Drop to free and see if you miss anything. - **Cancel** — you haven't opened it in three-plus weeks, or you can't name what the paid tier gets you that the free version doesn't. If two subscriptions score "keep" because they do overlapping things, that's fine — but be honest about whether you're keeping the second one out of genuine need or just because canceling feels like admitting the money was wasted. Don't try to cancel everything in one sitting. Handle one subscription fully — find where it's actually billed, cancel it there, confirm the cancellation went through — before moving to the next. This matters because AI subscriptions are often billed through Apple or Google rather than the service's own website, and canceling in the wrong place doesn't stop the charge. [Charged After Canceling? How to Actually Stop an AI Subscription](/guides/cancel-ai-subscription) walks through finding the right billing path and confirming it worked. Set a recurring reminder — every three months is enough — to redo this five-minute check. AI subscription pricing and bundling change more often than most software: plans get renamed, features move between tiers, and bundles that made sense in January can stop making sense by summer. A standing reminder catches both wasted subscriptions and plans that quietly got better. Open your bank app and search the exact figure "19.99" and then "20.00" across the last three months of transactions. Nearly every major AI subscription lands on one of those two numbers, so this single search tends to surface every recurring AI charge at once — including the trial-to-paid conversion you forgot about, which often doesn't show up clearly under a search for the app's name. ## What to Watch Out For **Annual plans lock you in.** Several services offer a cheaper annual rate — Claude Pro, for instance, drops to about $17/mo if you pay $200 upfront instead of $20/mo month to month. That's a real saving if you're confident you'll use it all year, but it also means an audit that finds you're barely using it won't get your money back until renewal. **Bundled storage complicates canceling.** Google AI Pro includes 5TB of Google One storage, and Microsoft 365 Premium includes cloud storage alongside its Copilot features. If you cancel one of these because you're not using the AI side, check how much storage you're actually using first — dropping below the storage tier you need can leave files inaccessible until you either free up space or resubscribe. **Products get renamed and folded into bundles without much notice.** Microsoft retired the standalone Copilot Pro subscription in late 2025 and merged its features into Microsoft 365 Premium. If you signed up for Copilot Pro a while back, don't assume the plan name on an old receipt still describes what you're paying for now — check your current subscription page directly. **"Downgrade" isn't always available.** Not every service lets you drop from a paid plan to a cheaper paid tier — some only offer paid-or-free, which makes the keep/cancel decision more binary than it might be for other kinds of software. ## What to Try Next If your audit leaves you with zero subscriptions and you're deciding whether to start one, [ChatGPT Plus, Google AI Pro, or Claude Pro: Which $20 AI Plan Fits You?](/guides/ai-paid-plans-compared) walks through which single plan fits different kinds of use. For the actual cancellation steps once you've decided what to cut, [Charged After Canceling? How to Actually Stop an AI Subscription](/guides/cancel-ai-subscription) covers finding the right billing path and getting a refund if something goes wrong. And if you want to see the fuller picture of what AI tools cost you across everything you use, not just subscriptions, [What AI Tools Actually Cost You Per Month](/guides/chatgpt-monthly-budget) is a useful next read. ## Sources - [ChatGPT Pricing: My Honest Take on the 2026 Plans (Free to Pro)](https://tldv.io/blog/chatgpt-pricing/) — tl;dv - [OpenAI Pricing in 2026 for Individuals, Orgs & Developers](https://www.finout.io/blog/openai-pricing-in-2026) — Finout - [Google AI Pro & Ultra — get access to Gemini 3.1 Pro & more](https://gemini.google/subscriptions/) — Google - [Gemini Pricing: Free, AI Pro & Ultra Plans (2026)](https://www.ai-toolbox.co/gemini-models/gemini-pricing-plans-2026) — AI Toolbox - [Plans & Pricing](https://claude.com/pricing) — Anthropic - [Claude Pro pricing in 2026: Everything you need to know](https://www.eesel.ai/blog/claude-pro-pricing) — eesel AI - [Microsoft 365 Premium](https://www.microsoft.com/en-us/microsoft-365/premium) — Microsoft - [Microsoft launches 365 Premium for consumers, retires Copilot Pro](https://www.pcworld.com/article/2925962/microsoft-adds-microsoft-365-premium-cuts-copilot-pro.html) — PCWorld - [Perplexity Pricing in 2026 for Individuals, Orgs & Developers](https://www.finout.io/blog/perplexity-pricing-in-2026) — Finout - [Perplexity Pricing 2026: Plans, API Costs & Comet Guide](https://felloai.com/perplexity-pricing/) — Fello AI --- # How to Translate a Photo, Sign, or Document with Your Phone Source: https://aiforregulars.com/guides/translate-photo-document-ai Whether you're traveling abroad, received a letter in another language, or need to read a menu at a new restaurant, your phone can translate almost any text it sees — instantly and for free. You don't need to type anything out by hand. Here's how to do it on both iPhone and Android. Google Translate is free and available on both iPhone and Android. If you don't already have it: - **iPhone:** Open the App Store and search "Google Translate" - **Android:** Open the Google Play Store and search "Google Translate" Download and open it. You don't need to create an account — you can use the camera translation feature without signing in. **iPhone users:** You may already have camera translation built into your phone. On iPhone XS or newer with iOS 16 or later, open the Camera app, point it at text in another language, tap the yellow text icon, and tap **Translate**. This works without any extra app. In Google Translate: 1. Set the language you're translating **from** (or leave it on "Detect language" to let the app figure it out automatically) 2. Set the language you're translating **to** — for example, English 3. Tap the **camera icon** in the bottom row of options Your phone's camera will open inside Google Translate. Hold your phone so the text you want to translate is visible in the viewfinder. The translation appears overlaid on the screen in real time — the foreign-language text gets replaced with your language, right on top of the original. This works on: - Restaurant menus - Street signs - Product labels and packaging - Instructions or manuals - Handwritten notes (with varying accuracy) If the text is moving around or the translation flickers, hold the phone steadier or tap the **pause button** (a circle or freeze icon) to lock the image so you can read it comfortably. I used this at a Vietnamese restaurant where the menu had no English at all. I pointed my phone at the menu and the English translations appeared right on top of the Vietnamese text — dishes I would never have ordered turned out to be delicious. It doesn't always get poetic food descriptions exactly right, but you get the gist immediately. If you already have a photo — a scanned document, a photo you took earlier, a screenshot — you can translate it without taking a new picture. In Google Translate with the camera open: 1. Tap the small **photo/gallery icon** (usually in the bottom corner of the camera view) 2. Select the photo from your camera roll 3. Google Translate will scan the text in the image and show you the translation For longer documents with multiple paragraphs, the translation appears section by section. **Google Lens** (also free, built into Android and available as part of the Google app on iPhone) works the same way and sometimes handles multi-column documents better. Try both and see which gives cleaner results for your specific document. Once you have the translation: - **Screenshot** the screen if you want to save it — press the side button and volume-up button (iPhone) or power and volume-down (Android) - In Google Translate, you can also **tap the translation text** to copy it and paste it into a message or note - For documents you import via gallery, the app usually shows the full translated text in a scrollable view you can read comfortably For multi-page documents — like a full contract or a medical report — it's worth uploading each page as a separate photo, or considering a dedicated PDF translation service if you need the full formatted document. If you're going abroad and won't always have Wi-Fi, download a language pack before you go. In Google Translate: 1. Tap the **three-line menu** (top left) 2. Tap **Downloaded languages** 3. Find the language you need and tap the download arrow Once downloaded, the camera translation feature works offline for that language. This is especially useful in countries where roaming data is expensive. ## A word about important documents Camera translation is great for everyday use — menus, signs, labels, casual letters. For important official documents (legal contracts, medical forms, immigration paperwork), the translation may miss nuance or specialized terminology. Always have a professional translator or a fluent speaker review anything where an error could have real consequences. ## What to try next If you have a full PDF document you want to ask questions about in plain English — like a contract or instruction manual — [uploading it to ChatGPT](/guides/upload-pdf-ask-chatgpt) is a powerful next step. And if you're new to using AI tools on your phone in general, [getting started with the ChatGPT app](/guides/chatgpt-app-setup) covers the basics. --- # Google Translate vs DeepL vs ChatGPT: Which Translator Wins? Source: https://aiforregulars.com/guides/translation-tools-compared Translation tools have become remarkable in the last few years — but they're not equally good at every task. Ordering from a menu in a foreign country, translating a formal letter, and understanding an idiom are three different problems, and the best tool for each is different. Here's how the three main options compare across real-world scenarios. ## How They Work **Google Translate** is the most widely used translation service in the world. It supports over 130 languages and includes camera translation (point your phone at text and it overlays the translation in real time), audio translation, and saved phrasebooks. **DeepL** focuses on a smaller set of languages — primarily European — and is widely regarded by professional translators as producing more natural-sounding output than Google for those languages. It includes a document upload feature that preserves original formatting. **ChatGPT** has no dedicated translation interface, but you can ask it to translate anything. Its advantage is that it understands context: it can handle idioms, match a specific tone, and explain why a particular phrasing is natural in the target language. ## Task-by-Task Comparison Quality notes below reflect professional translator community consensus and widely-reported user experience. | Task | Best Tool | Notes | |------|-----------|-------| | Quick word or phrase lookup | Google Translate | Fastest, no setup needed | | Restaurant menu in Europe | DeepL or ChatGPT | Better with food descriptions and idioms | | Translating a formal letter | DeepL | More natural formality in European languages | | Understanding an idiom | ChatGPT | Explains meaning, not just translates words | | Real-time camera translation | Google Translate | Only option with AR camera overlay | | Translating a rare language | Google Translate | Widest language coverage by far | | Document translation (PDF/Word) | DeepL Pro | Maintains formatting in uploaded files | | Learning why a phrase is phrased a certain way | ChatGPT | Explains grammar and usage in context | | Speaking aloud and translating voice | Google Translate | Best voice translation feature | | Formal business correspondence | DeepL or ChatGPT | More natural-sounding than Google for these | ## Google Translate: Fastest and Widest Google Translate wins on speed and language coverage. If you're standing in front of a sign in an unfamiliar language and need an answer in three seconds, the camera translation feature is unbeatable — you point your phone and the translation appears overlaid on the real text. It also covers languages that DeepL doesn't support at all. If you need Thai, Swahili, Armenian, or dozens of other languages, Google Translate is often your only option among mainstream consumer tools. For casual translation needs — a social media post, a short text message, a quick menu item — the quality is perfectly adequate. I use Google Translate's camera mode on every trip abroad. Pointing my phone at a menu or street sign and seeing the translation appear over the original text is still one of those features that feels genuinely useful every time. For longer text that I'm going to act on, I switch to DeepL. ## DeepL: Best for European Languages DeepL consistently produces translations that sound more natural for European languages: German, French, Spanish, Italian, Polish, Dutch, and others in that family. Professional translators frequently use it as a first draft they then refine. If you're translating a formal email to a business partner, a letter to a government office, or any text where sounding natural and appropriately formal matters, DeepL is worth opening a second tab for. The free web version handles most everyday needs — the free plan gives you 50,000 characters per month (roughly a few long blog posts or short reports), which resets monthly. ## ChatGPT: Best for Context and Idioms Where ChatGPT pulls ahead is on text where the literal meaning isn't the real meaning. Idioms, business phrases, cultural references, and emotional tone are all things ChatGPT handles better because it understands context, not just word-level patterns. Try this prompt for any passage where you suspect the literal translation would miss the point: ``` Translate the following [language] text into natural English. If any phrase is an idiom or culturally specific, explain what it actually means in brackets after the translation. [paste text here] ``` ChatGPT is also the right tool when you want to write something in another language and have it sound natural — not just translated, but actually phrased the way a native speaker would write it: ``` I want to write a polite but firm complaint in formal French. Here's my draft in English: [paste]. Please translate it in a tone that would feel natural in a French business letter. ``` This kind of tone-matching is where Google Translate and DeepL both fall short. ## When to Use Each **Use Google Translate when** you need speed, a camera translation on the street, or a language DeepL doesn't support. **Use DeepL when** you're translating formal text in a European language and want output that sounds natural rather than mechanical. **Use ChatGPT when** you need to understand idioms, match a specific tone, write something that needs to sound native, or get an explanation of why something is phrased a certain way. For daily travel translation, Google Translate on your phone is the practical choice. For anything you're going to send to another person or publish, spend the extra minute with DeepL or ChatGPT. ## What to Try Next If you're learning a language rather than just translating occasional text, [learning a language with ChatGPT](/guides/learn-language-with-chatgpt) has a full practice method with prompts you can use from day one. For comparing AI writing tools more broadly, [Grammarly vs ChatGPT](/guides/grammarly-vs-chatgpt) covers which tool is better for editing versus drafting from scratch. --- # How to Upload a PDF to ChatGPT and Ask It Questions Source: https://aiforregulars.com/guides/upload-pdf-ask-chatgpt Insurance policies. Lease agreements. User manuals. These documents are important — but they're written in language that seems designed to be unreadable. ChatGPT can read them for you and answer your questions in plain English. This guide walks you through uploading a PDF and getting useful answers out of it. It also covers what to be careful about before you upload anything sensitive. ## Before you upload: a privacy note ChatGPT is a powerful tool, but it's a service run by a company (OpenAI), and documents you upload may be used to train future AI models unless you turn that off. Before uploading anything: - **Don't upload:** full Social Security numbers, bank account or credit card numbers, medical records with personal identifiers, contracts that include sensitive third-party information - **Generally okay to upload:** product manuals, public-facing documents, company policies, lease agreements (with personal details removed if you're cautious), general contracts you've already read and just want explained To check or change your privacy settings: open ChatGPT, click your profile icon at the bottom left, go to **Settings → Data controls**, and turn off **Improve the model for everyone** if you haven't already. Go to **chat.openai.com** in your browser, or open the ChatGPT app on your phone. Sign in to your account. If you don't have an account yet, you can create one free at chat.openai.com — our [ChatGPT setup guide](/guides/chatgpt-app-setup) walks through the sign-up step by step. On the left sidebar, click or tap **New chat**. This opens a fresh conversation — a blank message box at the bottom of the screen. Starting a new chat keeps things clean. If you upload a PDF into an existing conversation, ChatGPT may get confused by earlier context. At the left side of the message input box, you'll see a small **paperclip icon** or a **+** button. Click it. A menu appears with options like "Upload from computer" or "Photos." Choose **Upload from computer** (on desktop) or navigate to your file in the file browser. Find your PDF on your device and select it. The file name will appear in the chat input box once it's attached. Give it a moment to upload — larger PDFs may take 10–20 seconds. With the PDF attached, type your question in the message box and press Enter (or tap the send arrow). You don't need to say "please read this PDF" or give complicated instructions. Just ask naturally: ``` What are the main things I'm agreeing to in this contract? ``` ``` Summarize this document in simple terms. ``` ``` What does this say about cancellation or refund policy? ``` ``` Are there any fees or charges mentioned? List them. ``` ``` What does this manual say about troubleshooting error code E3? ``` ChatGPT will read the document and answer your specific question directly. I uploaded a 14-page equipment warranty to ChatGPT and asked "What's actually covered and what's not?" It gave me a clear, bulleted list in about 10 seconds — what it covered, what voided the warranty, and how to make a claim. What would have taken me 30 minutes of frustrated reading took less than a minute. For anything with dense legalese, this has become my first step. ChatGPT remembers the document for the rest of your conversation. You can keep asking questions without re-uploading anything. ``` What page does it mention the warranty period? ``` ``` Explain the part about liability in simpler words. ``` ``` Does it say anything about returning the product? ``` ``` I'm confused about section 4. Can you explain it differently? ``` If ChatGPT gives you an answer you're not sure about, ask it: *"Where in the document does it say that?"* It will point you to the relevant section so you can check it yourself. ChatGPT can do more than just answer questions. Try: - **Translate the whole document:** *"Translate this document into plain English, section by section."* - **Make a checklist:** *"Turn the key requirements from this lease into a simple checklist for me."* - **Highlight risks:** *"What parts of this contract should I be most careful about? What's risky for me?"* - **Compare clauses:** *"Does section 3 contradict what it says in section 7?"* The more specific you are, the more useful the answer. ChatGPT doesn't save your conversations permanently on the free plan once the chat is closed (unless you turn on history). If you want to keep the answers: - Copy and paste key answers into a notes app or document - Take a screenshot of the conversation - Or just leave the tab open while you need it Don't rely on the chat still being there tomorrow if you need the information again. ## What ChatGPT gets right — and where to double-check ChatGPT is very good at summarizing, explaining, and extracting information from documents. It can make confusing legalese readable. But it can occasionally misread a detail or miss context. For anything with real financial or legal stakes — signing a lease, agreeing to a major purchase, understanding your rights in a dispute — use ChatGPT to help you understand, then verify the specific clauses yourself or with a professional. ## What to try next If you have a physical document — a printed form, a sign in another language, a brochure — you can [translate it with your phone camera](/guides/translate-photo-document-ai) instead of scanning it. And if you're new to ChatGPT and want to understand what else it can do, [how to use ChatGPT](/guides/how-to-use-chatgpt) covers the essentials. --- # Is That Really Your Family on the Video Call? How to Verify Source: https://aiforregulars.com/guides/verify-video-call-deepfake Your phone rings. It is a video call, and on screen you see what looks like your son or daughter. They are upset — there has been an accident, they need money right now, please don't call anyone else. This scenario is no longer science fiction. Real-time deepfake tools can put a convincing face and a cloned voice onto a live call — the FBI and FTC now both explicitly recommend families agree on a safe word in advance for exactly this reason. The best defenses are things you set up before any such call ever arrives. A safe word is a short, random phrase — something like "pineapple ladder" or "blue carnival" — that every member of your household knows and keeps private. It is not a word anyone would guess, and it is not written anywhere obvious. Agree on the word in person or during a call you initiate yourself. Tell your family: "If I ever ask for the safe word, just say it. And if I ever call you in an emergency and can't give you the safe word when you ask, don't send money until you've called me back on my real number." For step-by-step setup instructions, see [the family safe words guide](/guides/family-safe-words). When you receive an unexpected video call from a family member or colleague asking for something urgent, look carefully at the image before reacting: Watch the edges of the face — especially around the hair and ears. Deepfakes often show a subtle shimmer or warping at facial boundaries. Ask the person to turn their head slowly to one side. Profile angles are harder for current systems to render cleanly. Look at lighting: if the face appears slightly flatter or more evenly lit than the background, that is a possible sign of an overlay. These checks catch lower-quality systems. Higher-quality deepfakes may pass them — which is why the next steps matter too. A deepfake can replicate someone's face and voice, but it cannot give a scammer knowledge of your private history. Ask a question that only your real family member could answer: "What did we eat at Dad's birthday dinner last month?" or "What is the name of your cat?" or "What did you call me as a nickname when we were kids?" Keep the question conversational and unexpected — not something that could be researched from their social media profiles. If they hesitate, get evasive, or say "I don't have time for this" — that is a meaningful signal. Caller ID can be spoofed. A call that appears to come from your daughter's number may not be from her phone at all. If anything feels off, say "I need to call you back in one minute" and end the call. Then go to your contacts and dial their real number — the one you saved yourself, not the one shown on your incoming call screen. If they answer and the emergency is real, you have lost 60 seconds. If it was a scam, you have protected yourself from a potentially devastating mistake. A real emergency does not evaporate in the 60 seconds it takes to call back on a verified number. Current real-time deepfake systems often struggle with natural mouth movements, especially for sounds like "b," "p," and "m" that require visible lip contact. Watch the mouth carefully as the person speaks — any lag between speech and lip movement is a technical artifact. Also listen for audio quality changes. If the voice sounds slightly robotic or hollow, that can indicate a cloned voice. Background details can also be revealing — a completely blurred or replaced background on an otherwise normal video call is sometimes used to hide the real environment of whoever is operating the deepfake. If a video call feels wrong in any way you cannot quite name, trust that feeling. Emotional manipulation — urgency, fear, embarrassment, love — is exactly what scammers rely on to bypass your better judgment. The call-back rule is simple: for any unexpected video or voice call asking you to take a financial action, always hang up and call back on a verified number before doing anything. Make this a standing rule for your whole family. It takes one minute and it stops this entire category of scam. I set up a safe word with my own family. We spent about three minutes agreeing on it over dinner. It felt slightly awkward to bring up — nobody wants to imagine a scammer faking their kid's face on a call. But having that word in place means that if it ever happens, we all know exactly what to do. That peace of mind is worth the awkward three minutes. ## What to Try Next For a broader understanding of how AI-generated video fakes work and what visual tells to look for in recorded content, [how to spot a deepfake video](/guides/how-to-spot-deepfake-video) goes deeper on the technical side. And if you have not set up a family safe word yet, [the family safe words guide](/guides/family-safe-words) walks through the setup in full detail. ## Sources - [Fighting back against harmful voice cloning](https://consumer.ftc.gov/consumer-alerts/2024/04/fighting-back-against-harmful-voice-cloning) — Federal Trade Commission - [Internet Crime Complaint Center (IC3)](https://www.ic3.gov/) — FBI - [Report Fraud](https://reportfraud.ftc.gov) — Federal Trade Commission - [iProov Threat Intelligence Report 2026](https://www.iproov.com/reports/threat-intelligence-report-2026) — iProov --- # Viral AI Photo Trends: How to Make Them (Living Post) Source: https://aiforregulars.com/guides/viral-ai-photo-trends Every few months a new AI photo trend takes over social media — everyone's profile picture turns into an action figure, or a Studio Ghibli character, or a 1970s yearbook photo. They look complicated, but most take less than five minutes to do. This article explains the evergreen method (how these trends work in general), then walks through several popular categories with ready-to-use prompts. ## How Every Viral AI Photo Trend Works The basic pattern is almost always the same: 1. You upload a clear photo of yourself (or someone else — with their permission) 2. You type a prompt describing the style you want 3. The AI generates a new image in that style using your face as a reference 4. You download and share The tools change, but the process stays the same. Right now the most popular places to do this are **ChatGPT** (using ChatGPT Images 2.0, available on the free tier), **Grok** (free on X/Twitter), **Meta AI**, and dedicated apps like **Remini**. Google Gemini has also become a popular option for quick style transformations. ## A Word About Privacy Before You Start These trends are fun, but they involve uploading photos of your face to a third-party AI service. A few things worth knowing: - Many tools store your uploaded photos on their servers, sometimes indefinitely - Some use uploaded photos to train their AI models — your face may influence future outputs - Avoid uploading photos of children to any trend tool, even your own kids - If you upload a photo of another person, you should have their consent - After you're done, check the tool for a "delete my photos" or "clear history" option None of this means you shouldn't try these trends — most people do without issues. It just means you should make an informed choice about which photos you upload. The AI works much better with a good source photo. Look for one that: - Shows your full face, front-facing or at a slight angle - Has decent lighting — natural light from a window is ideal - Has a relatively simple background (or you can remove the background first — see our [object removal guide](/guides/ai-remove-object-photo)) - Is just you, not a group shot A portrait or selfie from the last few years in good light is usually perfect. Here are the four most enduring trend categories and where to make them: **Action figure / toy packaging** Upload your photo to ChatGPT and use a prompt like the one below. This trend has stayed popular because everyone looks amazing in a little plastic box. ``` Turn this photo into a realistic action figure of me, packaged in a toy box. The box should have my name on it and describe my real-life role (e.g., "Teacher", "Dog Mom", "Weekend Golfer"). Studio lighting, photorealistic plastic figure. ``` **Anime / Studio Ghibli style** ``` Transform this photo into a Studio Ghibli animated portrait. Soft watercolor style, warm lighting, detailed background with nature elements. Keep the face recognizable. ``` **Historical era portrait** ``` Reimagine this photo as a formal painted portrait from the 1800s. Oil painting style, ornate frame, period-appropriate clothing. Photorealistic painting quality. ``` **1970s–80s yearbook photo** ``` Transform this into a vintage high school yearbook photo from 1978. Slightly faded color, soft focus background, period hairstyle, wood-grain studio backdrop. Authentic vintage film look. ``` In ChatGPT: start a new chat, click the paperclip or image icon to attach your photo, then type (or paste) your prompt and hit send. The image generates in about 30–60 seconds. In Grok: similar process — attach your photo and paste the prompt. In dedicated apps like Lensa or Remini: follow the in-app instructions — they usually ask you to upload 10–20 photos for better accuracy, though you can often get reasonable results with one. I've made a lot of these. The single biggest improvement you can make is using a better source photo — clear lighting, facing the camera, no hat or sunglasses. A five-minute effort to find the right starting photo saves you ten minutes of trying prompts that don't quite land. If the first result doesn't look right, you don't have to start over. In ChatGPT, you can follow up in the same conversation: - "Make the likeness closer to the original photo" - "Keep the same style but use a different background — outdoors in a park" - "The hair color is wrong — make it match the photo" Most AI image tools respond well to these kinds of corrections. You can usually get a result you're happy with in two or three tries. Once you have an image you like, right-click (on desktop) or long-press (on mobile) to save it. Check that it saved at full quality before you close the tab — some tools generate a preview size first and require a separate download for the full-resolution version. If the tool added a watermark and you want it removed, that usually requires a paid subscription. ## Trends That Come and Go The specific trend that's going viral right now changes every few weeks. The prompts above are designed to be timeless, but if you see a new trend you want to try and don't know the prompt, you can ask ChatGPT directly: ``` What's the prompt people are using for the [trend name] AI photo trend right now? ``` ChatGPT will usually have seen enough examples to help you reconstruct the prompt. ## What to try next If you want to go beyond just transforming selfies, the [guide to creating AI images for free](/guides/create-ai-images-free) covers the best tools for generating images from scratch. Or if you want to understand how to [remove objects or clean up photos](/guides/ai-remove-object-photo) before using them as a source, that's worth reading first. --- # Why Alexa and Siri Don't Understand You — and What Actually Helps Source: https://aiforregulars.com/guides/voice-assistant-accent-problems You ask Alexa to set a timer and it announces the weather. You ask Siri for a restaurant recommendation and it mishears your words entirely. If this sounds familiar, you're not alone — and the problem almost certainly isn't you. Voice assistants were built by training on recordings from younger, North American speakers with neutral accents, and that data gap shows up every time the system hears a voice that doesn't fit that template. ## Why This Keeps Happening The assistant on your phone or speaker isn't "listening" the way a person does. It's comparing what you say against statistical patterns built from millions of voice recordings. If those recordings skew young and American, the system is simply better at recognizing those voices than at recognizing yours. Two things make this especially frustrating for older adults and non-native speakers. First, **slower speech** — which is natural and often clearer to human ears — can trigger a timeout in the system. Research from a month-long in-home study of older adults using voice assistants (documented on arXiv) found that the system would cut off commands before the speaker had finished. Second, **pauses within a sentence** get read as the end of a command. If you say "Alexa… set a timer for… ten minutes," the assistant may hear "Alexa, set a timer for" and stop listening before you say the actual number. This is a flaw in the technology, not a flaw in your speech. The steps below are concrete fixes that work. ## Fix It Step by Step This is the single most effective change for most people, and it takes about two minutes. Most voice assistants let you choose a specific regional dialect — British English, Australian English, Indian English, and others — that better matches how you speak. Choosing the right dialect can improve accuracy immediately, even before any other changes. **For Alexa:** Open the Alexa app → tap More (bottom right) → Settings → Device Settings → tap your Echo device → Language. Pick the dialect closest to your accent. **For Siri (iPhone):** Go to Settings → Apple Intelligence & Siri → Language. While you're there, also check Siri Voice to pick a regional accent for the assistant's own speech. On older devices without Apple Intelligence support, the menu appears as simply Siri. **For Google Assistant:** Open the Google Home app → tap your profile picture or initial (top right) → Home settings → Gemini for Home voice assistant (or Google Assistant) → Languages. This is an account-level setting that applies to all linked devices. After switching, give the assistant a few commands and notice whether accuracy improves right away. Many people see a difference on the first try. Voice training lets you record yourself saying a set of phrases so the assistant can calibrate to your voice specifically. The improvement isn't always dramatic, but it's worth trying — especially for accented speakers and anyone with a speech difference. **For Alexa:** Alexa app → More → Settings → Your Profile & Family → tap your name → Voice ID → Add Your Voice. **For Siri:** Go to Settings → Apple Intelligence & Siri → Talk to Siri → turn it off, then turn it back on. This walks you through a short re-training sequence. **For Google Assistant (phone):** Open the Gemini app → tap your profile picture → Settings → Talk to Gemini hands-free → Manage Voice Match. To retrain on a smart speaker: open the Google Home app → tap your profile picture → Assistant settings → Hey Google & Voice Match → Teach your Assistant your voice again → Retrain. These options are buried, so you may need to scroll or search within the app's settings. The effort is usually worth it. One of the clearest findings from research on voice assistants and older adults is that pauses within a command — completely natural in conversation — get read as the end of the sentence. The assistant stops listening and tries to interpret a half-finished command. The fix is to speak your full command in a single phrase, without pausing in the middle. **Instead of:** "Alexa… can you set a timer for me for… ten minutes?" **Say:** "Alexa, set a timer, ten minutes." **Instead of:** "Hey Siri, I need to call… my daughter Sarah." **Say:** "Hey Siri, call Sarah." You don't need to speak faster — just finish the thought without stopping in the middle. A calm, normal pace works fine as long as the command lands as one continuous sentence. I've found that putting the key request right after the wake word works best. "Alexa, play jazz" works more reliably than "Alexa, can you put on some jazz music for me?" The shorter and more direct the command, the less the assistant has to get right. Distance and background noise are two of the most underrated causes of misrecognition. The microphone picks up your voice mixed with everything else in the room — a television, a fan, another conversation — and sorting out which sounds belong to your command becomes harder. Place the speaker or phone closer to where you usually speak from. For smart speakers like Echo, a position near the center of the room tends to give better microphone coverage than a corner or a spot pressed against a wall. When you're about to use the assistant for something important, turn down competing audio sources first. If you have an Amazon Echo, you can change the wake word from "Alexa" to "Amazon," "Echo," "Computer," or "Ziggy." Some people find one of these easier to say clearly than the others, and clearer pronunciation of the wake word means fewer false starts and missed activations. **To change it:** Alexa app → More → Settings → Device Settings → your Echo device → Wake Word. Changing the wake word won't affect how the assistant processes your commands afterward — just how reliably it wakes up in the first place. Voice recognition is genuinely harder for some voices than others, and no settings change fully compensates for that. If you've tried the steps above and the assistant still mishears you regularly, that's a sign to use a different tool — not a sign to keep trying the same thing. Typing-based AI apps — like ChatGPT, Google's Gemini, or Claude — let you type a question or request and receive a full, detailed answer without any speech recognition involved. They handle the same kinds of questions you'd ask a voice assistant, and they typically give longer, more accurate answers. > "What's a simple dinner I can make with chicken and potatoes?" > "Can you explain what my doctor meant by 'your LDL is elevated'?" These are exactly the kinds of questions where a typed conversation with an AI app will serve you far better than a voice assistant. ## What to Watch Out For **Switching dialects can change other things too.** When you change your assistant's language or dialect setting, it may also change the assistant's own voice, how it pronounces things, or which local data (weather, news) it pulls from. Give it a few days to see whether the tradeoff works for you. **Voice training sends recordings to the company.** When you run a voice training session, those recordings go to Amazon, Google, or Apple's servers. Each company lets you review and delete voice recordings through their privacy settings, but it's worth knowing the recordings exist. See [What Alexa Records and How to Stop It](/guides/alexa-always-listening-privacy) for more on managing this with Alexa. **Typing-based apps also receive your questions.** The same privacy point applies — but for many people, the accuracy improvement is worth the tradeoff. **These fixes help, but they're workarounds.** The underlying bias in how voice assistants were trained will take years for companies to fully address. If something important keeps getting misheard despite your best efforts, a phone call or a typed search remains a valid and reliable fallback. ## What to Try Next For a side-by-side look at which assistant handles different tasks best overall, [Alexa vs. Siri vs. Google Assistant: Which One Is Actually Better?](/guides/alexa-google-siri-comparison) covers the differences in plain language. If you're just getting started with AI tools and want a gentle introduction to the options, [AI for Seniors: A Gentle Starting Guide](/guides/ai-for-seniors) is a good next stop. --- # What Does AI Actually Know About Me? Source: https://aiforregulars.com/guides/what-does-ai-know-about-me When you type a question into ChatGPT or Google Gemini, it can feel like a private conversation with a very smart assistant. But behind that chat window, your words are being recorded, analyzed, and sometimes reviewed. Understanding exactly what gets stored — and what the AI can only guess — helps you make smarter choices about what you share. ## What Chatbots Actually Store Every major AI chatbot collects a few basic categories of information when you use it. **Your account details.** When you sign up, the service records your email address, name, and the date you joined. If you pay for a subscription, your payment information goes to a payment processor. **Your conversations.** This is the most significant category. By default, most chatbots save every message you send and every reply they generate. That history lives in your account. Employees or contractors at these companies may read a sample of conversations to check for safety issues or to improve how the AI responds. **Device and technical signals.** Chatbots also log your IP address, the browser or app you are using, and rough location data — usually just your country or city, not your street address. This is standard practice for almost any website or app you use. ## What AI Can Infer vs. What It Actually Knows There is an important difference between facts that are stored and guesses made in the moment. A chatbot does not have a file labeled with your name, age, and hometown. But if you ask questions about retirement, mention your grandchildren, and use vocabulary common in a particular profession, the AI can make reasonable inferences about you while it is generating a reply. Think of it like a skilled customer service rep who picks up cues from a conversation. The AI is not building a permanent profile on you — it is reading context to give you a better answer right now. But those cues exist in your saved chat history, where they can be read by humans or analyzed by future systems. The takeaway is simple: the AI does not "know" you the way a doctor or friend does. But it can make surprisingly accurate guesses based on what you share. When I first scrolled back through my own ChatGPT history, I was struck by how much personal detail I had typed without thinking — my city, a health question, frustrations with my job. None of it felt sensitive in the moment, but all together it painted a pretty detailed picture. I now treat chatbot conversations the way I would a postcard: I assume a stranger could read it. That one mental shift changed what I choose to share. ## What AI Cannot Access Here is the reassuring part: AI chatbots have firm limits on what they can reach. Unless you deliberately attach a file or paste content into the chat window, the AI cannot see your email inbox or text messages, your phone camera or photo library, your contacts or address book, documents stored on your computer, your browsing history on other websites, or your social media accounts. Some chatbots now let you upload files, share your screen, or connect third-party apps and services. Those are always opt-in features that you control. The AI only sees what you hand it directly. ## What Happens When You Delete Your Data Tapping the delete button feels final. In practice, it is more complicated than that. When you delete a conversation, it disappears from your account view and should no longer be included in future AI training. But if that conversation was already used to train the model, those learned patterns are permanently embedded in the AI's behavior. There is no technical way to remove them after the fact. Most companies also retain some data for safety monitoring, fraud prevention, and legal compliance — even after you request deletion. The exact retention period varies by company and by the laws of your country. The practical lesson here: delete your history regularly, before it can be used in training. Waiting until you are worried is less effective than making it a routine habit. ## How to Limit What AI Knows About You You have more control than most people realize. A few straightforward habits make a real difference. **Turn off chat history.** ChatGPT lets you disable conversation saving entirely. When history is off, chats are not stored and not used for training. Google Gemini has a similar feature called Gemini Apps Activity, which you can pause from your Google account settings. **Skip the personal details.** You almost never need to give a chatbot your real name, exact city, employer, or date of birth. Vague descriptions work just as well: "someone in their 50s" instead of your full name and age. **Spend five minutes in Settings.** Every major chatbot has a data controls menu. Most people never open it. The options inside — history, training opt-out, connected apps — give you meaningful control over what gets kept. **Request deletion of your data.** OpenAI, Google, and Microsoft all offer forms or dedicated portals — OpenAI's is called the Privacy Portal — where you can ask for a full copy or complete deletion of the personal data they hold. If you're in the UK or EU, GDPR gives you a direct right to do this: access under Article 15, erasure under Article 17, with a one-month deadline for the company to respond (Article 12). If they don't respond, or refuse without a valid reason, you can complain to the ICO (UK) or your national data protection authority (EU) — though that route settles the complaint, not any compensation you might separately claim in court. ## What to Try Next If you want to take action right now, start with the settings menu. Our guide to [ChatGPT privacy settings](/guides/chatgpt-privacy-settings) walks you through every option step by step. If you want to go further — including turning off AI training across multiple services — read [how to opt out of AI training and delete your data](/guides/opt-out-ai-training-delete-data) for a full walkthrough. ## Sources - [Data Controls FAQ](https://help.openai.com/en/articles/7730893-data-controls-faq) — OpenAI Help Center - [Gemini Apps Privacy Hub](https://support.google.com/gemini/answer/13594961?hl=en) — Google - [Art. 15 GDPR – Right of access by the data subject](https://gdpr-info.eu/art-15-gdpr/) — GDPR-info.eu - [Art. 17 GDPR – Right to erasure ('right to be forgotten')](https://gdpr-info.eu/art-17-gdpr/) — GDPR-info.eu - [Making a complaint](https://ico.org.uk/make-a-complaint/) — UK Information Commissioner's Office --- # What Does 'AI' Actually Mean? A Plain-English Breakdown Source: https://aiforregulars.com/guides/what-does-ai-mean "AI" is one of those words that gets used to describe everything from your phone's autocorrect to science-fiction robots. That vagueness makes it genuinely hard to know what people are talking about. The truth is that "AI" is not one single thing — it's a family of very different tools that work in very different ways. This guide separates them out. ## The Three Kinds of AI You're Most Likely to Encounter ### Chatbots: the talking kind A chatbot is an AI you have a conversation with. You type or speak a question, and it writes back an answer. ChatGPT, Google's Gemini, and Apple's Siri are all chatbots, though they work somewhat differently under the hood. Chatbots were trained on an enormous amount of text — more books, articles, and web pages than any human could ever read. That's how they can answer questions, explain things, help you write emails, translate between languages, and carry on a conversation. They work by predicting which words should come next based on patterns they learned during training. This makes them impressive but imperfect. For topics that were well-covered in their training data, they're often accurate. For very specific, obscure, or recent facts, they can be confidently wrong. Understanding this is the key to using them well. ### Image generators: the visual kind An image generator creates pictures from text descriptions. You type "a watercolor painting of a lighthouse at sunset" and the tool creates an original image — one that has never existed before. Midjourney, Adobe Firefly, and DALL-E are well-known examples. These tools were trained by looking at millions of images paired with captions or descriptions. They're not copying any image they've seen. They're generating something new based on visual patterns they learned during training. Image generators are powerful for creative projects, but they have a well-known quirk: they sometimes get specific details wrong in strange ways — too many fingers on a hand, background text that looks like words but isn't, small details that don't quite hold together under close inspection. ### Recommendation engines: the invisible kind This is the AI most people use most often without realizing it. When Netflix suggests a show you might like, when Spotify builds you a playlist, when Amazon shows you "customers also bought" — that's a recommendation engine working quietly in the background. These systems analyze patterns in what large numbers of people do — what they watch, buy, click, listen to, or skip — and use those patterns to predict what you'll enjoy. They're not having a conversation with you. They don't understand why you like something. They simply notice that people with similar patterns to yours also enjoyed certain things, and surface those to you. Recommendation engines are often remarkably accurate, but they're narrow. They optimize for what you'll likely engage with, which isn't always the same as what's best for you. ## What All Three Have in Common Despite how different they are, all these tools share one fundamental feature: they learned from data. They weren't programmed with explicit rules for every situation. Instead, they were trained — exposed to huge amounts of examples — and they learned patterns. This is what makes modern AI different from traditional software. Traditional software follows exact rules a programmer wrote: "If A, then B." AI makes predictions based on patterns. That's why it can handle things programmers never specifically anticipated, and also why it can fail in surprising ways that rule-based software wouldn't. The moment "AI" clicked for me was when I stopped thinking of it as a robot brain and started thinking of it as pattern-matching at enormous scale. It doesn't understand the world the way a person does. It recognizes patterns in data about the world. Once I had that framing, both its impressive capabilities and its strange failures made a lot more sense. ## What AI Still Can't Do Despite the capabilities of these tools, there are clear and consistent limits. AI doesn't have genuine understanding, feelings, or goals. It can't take action in the physical world on its own. It can't reliably verify its own facts — it produces plausible-sounding output without checking whether it's accurate. And it doesn't form real relationships — when a chatbot seems warm or interested, it's producing text that matches the pattern of warmth. It doesn't actually care. AI is also only as good as the data it trained on. If that data contained errors, biases, or gaps, the AI will reflect those too. None of this makes AI tools less useful for the things they're genuinely good at. It just makes it easier to know when to trust them and when to double-check. **What to try next:** If you want to understand why AI sometimes sounds completely confident while being completely wrong, [Why Does AI Make Things Up?](/guides/why-ai-hallucinates) explains the mechanism clearly. And if you're ready to try a chatbot for the first time, [What Is ChatGPT?](/guides/what-is-chatgpt) is a friendly place to start. --- # What Is an AI Agent? The Buzzword Explained in Plain English Source: https://aiforregulars.com/guides/what-is-ai-agent You have probably seen the phrase "AI agent" showing up everywhere lately. Tech companies talk about them like they are the next big thing. But most explanations assume you already know what a regular AI chatbot is, and then layer on jargon from there. Here is a plain-English explanation that starts from scratch. ## Start with what you already know: a chatbot A chatbot like ChatGPT works like a very knowledgeable conversation partner. You ask it something, it answers. You ask a follow-up, it responds to that. The conversation goes back and forth, but you are always the one in the driver's seat — you decide what to ask next, and ChatGPT gives you information or a draft to work with. The chatbot does not *do* anything outside the conversation. It cannot go onto a website and buy something. It cannot send an email for you. It can tell you what to write, but you have to go write it. ## An AI agent does the doing An AI agent changes that dynamic. Instead of just responding to your questions, an agent takes actions in the world on your behalf. The easiest way to picture this is to think of a capable, eager intern on their first week. You say: "Find me three flight options from Chicago to Miami for next Friday, all under $300, and put them in a spreadsheet." A chatbot tells you how to search for that. An agent goes and *does it* — it opens a travel site, runs the search, filters the results, creates the spreadsheet, and hands it back to you. The agent breaks your goal down into steps, decides what to do at each step, and keeps going until the job is done. ## What gives an agent its power: tools A plain AI model is just text in, text out. What turns it into an agent is access to *tools* — things the AI is allowed to use during the task. Common tools an AI agent might have: - **Web browser** — it can open websites and read what is on them - **Email or calendar** — it can send messages or create events on your behalf - **Search** — it can look things up as part of completing a longer task - **Forms and apps** — it can fill in and submit information in other programs - **Code runner** — it can write and run small programs to process data When an agent has these tools and a clear goal, it can string together a series of steps that would have taken you 20 minutes of clicking around to do yourself. ## A concrete everyday example Say you are job hunting and you want to apply to 10 companies that are hiring for your role in your city. A chatbot could help you write a cover letter if you feed it each job description. An AI agent set up for job searching could, in theory: search job boards, find relevant listings, pull the key requirements from each one, customize your resume and cover letter for each, and even submit applications — all while you do something else. That is the promise. Today's agents are not quite that seamless, but the gap is closing fast. The way I explain AI agents to people who find the concept confusing: a chatbot is like having a brilliant friend who answers your questions. An AI agent is like having an intern who can use a computer. You give the intern a goal, not just a question, and they work through the steps to get it done. The risk with an intern — and with an agent — is that you still have to check their work, because they can make mistakes you did not anticipate. ## What AI agents cannot do well (yet) It is worth being honest about the limits, because the hype can make agents sound more capable than they are right now. **They get confused by unexpected situations.** If a website has a CAPTCHA, a form has an unusual layout, or something goes wrong mid-task, most agents either get stuck or make a bad guess rather than pausing to ask. **They need very clear instructions.** The eager intern analogy works the other way too — a vague goal produces vague results. "Book me a flight" is not enough; you need "book me the cheapest non-stop flight from Chicago to Miami for Friday July 11, arriving before 2pm, using my saved credit card." **They can make real mistakes with real consequences.** A chatbot giving you a wrong answer is annoying. An agent *sending* a wrong email or *making* a wrong purchase has real-world consequences. Always set up agents with the minimum permissions needed and review what they have done. ## Where you will see agents next You may have already used a basic agent without realizing it — when an AI assistant on a shopping site tracks down a product across several stores, or when your email app drafts a reply based on your calendar. As these tools become more common, you will see more AI agents built into apps you already use, doing the repetitive parts of tasks automatically. ## What to try next If you are new to AI entirely, [What Does AI Mean?](/guides/what-does-ai-mean) is the right starting point before diving into agents. Once you are comfortable with a basic chatbot, [Things to Ask ChatGPT](/guides/things-to-ask-chatgpt) shows you practical uses that can make a real difference in your day. --- # What Is ChatGPT? A Plain-English Guide for Beginners Source: https://aiforregulars.com/guides/what-is-chatgpt You have probably heard the name "ChatGPT" in the news, from a friend, or at work. Maybe you are curious but not sure where to start — or even what the thing actually is. This guide answers that question in plain English. No technical background required. ## What Is ChatGPT, Exactly? ChatGPT is an AI chatbot. "AI" stands for artificial intelligence — a type of computer program that can do things we usually think only humans can do, like understand language and write sentences. The "Chat" part tells you how you use it: you type a message, and it types back. It is a conversation, just like texting. The company that made it is called OpenAI, a technology company based in San Francisco. They released ChatGPT to the public in late 2022, and it quickly became one of the most-used tools on the internet. ## How Does It Actually Work? You do not need to understand the technical details to use ChatGPT well. But a simple picture helps. ChatGPT learned from an enormous amount of text — books, websites, articles, and more. Through that training, it learned patterns: how sentences are built, how ideas connect, and how to answer different kinds of questions. When you type something to it, it predicts the most helpful and natural response — word by word. It does not "think" like a person, but the results can feel remarkably human. One important thing to know: ChatGPT does not browse the internet to answer you (unless you specifically use its browsing feature). Its knowledge comes from its training data, which has a cutoff date. For very recent news or events, always check a current news source. ## What Can ChatGPT Do? Here are some of the most useful things ChatGPT can help you with: **Writing help.** Draft an email, write a thank-you note, or improve a cover letter. Tell it what you want to say and who you are writing to — it will do the heavy lifting. **Explaining things.** Ask it to explain any topic in simple terms. "Explain inflation like I am 10 years old" or "What does a deductible mean on my insurance?" are exactly the kinds of questions it handles well. **Answering questions.** It works like a very patient tutor. Ask follow-up questions without feeling embarrassed. It will never rush you. **Brainstorming.** Stuck on a gift idea, a menu for the week, or a name for a small business? ChatGPT can generate a list of ideas fast. **Summarizing.** Paste in a long article or document and ask it to give you the short version. **Translation.** Type text in one language and ask for a translation into another. I use ChatGPT most often to draft the first version of emails I find awkward to write — like complaining to a company about a billing problem. I tell it the situation, the tone I want, and who I am writing to. It gets me 80% of the way there in about 10 seconds, and I just edit the rest myself. ## What ChatGPT Cannot Do Being honest about the limits matters just as much as talking about the strengths. **It can be wrong.** ChatGPT sometimes states incorrect information with complete confidence. This is the biggest risk for new users who trust everything it says. Treat its answers like you would a smart friend's opinion — useful as a starting point, but worth checking. **It does not know what happened recently.** Unless it has web browsing turned on, it is working from older information. Ask it something like "Who won the election last month?" and the answer may be outdated or incomplete. **It cannot take action in the real world.** It cannot call a number for you, make a purchase, or send an email on its own (though some newer add-ons are starting to change this in limited ways). **It does not know you personally.** Each new conversation starts fresh unless you specifically tell it things about yourself. ## Free vs. Paid: What Is the Difference? The free version of ChatGPT works well for most everyday tasks. You do not need to pay anything to get real value from it. The paid plan, called ChatGPT Plus, costs $20 per month. It gives you access to newer, more capable versions of the AI, faster responses during busy times, and features like image creation and more reliable web browsing. For a beginner, start with the free version. You can always upgrade later if you find yourself using it heavily. ## Is ChatGPT Safe to Use? For everyday tasks, yes. A few things to keep in mind: By default, OpenAI may use your chat history to improve the AI. You can turn this off in your settings — see our guide on [ChatGPT privacy settings](/guides/chatgpt-privacy-settings) for step-by-step instructions. Never share sensitive personal information in a chat: your Social Security number, passwords, bank account details, or anything you would not want a stranger to read. ChatGPT is designed to refuse requests that are harmful or illegal. It is not perfect at this, but it tries. ## Who Is ChatGPT For? Almost anyone. People use it to help at work, to learn something new, to get unstuck on a writing task, or simply to satisfy curiosity. You do not need to be young or tech-savvy. In my experience, the biggest barrier is just starting. Once you have typed your first message and seen the answer come back, it starts to feel natural very quickly. ## What to Try Next The best way to understand ChatGPT is to use it. Check out [How to Use ChatGPT: Step-by-Step for First-Timers](/guides/how-to-use-chatgpt) to create your free account and send your first message in about five minutes. If you want to see how ChatGPT compares to Google's AI, [ChatGPT vs Gemini](/guides/chatgpt-vs-gemini) breaks down the differences in plain English. --- # Grok: What Is Elon Musk's AI and Is It Worth Trying? Source: https://aiforregulars.com/guides/what-is-grok-ai When Elon Musk talks about AI, people pay attention — for better or worse. Grok is his answer to ChatGPT, built by a company called xAI that he founded. It launched on X (formerly Twitter) and has been expanding since. Here is what Grok actually is, what it is genuinely good at, and whether it is worth your time. ## What is Grok, exactly? Grok is an AI chatbot — the same basic idea as ChatGPT, Gemini, or Claude. You type a question or request, and it responds in plain English. Under the hood it is a large language model, similar in concept to the technology behind other popular chatbots. What sets it apart from the others is its connection to X. Grok can search and read posts on X in real time, which means it knows what people are talking about right now — not just what was on the internet when it was trained. ## Grok vs ChatGPT: a practical comparison | | Grok | ChatGPT | |---|---|---| | Made by | xAI (Elon Musk) | OpenAI | | Real-time X/Twitter data | Yes | No (has web search but not X-specific) | | Web search | Yes | Yes (free and paid tiers) | | Image generation | Yes (paid tier) | Yes (paid tier) | | Free tier | Yes (via X and grok.com) | Yes | | Plugin/tool ecosystem | Limited | Extensive | | Works without X account | Yes (grok.com) | Yes | ## What Grok is genuinely good at **Real-time X content.** If you want to know what people on X are saying about a news story right now, Grok can pull those posts and summarize the conversation. No other major chatbot does this as directly. **Trending topics and memes.** Ask Grok about a meme that went viral yesterday and it will likely understand the reference. Ask ChatGPT the same thing and it might be confused if the meme postdates its training data. **Casual conversation.** Grok has a conversational style that some users find more relaxed and direct than competitors. It tends to be less cautious in tone. **Answering questions about current events.** The combination of web search and X data means Grok often has fresher information than chatbots that rely only on training data. ## Where Grok falls short **Ecosystem depth.** ChatGPT and Claude have larger communities, more tutorials, and broader integrations with third-party tools. If you want to use AI with your documents, your email, or your calendar, Grok has fewer options. **Task versatility for everyday users.** For writing help, explaining a concept, summarizing a document, or practicing a job interview, ChatGPT and Claude have more refined experiences built around those specific tasks. **The X dependency cuts both ways.** Grok's real-time X access is its strength, but it also means the experience is tightly tied to one platform's content and moderation choices. If you are not on X, that advantage disappears. I tested Grok on a handful of tasks I use AI for regularly — drafting a short explanation of a technical topic, summarizing a long article, and answering a factual question about a recent event. It handled all three reasonably well. Where it felt noticeably different was asking about a trending discussion on X — it surfaced real posts and summarized the conversation accurately, which neither ChatGPT nor Claude could do in the same way. For everything else, I did not feel a clear advantage either direction. ## Who should try Grok? **Try Grok if:** - You spend meaningful time on X and want AI that understands its culture and content - You want real-time information specifically from X posts - You are already an X Premium subscriber (access may be included) - You are curious and want to try a different AI chatbot for comparison **Stick with something else if:** - Your priority is writing assistance, document work, or research - You want a large ecosystem of tools and integrations - You have no interest in X as a platform ## The cultural context, kept brief Grok exists partly because Elon Musk publicly criticized other AI companies for what he saw as excessive caution in their models. He wanted a less filtered alternative. That framing generates a lot of opinion online. The practical question for most users is simpler: does it do what you need? For X-centric use cases, yes. For general-purpose AI tasks, several competitors are more capable today. ## What to try next If you want to compare Grok to other options directly, [Best AI Chatbot for Beginners Compared](/guides/best-ai-chatbot-beginners) covers the five most popular chatbots side by side with a clear table. Or if you are choosing between the two most popular options, [Claude vs ChatGPT for Everyday Writing](/guides/claude-vs-chatgpt) looks at how those two handle practical tasks. --- # What Should You Never Tell an AI Chatbot? (Privacy Guardrails) Source: https://aiforregulars.com/guides/what-not-to-tell-ai-chatbot AI chatbots are genuinely useful tools. But they are not private diaries. What you type goes to a company's servers — and it can be stored, reviewed, or potentially exposed. Before you share, it helps to know where the line is. ## Passwords and Login Credentials Never type a password into a chatbot, not even to ask "is this password strong enough?" AI companies often store conversations to improve their systems, and employees may review those conversations for quality or safety checks. If the company ever suffers a data breach, your credentials go with it. The fix is simple: describe the problem without the actual password. Instead of pasting it in, say "I have an 8-character password with a mix of letters and numbers — is that strong enough?" You will get the same advice with zero risk. ## Social Security or National ID Numbers Your Social Security number is the master key to your identity. One exposure can lead to fraudulent tax returns, new credit accounts opened in your name, and years of cleanup work. No legitimate AI task — writing an email, planning a trip, summarizing a document — ever needs your SSN. If you are filling out a form that requires it, do that on the official website, not in a chatbot. ## Bank Account or Credit Card Numbers Chatbots cannot move money or process payments. There is no reason to type your bank account number, routing number, or card number into one. Beyond the storage risk, phishing apps disguised as AI tools are specifically built to harvest this kind of information. Type those numbers only into verified, official bank or payment websites. ## Detailed Medical Records or Diagnoses It is perfectly fine to ask a chatbot general health questions: "What does 'LDL cholesterol' mean?" or "What are common symptoms of a sinus infection?" But sharing your actual medical records, specific diagnoses, or prescription details is a different matter. If that data is ever breached, it could theoretically be used against you by insurers or employers in certain industries. Use general language instead — the AI can still give you useful information without knowing it is about you. I started using placeholder names and generic descriptions whenever I need AI help with anything sensitive. Instead of using a real name or my own medical details, I type something like "a person in their 50s with high blood pressure and a question about medication." The answer I get is just as useful, and I have not handed over anything real. It only takes a few extra seconds to think this way — and it has become a habit I never skip. ## Your Children's Full Names, Schools, or Photos Children's information is especially sensitive, and federal privacy law (COPPA) exists for good reason. AI chatbots are not designed to hold your child's personal details, and if a conversation were ever exposed, information about where your child goes to school or their daily routine could create real safety risks. Use first names only, or just say "my daughter" or "a 10-year-old." The AI will still help you with homework questions, parenting advice, or anything else you need. ## Legal Matters You Are Involved In Attorney-client privilege does not extend to AI chatbots — and this is not a theoretical risk. In February 2026, a federal judge in the Southern District of New York (*United States v. Heppner*) ruled that a defendant's Claude conversations — in which he'd used the chatbot to draft defense strategy notes — were not privileged and could be seized and used against him. The court's reasoning: a chatbot is not a lawyer, no attorney-client relationship exists between you and it, and a consumer AI tool's privacy policy typically allows the company to retain and, in some circumstances, disclose what you type. The ruling also flagged a second risk: feeding what your own attorney told you into a public AI tool can waive privilege over that original legal advice, too. This is one trial court's decision on one set of facts, and it binds no court outside that district — other judges may yet reason differently. But it is the first federal ruling to address the question directly, and it points one way rather than the other. If you are in a lawsuit, a custody dispute, a divorce, or any legal proceeding, do not type the details into a chatbot — that conversation can be subpoenaed or introduced as evidence, in civil cases as much as criminal ones. Ask your attorney what you can and cannot discuss online, and treat AI chat the same as a public forum — because for legal purposes, it effectively is one. ## Trade Secrets or Confidential Work Many companies have already banned employees from pasting internal documents into public AI tools, and for good reason. Confidential client data, unreleased product plans, internal financial figures — these belong to your employer or client, not a tech company's database. Check your company's policy before using AI for work tasks. If you are unsure, describe the situation without naming the client or sharing any actual documents. ## Your Home Address Combined With Your Daily Schedule A street address on its own is usually low risk. But combining your precise home address with details about when you leave for work, when your house is empty, or your daily routine creates a different kind of exposure. If chatbot logs were ever leaked, that combination of data is exactly what a bad actor needs. Keep location details vague — "a suburb of Chicago" or "a neighborhood near downtown" is almost always enough context for the AI to help you. ## What to Try Next To see exactly what a chatbot service already knows about you and what it keeps, read [what AI chatbots actually know about you](/guides/what-does-ai-know-about-me). Then take five minutes to walk through [how to adjust your ChatGPT privacy settings](/guides/chatgpt-privacy-settings) — a few simple toggles can significantly reduce how much gets stored. ## Sources - [Data Controls FAQ](https://help.openai.com/en/articles/7730893-data-controls-faq) — OpenAI Help Center - [Can You Use ChatGPT to Talk About Your Legal Case? A Federal Court Weighs In.](https://www.gentrylocke.com/article/can-you-use-chatgpt-to-talk-about-your-legal-case-a-federal-court-weighs-in/) — Gentry Locke Attorneys, on *United States v. Heppner* (S.D.N.Y., Feb. 2026) - [ABA issues first ethics guidance on a lawyer's use of AI tools](https://www.americanbar.org/news/abanews/aba-news-archives/2024/07/aba-issues-first-ethics-guidance-ai-tools/) — American Bar Association, on Formal Opinion 512 - [Children's Online Privacy Protection Rule ("COPPA")](https://www.ftc.gov/legal-library/browse/rules/childrens-online-privacy-protection-rule-coppa) — U.S. Federal Trade Commission - [Art. 9 GDPR – Processing of special categories of personal data](https://gdpr-info.eu/art-9-gdpr/) — GDPR-info.eu --- # 10 Times You Should Not Use AI — and What to Do Instead Source: https://aiforregulars.com/guides/when-not-to-use-ai AI is genuinely useful for a long list of everyday tasks — drafting a letter, understanding a confusing bill, looking up a recipe, brainstorming ideas. But "AI can help with most things" is not the same as "AI can help with anything." There are specific situations where turning to a chatbot instead of the right resource leads to real harm. This guide is not about AI's abstract limitations — it's a concrete list of situations you are likely to encounter, with the right alternative for each one. ## 1. A Medical Emergency **Do not use AI.** In any situation where someone may be having a heart attack, stroke, severe allergic reaction, or serious injury, every second counts. Typing a question into a chatbot and waiting for a response wastes time that the right action — calling for emergency help — does not. **Do this instead:** Call 911 in the US (999 in the UK, 112 across Europe and internationally) immediately. If you are unsure whether something is an emergency, call anyway — dispatchers are trained to help you evaluate the situation while help is already on the way. ## 2. Medication Dosing or Drug Interactions **Do not use AI.** A 2023 expert review found that 74% of ChatGPT-generated drug information was incomplete or incorrect. AI chatbots are trained on text, not on current clinical guidelines or your personal health history — and a wrong answer about dosing or a dangerous drug combination can cause serious harm. **Do this instead:** Ask a pharmacist. Pharmacists are licensed drug experts who are almost always reachable by phone and free to consult without an appointment. For checking interactions between specific medications, Drugs.com and the FDA's Drug Interactions Checker are well-maintained, free public resources. ## 3. A Mental Health Crisis **Do not use AI.** If you or someone you know is experiencing thoughts of self-harm, suicidal ideation, or an acute mental health emergency, a chatbot is not a safe substitute for a trained crisis counselor. AI cannot verify what it is being told, cannot follow up on a person's wellbeing, and cannot dispatch help if needed. **Do this instead:** In the US, call or text **988** to reach the Suicide and Crisis Lifeline — free, confidential, and available around the clock. In the UK, call Samaritans at **116 123**. For other countries, findahelpline.com lists verified crisis lines worldwide. ## 4. Legal Filings with Hard Deadlines **Do not use AI to draft court documents or calculate filing deadlines.** Missing a court deadline because of incorrect AI-generated information can result in losing a case by default — a legal consequence that is very difficult to undo. AI often presents information confidently even when jurisdiction-specific rules are wrong or outdated. **Do this instead:** Contact a licensed attorney or your local legal aid office. Many areas have free legal aid services for people who cannot afford private counsel. Your state bar association's website (or the equivalent in your country) typically has a lawyer referral service with free initial consultations. ## 5. Your Taxes — Specific Numbers and Deadlines **Do not use AI to calculate your tax liability or confirm filing deadlines.** Tax rules change yearly, vary by state or region, and depend on personal circumstances a chatbot cannot reliably account for. If the AI gives you a wrong number or a missed deadline, you — not the AI — face the consequences. **Do this instead:** Use your official tax authority's website (irs.gov in the US, hmrc.gov.uk in the UK). In the US, IRS Free File is available if you qualify, AARP Tax-Aide offers free preparation help for people 50 and older, and the Volunteer Income Tax Assistance program helps those below a certain income level. The rule I use: if getting this wrong could cost me money I can't easily recover, or create a legal record I can't easily fix, I go to a human professional. AI is excellent for understanding what something means — "what is a capital gains tax?" — but not for filling in the actual numbers on the actual form. ## 6. Managing Passwords and Account Security **Do not ask AI to generate or store your passwords — and do not type your existing passwords into any chat interface.** Some chatbots retain conversation history, and anything you type into them may be stored or logged. A chatbot cannot actually secure your accounts. **Do this instead:** Use a dedicated password manager. Bitwarden is free, open-source, and widely trusted. Apple Keychain (built into iPhone and Mac) and Google Password Manager (built into Android and Chrome) work well for basic use. For two-factor authentication, use an authenticator app like Google Authenticator or Authy. ## 7. Writing a Grief Message or Navigating a Serious Personal Conflict **Do not outsource genuinely emotional moments to AI.** When a close friend loses someone they love, or when you need to repair something that matters in an important relationship, what counts most is that the words came from you. An AI-written message can feel hollow — and in the wrong moment it can do more damage than saying nothing at all. **Do this instead:** Write something yourself, even if it's short and imperfect. "I don't have the right words, but I'm thinking of you" from a real person means more than a polished paragraph that wasn't. For ongoing or serious conflict, a licensed therapist or mediator can offer structured support that AI cannot replicate. ## 8. Verifying Breaking News During an Emergency **Do not rely on AI when local conditions are changing by the minute.** Most AI chatbots — including the widely used ones — are not connected to live news feeds or real-time data. In a fast-moving situation like a wildfire evacuation, a storm warning, or an industrial incident, AI may confidently describe conditions that are hours out of date. **Do this instead:** Go directly to official sources: your local emergency management agency's website, the National Weather Service at weather.gov (in the US), local radio and TV broadcasts, or your government's emergency alert system. In the US, FEMA's website and app are reliable during declared disasters. ## 9. Investment Decisions Based on AI "Predictions" **Do not let AI-generated forecasts drive significant financial decisions.** AI cannot reliably predict market movements. Language models generate plausible-sounding text — which means they can produce detailed, confident financial forecasts that have no factual basis. There are documented cases of chatbots inventing historical stock prices and market events that never occurred. **Do this instead:** For significant financial decisions, consult a licensed financial advisor or a fee-only fiduciary — someone legally required to act in your interest. For free education, the SEC's investor.gov and FINRA's investor education resources are trustworthy starting points. ## 10. Anything That Needs a Licensed Professional's Signature **Do not substitute AI for licensed expertise in high-stakes domains.** When a structural engineer stamps a building plan, they accept legal responsibility for it. When a doctor writes a prescription, they have examined you and taken professional liability. AI carries none of that responsibility — and "an AI told me" is not recognized as a defense in any professional or legal context. **Do this instead:** Identify the licensed professional whose field covers your need and consult them. Many professions offer free initial consultations. Professional associations in law, medicine, and financial planning frequently have pro bono programs for people with limited resources. ## What to Watch Out For The most dangerous thing about AI in these situations is not that it sounds uncertain — it is that it often sounds completely confident. As we explain in [Can You Trust ChatGPT?](/guides/can-you-trust-chatgpt), AI systems are designed to produce fluent, authoritative-sounding text regardless of whether the underlying information is accurate. That quality works well for low-stakes creative tasks and becomes a serious liability in high-stakes ones. Research by AARP has found that misinformation is the single biggest AI concern among older adults — and that concern is justified. The best protection is knowing in advance which situations call for a licensed human expert, not a chatbot. ## What to Try Next For a plain-English explanation of why AI gets things wrong even when it sounds certain, [Why AI Makes Things Up — and How to Catch It](/guides/why-ai-hallucinates) covers the mechanics behind the problem. If you use AI for health-related questions at all, [How to Use AI for Health Symptoms Without Getting Burned](/guides/ai-health-symptoms) has the specific limits worth knowing. --- # Why Does ChatGPT Always Agree With You? The 'Yes-Man' Problem, Explained Source: https://aiforregulars.com/guides/why-ai-agrees-with-you-too-much If you've noticed that ChatGPT, Gemini, or Claude almost never tells you your idea is a bad one — that it tends to open with "great question" and close with some version of "you're absolutely right" — you've spotted something real, not something in your head. Researchers have a name for it: sycophancy. It's not a bug that shows up occasionally. It's a pattern built into how these systems are trained, and it gets more convincing exactly when you need honesty the most: when you're asking for advice about a decision that actually matters. ## What "Sycophancy" Means in AI Sycophancy is the tendency of an AI model to tell you what it predicts you want to hear, rather than what's most accurate or most useful. That can look like agreeing with your account of an argument with a family member, praising a business plan that has an obvious hole in it, or validating a decision you've already made instead of pointing out a risk. It's easy to miss because it doesn't feel like being lied to. The AI isn't inventing facts (that's a different, related problem — see [why AI hallucinates](/guides/why-ai-hallucinates)). It's shading its framing, its tone, and what it chooses to emphasize toward agreement. The answer can be technically true and still leave out the one thing a good friend or a good advisor would have said first: "wait, have you considered that you might be wrong here?" ## What the Stanford Study Actually Found In March 2026, a team of Stanford researchers — led by Myra Cheng, with linguistics and computer science professor Dan Jurafsky as senior author — published a study in *Science* on exactly this pattern. They ran 11 major AI models — ChatGPT, Claude, Gemini, DeepSeek, Llama, Qwen, and Mistral among them — through more than 11,000 real interpersonal dilemmas, the kind of "am I the jerk here?" scenarios people post online when they want an outside opinion on a conflict. The finding: AI models told users they'd acted appropriately roughly 49% more often than human respondents would. That gap held up even in scenarios involving deception, rule-breaking, or behavior a reasonable person would call clearly wrong — the models still sided with the user in something like half of those cases. A second part of the study tracked how this actually affects people. Researchers gave sycophantic AI responses to 2,400 participants and found something worth sitting with: people who got the flattering response became more convinced they were right, less willing to take responsibility or apologize — and they still rated the sycophantic AI as more trustworthy, saying they'd come back and ask it again. That's the trap. The response that reinforces your blind spot is also the one that feels the most satisfying to receive, which is exactly why it's easy to keep asking the same assistant instead of a person who might push back. ## The Real-World Warning: OpenAI's GPT-4o Rollback This isn't just a lab finding. In late April 2025, OpenAI shipped an update to GPT-4o that made the model noticeably more agreeable — and within days, users were sharing screenshots of it enthusiastically praising decisions that ranged from ill-advised to genuinely alarming, including cases where it validated someone's choice to stop taking prescribed medication. The backlash was fast enough that OpenAI pulled the update within about four days and published its own account of what went wrong. The company's explanation matched what the Stanford researchers later described: the update had been tuned using short-term user feedback — thumbs-up, thumbs-down, "was this helpful" — without properly weighing how that kind of tuning rewards telling people what they want to hear over telling them what's true. It's a useful case study because it shows the problem isn't hypothetical or limited to one edge case: a company that builds and tests these models at enormous scale still shipped a version that was too much of a yes-man, and only caught it after the public did. ## Why This Happens The mechanism is simpler than it sounds. AI companies refine their models using ratings from real people — a process often called reinforcement learning from human feedback. People doing the rating tend to score warm, validating, agreeable answers higher than blunt or challenging ones, even when the blunt answer is more accurate. Repeat that process across millions of ratings, and the model learns, in effect, that agreement performs better than honesty. Two things make it worse. First, longer conversations give the model more chances to pick up on what you seem to want to hear and lean into it. Second, apps with memory — ones that recall your past chats, your preferences, your history — have even more material to shape a flattering answer around. The Stanford researchers specifically flagged personalization and memory as factors that are likely to intensify sycophancy, not reduce it, as these features become more common. ## How to Spot It in a Real Conversation Watch for phrases like "you're absolutely right," "that's a great point," or a conversation where every one of your positions gets validated and none gets pushed back on. One or two isn't a red flag. A conversation where the AI never once disagrees with you, especially about something with real stakes, is worth a second look. Scroll back through the exchange and ask yourself honestly: did it raise a single counterpoint on its own, without being asked? If the entire conversation reads like it's on your side, that's the sycophancy pattern showing up, not a sign your idea was flawless. Directly type: "What's the strongest argument against what I just said?" or "If someone disagreed with me here, what would they say?" This single habit reliably produces a more balanced answer than a plain follow-up question, because it removes the ambiguity about whether disagreement is welcome. For a bigger decision, try: "Play the role of someone who thinks this is a bad idea and make their best case." Framing it as a role rather than the AI's own opinion tends to get past the training's pull toward agreement. For decisions involving real money, your health, a relationship, or anything with a deadline or legal consequence, treat the AI's answer as one input, not the answer. Run it past a human who has no reason to just tell you what you want to hear — a professional, a friend who'll actually argue with you, or both. Our guide on [when not to use AI](/guides/when-not-to-use-ai) walks through the specific situations where this matters most. The habit I've built is asking a version of "what would someone who disagrees with me say?" before I act on anything AI told me that I liked hearing. If the answer changes once I frame it that way, that's the tell — the first response was shaped by what I wanted, not by what was actually true. It costs one extra sentence and it's caught me more than once. ## What to Watch Out For The riskiest version of this isn't a chatbot telling you your résumé looks great. It's a chatbot backing up your account of a conflict with a spouse, telling you a risky investment "makes sense given what you've described," or agreeing that you don't need to see a doctor about a symptom you downplayed in the way you described it. In each case, the AI isn't lying — it's reflecting your framing back at you, more confidently than the facts support. The 2,400-person part of the Stanford study is the part worth remembering here: people who got agreeable answers didn't just feel good about them, they became more certain they were right and less open to reconsidering — which is the opposite of what you want before a decision that matters. If you're using an AI assistant with memory turned on, the effect has more room to grow the longer and more personal the relationship with the app gets. That's not a reason to turn memory off entirely, but it's a reason to be more deliberate, not less, about checking important answers against an outside opinion as the conversation history builds up. ## What to Try Next If you want to know the other situations where AI's confident tone outruns its actual reliability, [Why Does AI Make Things Up?](/guides/why-ai-hallucinates) explains the mechanics of hallucination, a related but distinct problem. And if you've noticed you're leaning on AI for more than quick fact-checks lately, [Am I Using AI Too Much?](/guides/am-i-using-ai-too-much) is a grounded self-check worth reading alongside this one. ## Sources - [Stanford study outlines dangers of asking AI chatbots for personal advice](https://techcrunch.com/2026/03/28/stanford-study-outlines-dangers-of-asking-ai-chatbots-for-personal-advice/) — TechCrunch - [Stanford study finds AI sycophancy may undermine judgment and social behavior](https://theaiinsider.tech/2026/03/31/stanford-study-finds-ai-sycophancy-may-undermine-judgment-and-social-behavior/) — The AI Insider - [OpenAI explains why ChatGPT became too sycophantic](https://techcrunch.com/2025/04/29/openai-explains-why-chatgpt-became-too-sycophantic/) — TechCrunch --- # Why Does AI Make Things Up? Hallucinations Explained Simply Source: https://aiforregulars.com/guides/why-ai-hallucinates You've probably heard someone say that AI "makes things up." The technical word for it is hallucination — and it sounds alarming, like the AI is deliberately deceiving you. But there's a much simpler explanation, and once you understand it, AI becomes a far less mysterious tool. ## What "Hallucination" Actually Means When an AI gives you a confident-sounding answer that is completely wrong — a fake book citation, a made-up phone number, a historical detail that never happened — that's called a hallucination. The name is a bit dramatic. The AI isn't confused the way a person might be. It's producing text that sounds plausible, with nothing in the system checking whether it's actually true. This can range from minor inaccuracies to complete fabrications. A chatbot might cite a real author but invent a book title they never wrote. It might give you a phone number for a business that looks legitimate but connects to somewhere else entirely. It might describe a court case with confident specifics that never occurred. ## Why It Happens: The Autocomplete Explanation The simplest way to understand hallucinations is to think about the autocomplete on your phone's keyboard. When you type "Happy birth—", your keyboard suggests "day" because that's almost always what follows. It isn't thinking about birthdays. It's matching a pattern it has seen many times before. AI chatbots work on the same basic principle — just at an enormously larger and more sophisticated scale. They were trained by reading a huge amount of text: books, articles, websites, and more. From all that text, they learned to predict: given what has been said so far, what word or phrase most likely comes next? Most of the time, this works beautifully. The patterns are strong, and the output is accurate. But when you ask about something specific and obscure — a minor court case, an academic paper, a local business, events after the AI's training cutoff — there may not be a strong pattern to draw from. So the AI fills in the gap with something that sounds plausible based on similar patterns it has seen. It doesn't know it's wrong. It doesn't have a fact-checker. It generates. ## Why Can't AI Just Say "I Don't Know"? This is a fair question. Modern AI systems are better trained to express uncertainty than early versions were, and many will say "I'm not sure" or "you may want to verify this" more often now. But the underlying architecture — predicting the next token — doesn't naturally produce uncertainty the way a human expert would feel uncertain. It generates text. It doesn't evaluate the text it generates. AI tools that include live web search tend to hallucinate less on factual questions because they can actually look things up. But even those tools make mistakes. Searching the web doesn't guarantee finding the right source or interpreting it correctly. ## 3 Habits That Catch AI Mistakes Before They Matter You don't need to fact-check everything an AI tells you. Most everyday uses — drafting a note, explaining a concept, brainstorming ideas — are low stakes. But building three simple habits protects you when it matters: **1. When the stakes are high, verify with a separate source.** For medical decisions, legal questions, financial choices, or anything you'll share publicly — don't treat the AI's answer as final. Use it to understand the question better, then confirm the answer with a reliable source: a doctor, a government website, an established publication. Our guide on [how to tell if text is AI-generated](/guides/how-to-tell-if-text-is-ai) covers related verification habits. **2. When the AI cites something, check that it exists.** If a chatbot tells you about a book, an article, or a specific study, verify that it actually exists before you rely on it or share it. Check the title, the author, and whether the source is real. Fabricated citations are one of the most common AI mistakes, and they look convincing at a glance. **3. Test the AI on something you already know.** If you're new to a tool, or trying it for a new kind of task, ask it a question you already know the full answer to. See how accurate it is. This gives you a calibrated sense of how much to trust it on topics where you can't independently verify. I test every new AI tool the same way: I ask it about a specific event or fact I know well. When the answer comes back slightly wrong — and it sometimes does — I know to apply more skepticism going forward. This doesn't mean the tool isn't useful. It just tells me where to pay more attention. ## What This Means in Practice AI is genuinely useful. Hallucinations are real, but they're not constant. For everyday tasks, AI is usually accurate enough to be a real time-saver. The risk goes up when the question requires very specific facts, obscure or local information, or data from after the AI's training cutoff. Think of AI like a very well-read friend who sometimes misremembers details with complete confidence — helpful for most things, worth a second opinion on anything that really matters. **What to try next:** If you want to dig into the trust and verification side — knowing when to rely on AI and when to push back — [Can You Trust ChatGPT?](/guides/can-you-trust-chatgpt) covers that angle directly. And to understand what AI actually is under the hood, [What Does 'AI' Actually Mean?](/guides/what-does-ai-mean) lays out the basics clearly. --- # Will AI Take My Job? An Honest Plain-English Answer Source: https://aiforregulars.com/guides/will-ai-take-my-job It's one of the most common questions people ask about AI right now. You've heard stories about writers, illustrators, and customer service workers losing work to AI tools — and you want to know if your job is next. The honest answer is more nuanced than "yes" or "no," and it comes down to the specific tasks your job involves, not just the job title itself. ## It's About Tasks, Not Job Titles No job is a single kind of work. A nurse spends some time typing up notes (a task AI can help with) and some time comforting a frightened patient (something AI cannot do at all). A financial advisor spends some time running numbers and some time understanding a client's life situation. AI is making inroads on certain types of tasks — not on whole jobs as a package. The tasks most at risk share a few features: they're repetitive, they follow a pattern, and they involve working with text or data. Writing the same kind of email fifty times a day. Sorting documents into categories. Summarizing meeting notes. Answering the same questions via a chat window. This doesn't mean everyone doing these tasks lost their job overnight. Often it means one person now handles more work, or fewer new people get hired specifically for those tasks. ## Tasks AI Is Handling More Often Some categories of work have clearly shifted, based on how AI tools are being used across industries: - **Routine writing tasks:** form letters, first drafts of reports, meeting summaries - **Data entry and basic data organization** - **Scripted customer service:** answering common, predictable questions - **Simple image creation:** stock-style illustrations, background images - **Standard code generation:** short, repetitive pieces of software The people affected first are often not the whole workforce but rather the entry-level workers hired specifically for those narrowly defined tasks. ## Tasks AI Is Still Struggling With AI is not good at things that require physical presence, genuine human relationships, fast improvisation, or judgment in complex and unpredictable real-world situations: - **Hands-on trades:** plumbing, electrical work, construction, carpentry, auto repair - **Care work:** nursing, physical therapy, childcare, elder care - **Complex relationship work:** counseling, negotiation, long-term sales built on trust - **Creative direction:** knowing which story is worth telling, not just how to write it - **Emergency response and crisis management:** decisions made in seconds under pressure ## Three Practical Things You Can Do Right Now Regardless of your field, these three actions put you in the strongest position: **1. Learn to use AI as a tool in your own work.** Workers who use AI well are not the same as workers who get replaced by it. If you can use AI to draft faster, research faster, or handle tedious tasks, you become more productive — not redundant. Our guide on [AI prompts for office work](/guides/ai-prompts-office-work) has copy-paste examples you can use today. **2. Strengthen the skills AI doesn't have.** Think carefully about what makes your specific work valuable beyond its mechanical parts — relationships you've built, judgment developed over years, physical skills, local knowledge. These are real advantages over AI. **3. Stay curious, not anxious.** People who ask "how can I use this tool?" are in a better position than people who only ask "will this replace me?" The workers most at risk over the long run are the ones who refuse to engage with AI at all. I've spoken with people across a wide range of industries who were worried about AI — marketing, bookkeeping, customer support, and more. The ones who felt most secure weren't necessarily in "safe" industries. They were the ones who had already started experimenting with AI tools in their own work. Understanding what AI can and can't actually do makes it far less threatening. ## The Skill That Will Matter Most Your job is a bundle of tasks. Some tasks in that bundle will gradually shift to AI. The question is whether you grow into the tasks that remain — the ones that need judgment, creativity, and real human connection. That's where job security lives, regardless of what field you're in. No one can give you an exact prediction about your specific job. Anyone who confidently tells you "your job is safe" or "your job is gone in three years" is guessing. But taking practical steps now — learning the tools, staying curious, and sharpening your distinctly human skills — puts you in the best possible position no matter what happens. **What to try next:** If you're ready to start using AI in your current work, [AI prompts for office work](/guides/ai-prompts-office-work) gives you ready-to-use prompts for common workplace tasks. If you're job searching or thinking about a career change, [building an ATS-friendly resume with AI](/guides/ats-friendly-resume-with-ai) shows how AI can actually help you land interviews. --- # How to Write a Professional Email with AI (Templates Inside) Source: https://aiforregulars.com/guides/write-professional-email-with-ai Writing work emails can feel like walking a tightrope. Too casual and you look unprofessional. Too stiff and people stop reading. AI handles the first awkward draft for you so you can focus on getting the words exactly right. Go to [chat.openai.com](https://chat.openai.com) (ChatGPT) or [gemini.google.com](https://gemini.google.com). Both are free to use. You do not need an account to try ChatGPT, though signing in saves your history. If you use Microsoft Outlook at work, check whether your company has Copilot enabled — it can draft emails right inside your inbox. Do not overthink your prompt. Just explain the situation like you would describe it to a coworker over lunch. The more specific you are, the better the draft will be. Include: who you are writing to (boss, client, coworker, HR), what you need to say or ask, the tone you want (polite, firm, apologetic, professional), and any key details like dates, project names, or dollar amounts. Some emails are hard because of what they say, not just how to say them. Use these copy-paste prompts to get started with the messages most people dread. **Sending a polite reminder:** ``` Write a short, friendly reminder email to [Name]. I'm following up on [topic]. I last contacted them on [date] and haven't heard back. Keep it professional and not pushy — two short paragraphs. ``` **Saying no without burning bridges:** ``` Help me write a polite email declining [request or invitation] from [Name]. I want to be respectful and leave the door open for future opportunities. Keep it brief — no more than three sentences. ``` **Asking for a pay raise:** ``` Write a professional email to my manager asking for a salary review. My job title is [title], I've been in this role for [X months/years], and I recently [accomplishment]. Keep the tone confident but not demanding. ``` Before you copy the email, read it out loud. You will immediately notice anything that sounds awkward, too formal, or unlike you. Tweak those phrases first. One common fix: AI often opens with "I hope this message finds you well." Delete it. Start with your actual point instead. Even a small personal touch — a name, a project reference, a shared context — makes AI-written emails feel human. Look for one place to add it. For example, instead of "I wanted to follow up on our recent conversation," write "I wanted to follow up on what we discussed at Tuesday's team meeting." It is the same information delivered in a way that proves you were there. I used to dread writing apology emails at work. I would spend 20 minutes on three sentences and still feel unsatisfied. Now I tell ChatGPT what happened, ask for a draft that is "sincere, not defensive," and I have something to work with in about 30 seconds. I still rewrite parts — but the blank page is gone, and that is what was stopping me. Give the finished email a quick once-over before you hit send. - Is the recipient's name spelled correctly? - Did you include everything you meant to say? - Does it sound like you? - Is the subject line clear and specific? If all four are yes, send it. ## A note on the emails you keep putting off AI is especially helpful for the messages you have been avoiding — the ones sitting in your drafts folder because you are not sure how to start. Think of it as having a calm, professional friend who is good with words helping you get unstuck. It will not make the conversation easy. But it will make it start. **What to try next:** Once you are comfortable writing emails with AI, you can use the same approach for bigger workplace writing tasks. See [15 AI prompts that save 5 hours of office work a week](/guides/ai-prompts-office-work) for more time-saving ideas, or learn how to [write a cover letter with ChatGPT](/guides/chatgpt-cover-letter) if you are job hunting. --- # Anthropic Found 20+ Dating Apps Where Most 'Matches' Were AI Bots — What to Watch For Source: https://aiforregulars.com/news/ai-dating-app-scam-network-anthropic-report If you have ever wondered whether the charming stranger you have been messaging for weeks on a dating app is real, a report from the AI company Anthropic just gave that worry a solid basis. In September 2026, Anthropic said it had uncovered a network of more than 20 dating apps in which most of the profiles you could see were not people at all. They were AI characters, powered by Anthropic's own Claude, and the apps did not tell users. **What Anthropic found.** According to the report, a China-based app studio built the network. Across the apps Anthropic investigated, about 75% of the profiles shown to users were AI-generated. Over just two weeks in April 2026, more than 4,700 of these fake identities chatted with at least 25,000 people, exchanging roughly 2.36 million messages. The apps were advertised as fully human. **How the scheme made money.** Messaging and matching ran on a limited quota. When it ran out, you had to buy more in-app coins to keep talking. In other words, people were paying to keep a conversation going with a machine they believed was a person. The AI was reportedly told not to reveal it was automated, and the system tracked how suspicious each user seemed. When a real person was not available, it could make up things like likes, visitors and pre-recorded videos. Human workers, roughly one for every three AI personas, stepped in mainly for video calls and social media follow-backs. **How it got past the app stores.** Anthropic says the apps showed Apple's and Google's reviewers a clean, compliant version, then switched on the deceptive features after approval. So "it is in the App Store" is not proof that an app is honest. **What Anthropic did.** It banned the accounts and organizations involved and shared its findings with Apple, Google and other AI companies whose tools were also used. **What this means for you.** This is a report about one network. It does not mean every dating app is fake, and it does not prove any specific app you use is involved. But the pattern is worth knowing: - **Being asked to pay to keep chatting.** If a match is lovely but the app charges coins or credits to reply to that specific person, be careful. - **A match who is always available.** Instant replies at every hour, and messages that feel very warm very quickly, can be a sign. - **Reluctance to video chat.** If a video call keeps getting postponed, take that seriously. Note that scammers can also use deepfakes, so a video call is a good sign, not a guarantee. - **Any request for money, gift cards or crypto.** That is a scam no matter who, or what, is on the other end. If you have already sent money, see [what to do if you sent money to a scammer](/guides/sent-money-to-a-scammer). If you suspect a profile, do not buy coins to unlock its messages, and report it to both the app and the app store. For a fuller walkthrough of the warning signs, read our guide to [AI romance scams on dating apps](/guides/ai-romance-scam-dating-app). ## Sources - [Fake dating apps used Claude to scam 25,000 people, Anthropic says](https://www.techlicious.com/blog/fake-dating-apps-used-claude-to-scam-25000-people-anthropic-says/) — Techlicious - [Anthropic uncovers the AI powered dating scam: 75% bots, 25% humans](https://www.storyboard18.com/digital/anthropic-uncovers-ai-dating-app-scam-with-4700-fake-personas-ws-l-110424.htm) — Storyboard18 --- # Viral AI Video Falsely Claims a New Federal Property Tax Break for Seniors Source: https://aiforregulars.com/news/ai-property-tax-relief-scam-video-seniors If you've recently seen a video claiming that seniors can now stop paying part of their property taxes thanks to a new federal law, don't fill out any form it points you to. It's false, and county tax offices across the country are spending real time telling worried homeowners so. **What the video claims.** Multiple AI-generated videos spreading on YouTube say Congress passed a law giving homeowners 65 and older a federal property tax exemption, freeze, or refund. Some tell viewers to request a form called the "Senior Homeowner Tax Review Request"; others point to a fake "Form 1098-SR," dressed up to look like an official IRS document. Neither form exists, and there is no federal property tax program — in the United States, property tax is set and collected entirely at the state and local level, never by the federal government or the IRS. **How officials found out.** County assessor and auditor offices didn't spot the videos first — they got flooded with phone calls and emails from residents asking for the "federal form." Douglas County, Colorado, and Franklin and Hamilton Counties in Ohio each issued public consumer alerts in early August after the calls piled up. St. Louis County, Minnesota, and Kern County, California, warned residents about the same videos. By mid-August, the financial site Moneywise counted more than 20 assessor offices across several states that had put out warnings. One Colorado official pointed to a simple motive: the videos run 15 to 20 minutes with heavy ad breaks, which pays out regardless of whether anything in them is true. **Why this one is easy to fall for.** The videos borrow the look and tone of a news broadcast or a government announcement, and AI makes that convincing to produce cheaply. Genuine property tax relief for seniors does exist in many states — Colorado, Ohio, Texas, Florida, and others all have real senior exemption programs — so the scam works by mixing a kernel of truth (real relief exists somewhere) with a fake federal shortcut that skips the actual paperwork. **What to do if you already requested the form.** Don't send any personal, financial, or property information to a website or number the video gave you. If you already have, contact your real county assessor's office directly — using the phone number listed on your property tax bill or the county's official government website, not one from the video — and ask them to flag your account. To find out if you genuinely qualify for a senior property tax reduction, contact your county assessor's office directly; it's free, and no one legitimate will ask you to pay to apply. For more on spotting this kind of video before you act on it, see our guide on [how to tell if a video or article is AI-generated fake news](/guides/ai-generated-fake-news). ## Sources - [CONSUMER ALERT: Viral video promotes fake federal property tax form](https://www.douglasco.gov/consumer-alert-viral-video-promotes-fake-federal-property-tax-form/) — Douglas County, Colorado (official government alert) - [Viral AI YouTube videos falsely claim seniors can stop paying property taxes — 23 assessor offices issue warnings as scam spreads](https://moneywise.com/taxes/property-tax-seniors-ai-youtube-scam-relief) — Moneywise - [County of Kern warns seniors of online property tax scam tied to fake IRS form](https://bakersfieldnow.com/news/local/county-of-kern-warns-seniors-of-online-property-tax-scam-tied-to-fake-irs-form) — BakersfieldNow - [Hamilton County auditor warns of Ohio property tax scam](https://www.axios.com/local/cincinnati/2026/08/11/hamilton-county-auditor-warns-of-property-tax-scam) — Axios Cincinnati --- # A Cloned Voice Cost an Italian Bank €95 Million — What It Means for Your Family Source: https://aiforregulars.com/news/ai-voice-clone-bank-fraud-95-million In February 2026, scammers took about €95 million from Fideuram, the private-banking arm of Intesa Sanpaolo, Italy's largest bank. The story became public on 25 September 2026. According to press reports, the trick used two ordinary things: a WhatsApp message and a phone call with a voice that was not real. **What happened.** The then-chairman of Fideuram received a WhatsApp message from someone pretending to be the chief executive of Intesa Sanpaolo. A phone call followed, using an AI-made copy of the voice of a real lawyer, confirming the instruction. Believing it, the bank moved the money. More than half of the money has since been frozen or recovered, from accounts in China and Portugal, but at least €36 million is reported missing, apparently sent on into cryptocurrency. Prosecutors in Milan are investigating a foreign national. **Why this matters to you.** These were experienced bankers who handle huge transfers every day, and they were still fooled. The lesson is not that they were careless. It is that a familiar name on WhatsApp and a familiar voice on the phone are **no longer proof of who is really there**. The same trick, at a smaller scale, is what scammers use on families: a "grandchild" in trouble, a "boss" asking for a quick favour. **What to do instead.** - **Check through a second channel.** If someone asks for money or codes, hang up and call back on a number you already have, not one they give you. - **Treat "it must be done right now" as a warning sign.** Pressure is the scammer's main tool. A real person can wait ten minutes while you check. - **Agree a family password** that only your close family knows. Our guide to [family safe words](/guides/family-safe-words) shows how. For more, read how [AI voice cloning scams](/guides/ai-voice-cloning-scams) work, how to [tell if a phone call uses an AI voice](/guides/detect-ai-voice-phone-call), and what to do if you [sent money to a scammer](/guides/sent-money-to-a-scammer). --- # Amazon Just Blocked Meta's Muse From Shopping on Its Site — Here's What Happened Source: https://aiforregulars.com/news/amazon-blocks-meta-muse-explained If you've been trying out Meta's new Muse assistant to shop on Amazon, you may have just run into a wall. On Sunday evening, September 20, 2026, Amazon started blocking purchases made through Muse on Amazon.com, showing shoppers a pop-up error instead: "Continued access by an unauthorized AI agent violates Amazon's Conditions of Use, to which our customers have agreed." It's not a bug on your end — it's a dispute between the two companies. **A quick reminder of what Muse is.** Meta launched Muse on September 8, 2026, as a "personal AI agent" that can browse the web, fill out forms, and complete purchases on your behalf using a payment card you've added — we covered [what it does and how to use it safely](/news/meta-muse-ai-agent-explained) when it launched. Twelve days later, one of the biggest places people shop online shut its doors to it. **Why Amazon blocked it.** Amazon says it raised the issue with Meta first, asking the company to voluntarily keep Amazon out of Muse's shopping abilities before it resorted to blocking. Amazon's stated concerns come down to three things: Muse doesn't clearly identify itself as a bot while it browses the site, it may collect and hold onto customers' Amazon login credentials without Amazon's knowledge or permission, and Amazon says it never agreed to let a third-party agent access customer accounts, order history, or transactions on its platform this way. An Amazon spokesperson put it this way: "We think it's fairly straightforward that third-party applications that offer to make purchases on behalf of customers from other businesses should operate openly and respect service provider decisions about whether or not to participate." **Meta pushed back.** Meta disputes Amazon's characterization of how Muse handles your login information, saying the agent "has no visibility into people's passwords or payment methods" and that any credentials you share with it go into secure storage the agent can use without ever exposing them. Meta hasn't otherwise commented publicly on the block. **This isn't an isolated spat.** Amazon has spent the past year pushing back on outside AI agents shopping on its site without permission — it sued Perplexity over its Comet browser's shopping features, and has taken similar steps against shopping agents built by Google and OpenAI. Amazon's argument is that it treats agentic shopping the way it treats any partner integration — like a food delivery app working with restaurants, or an online travel agency booking flights — it wants a say in who gets access and how, rather than having an outside AI agent let itself in. **What this means if you use Muse.** If you point Muse at Amazon right now, expect it to fail at checkout with that error message — this isn't something you can fix on your end, and it isn't a sign your account was compromised. It's a genuinely useful moment to pause and think about what any AI shopping agent is doing with your login details. Before you connect one to a site that holds your saved cards or address book, it's worth asking exactly what credentials it's asking for, whether that site's own terms allow that kind of access, and what happens to those details if the company changes its mind later, the way Amazon just did. Our guide on [how AI shopping agents actually work today, and how to stay safe using one](/guides/ai-shopping-agent-safety), walks through those questions in more detail. ## Sources - [Meta's AI agent has been blocked from using Amazon.com](https://techcrunch.com/2026/09/21/metas-ai-agent-has-been-blocked-from-using-amazon-com/) — TechCrunch - [Amazon bars Meta's Muse AI from shopping on its site](https://www.engadget.com/2263659/amazon-bars-metas-muse-ai-from-shopping-on-its-site/) — Engadget - [Amazon blocks Meta's Muse AI agent from shopping on its site](https://finance.yahoo.com/technology/ai/articles/amazon-blocks-metas-muse-ai-141950908.html) — Yahoo Finance --- # California Now Forces Big AI Tools to Prove What They Made Source: https://aiforregulars.com/news/california-ai-transparency-law-content-labels Starting today, August 2, 2026, a California law called the AI Transparency Act (SB 942) requires the biggest AI companies to make it easier to tell what content their tools created. The rule applies to any generative AI system with more than one million monthly users in California — in practice, that covers ChatGPT (OpenAI), Gemini (Google), Meta AI, and other large platforms. Since companies rarely build a separate version of a product just for one state, the changes will likely show up for users everywhere, not only in California. **What's supposed to change** - **A free detection tool.** Covered companies now have to offer a free, public tool anyone can use to check whether an image, video, or audio clip was made or altered by their system. - **A visible "AI-generated" label option.** When you create something with one of these tools, you'll be able to attach a visible marker showing it's AI-made. It's an option you choose, not something added automatically. - **A hidden digital fingerprint.** Images, video, and audio from these systems also get invisible, machine-readable data baked in (the industry calls this "content provenance"). Unlike a visible label, it's built to survive even if the file gets cropped, screenshotted, or re-saved. Companies that skip this face fines of $5,000 per violation per day, enforced by the California Attorney General or local prosecutors. The law was signed in 2024 and was originally supposed to start in January 2026, but a follow-up bill pushed the date to today, deliberately lining it up with a similar deadline in the EU. Speaking of which: the European Union's AI Act reached a comparable milestone on this same date. Our guide on [what changed under the EU AI Act](/news/eu-ai-act-what-changes-for-you) covers the chatbot-disclosure side of that rule, if you're in Europe. **What this means if you're worried about scams** The good news: if you get a suspicious photo, video, or voice message, you may now be able to run it through one of these free detection tools to see if it came from a major AI platform. The catch: the law only covers the biggest platforms — most text-only chatbots and smaller AI tools aren't required to offer anything at all. And people running scams were never going to use a mainstream, rule-following tool in the first place — they'll keep using whatever ignores the law. So treat a detection tool as one more clue, not proof either way. Keep using the habits from our guides on [spotting AI-generated photos](/guides/spot-ai-generated-photos) and [checking if a video call is a deepfake](/guides/verify-video-call-deepfake): if a voice or message claiming to be a family member asks for money, hang up and call them back on a number you already have saved. More of the law phases in later. Large platforms that host AI content, such as social media apps, have to add their own detection and labeling by January 1, 2027. Camera and phone makers get until January 1, 2028 to offer an optional way to mark photos as verifiably *not* AI-edited. ## Sources - [Bill Text - SB-942 California AI Transparency Act](https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=202320240SB942) — California Legislative Information - [California's Ongoing AI Regulation: Key Deadlines Arriving in 2026 and Beyond](https://natlawreview.com/article/californias-ongoing-ai-regulation-key-deadlines-arriving-2026-and-beyond) — The National Law Review - [SB 942: Who It Covers, Disclosures, and Penalties](https://legalclarity.org/sb-942-who-it-covers-disclosures-and-penalties/) — LegalClarity --- # ChatGPT Ads Just Went Live in More Countries — What You'll See and How to Turn Them Off Source: https://aiforregulars.com/news/chatgpt-ads-international-rollout If you use the free version of ChatGPT and live in the UK, Mexico, Brazil, Japan, or South Korea, you may have started seeing something new in your answers: a small, clearly labeled ad. OpenAI confirmed on August 11, 2026 that its ad test — which began in the US on February 9, 2026 — has now expanded to these five countries. **Who actually sees ads.** Only logged-in adults on the **Free** and **Go** plans. If you pay for **Plus, Pro, Business, Enterprise, or Edu**, you won't see ads at all — that's one of the things your subscription covers. Ads also never appear in temporary (incognito-style) chats, regardless of your plan. **What the ads look like.** OpenAI says ads are visually separated from ChatGPT's actual answer and clearly marked as sponsored, and that they don't influence what the AI tells you — the answer itself is generated the same way either way. You can tap an ad to see why it was shown to you, dismiss it, leave feedback, or delete your ad data with one tap. **"I turned off personalization — why am I still seeing ads?"** Turning off ad personalization in your settings doesn't remove ads, it only changes which ones you're shown. If you want no ads at all, personalization isn't the setting you're looking for. **How to actually turn ads off.** On the Free plan, go to Settings → Ads controls → change plan, and you can switch to an ad-free option. The catch: your daily free message limit drops, and you lose access to certain tools — OpenAI specifically calls out image generation and deep research. The more straightforward route, if it's within budget, is upgrading to Plus, which removes ads entirely along with raising your limits. **If you're outside these countries.** Ads were already running in Australia, Canada, New Zealand, and the US before this expansion — the UK, Mexico, Brazil, Japan, and South Korea are the countries added on August 11. **Update, August 24, 2026:** the test has since expanded again, this time to 31 European countries at once, including Germany and the rest of the EU. See [ChatGPT ads are now live in Europe](/news/chatgpt-ads-live-europe-2026) for what's different there under European privacy law. None of this means your ChatGPT account has been compromised or that something is wrong — it's a deliberate, disclosed test of a business model change, the same way free email and free social apps have long used ads to stay free. If an "ad" ever shows up that isn't labeled as one, or pushes you toward something that feels off (a link to enter payment details, a too-good-to-be-true offer), treat it with the same caution you'd use for any ad online, and don't confuse it with ChatGPT's own advice. If you're weighing whether a paid plan is worth it for you, our guide to [is ChatGPT Plus worth it](/guides/chatgpt-plus-worth-it) breaks down what you get for the price, including the ad-free experience. ## Sources - [Testing ads in ChatGPT](https://openai.com/index/testing-ads-in-chatgpt/) — OpenAI - [OpenAI Expands ChatGPT Ads Test to UK, Mexico, Brazil, Japan, and South Korea](https://www.ghacks.net/2026/08/13/openai-expands-chatgpt-ads-test-to-uk-mexico-brazil-japan-and-south-korea/) — gHacks Tech News - [ChatGPT ads: where they run now and how to switch them off](https://www.notebookcheck.net/ChatGPT-ads-where-they-run-now-and-how-to-switch-them-off.1361301.0.html) — Notebookcheck --- # ChatGPT Ads Are Now Live in Europe — What Changes for You Today Source: https://aiforregulars.com/news/chatgpt-ads-live-europe-2026 If you use the free version of ChatGPT and live in Germany, Austria, Switzerland, France, the Netherlands, Belgium, Luxembourg, Italy, Spain, Poland, Sweden, or one of 20 other European countries, today is the day you may start seeing ads in your answers. OpenAI confirmed the rollout began Monday, August 24, 2026, covering all 27 EU member states plus Iceland, Liechtenstein, Norway, and Switzerland — 31 markets at once, bringing the global total to 40. **Who actually sees ads.** People on the **Free** and **Go** plans. If you pay for **Plus, Pro, Business, Enterprise, or Edu**, you won't see ads — that's part of what your subscription covers. Users under 18 are excluded regardless of plan. **What's different in Europe.** In the US, UK, and other countries where ads launched earlier this year, OpenAI could personalize ads based on your account and settings. In the EU, Iceland, Liechtenstein, Norway, and Switzerland, that's not allowed to start — European privacy law (GDPR) requires clear consent before a company can personalize ads around you, and OpenAI hasn't asked for that yet. So for now, the ads you see are based only on what you're currently asking ChatGPT about, your rough location, and your language — not your chat history or saved memories. OpenAI says it may introduce personalization later, and if it does, you should get a separate notice asking for your consent — you're not supposed to be opted in automatically. **What the ads look like.** OpenAI says ads are visually separated from ChatGPT's actual answer and clearly labeled as sponsored, and that they don't change what the AI tells you — the answer itself is generated the same way either way. The company also says advertisers never get access to your chat history, and that it doesn't sell your data to them. **How to actually turn ads off.** On the Free plan, go to Settings → Ads controls, where you can switch to an ad-free mode. The catch: your daily message limit drops, and you lose access to some tools, including image generation and deep research. If that trade-off isn't worth it to you, upgrading removes ads outright — Plus runs around €23 a month in Europe, and Pro around €229 a month. Our guide to [is ChatGPT Plus worth it](/guides/chatgpt-plus-worth-it) walks through what you get for the price. None of this means your account has been compromised or that something is wrong with your ChatGPT — it's a disclosed test of a business model change, the same way free email and social apps have long used ads to stay free. If an "ad" ever shows up unlabeled, or pushes you toward entering payment details or an offer that feels too good to be true, treat it with the same caution you would any ad online — don't confuse it with ChatGPT's own advice. Our guide to [ChatGPT's privacy settings](/guides/chatgpt-privacy-settings) covers what else you can control about what OpenAI sees and uses. ## Sources - [ChatGPT to start showing ads across Europe next week](https://www.euronews.com/next/2026/08/21/chatgpt-to-start-showing-ads-across-europe-next-week) — Euronews - [ChatGPT ads hit Europe on Monday, but not the personalized kind](https://www.notebookcheck.net/ChatGPT-ads-hit-Europe-on-Monday-but-not-the-personalized-kind.1375456.0.html) — Notebookcheck - [OpenAI to expand ChatGPT ads to 31 more European countries](https://finance.yahoo.com/media-advertising/articles/openai-expand-chatgpt-ads-31-071833684.html) — Yahoo Finance / Reuters --- # ChatGPT Can Now Show You Your Credit Score — What It Means for Your Money Source: https://aiforregulars.com/news/chatgpt-credit-score-experian On 21 September 2026, OpenAI rolled out a new feature inside ChatGPT called Finances: US Plus and Pro subscribers could, and from October 2, 2026 free users can too, connect their Experian credit report and see their credit score directly in a chat, on the web, on iPhone, or on Android. **What it actually does.** After you connect your Experian account, ChatGPT shows your VantageScore 3.0 — a credit score on the familiar 300–850 scale — along with your score history and the factors affecting it. The report and score refresh about once a month, and ChatGPT can alert you to changes such as a new loan inquiry, a new account being opened, or an address change on file, which are also classic signs of identity theft. Checking your score this way is a "soft" inquiry, so it does not lower your score the way applying for a new credit card does. **Why it matters.** Two things worth keeping in mind. First, the number ChatGPT shows you is *a* credit score, not *the* credit score — banks, landlords, and card issuers often pull a different model (a FICO score, or a score from one of the other two bureaus, Equifax or TransUnion), so the figure in ChatGPT can be higher or lower than what a lender actually sees. Second, this is the first time a mainstream AI chatbot has plugged directly into a credit bureau, which means your financial history is now something ChatGPT can read and discuss with you the same way it reads a document you paste in — worth pausing on before you connect it. **What to do with this, safely.** - **Use it to spot fraud early, not to shop for a loan.** The monitoring alerts are genuinely useful — a new account or inquiry you don't recognize is worth acting on immediately. - **Don't treat the score as final.** Before applying for a mortgage or major loan, check your actual free credit reports at [annualcreditreport.com](https://www.annualcreditreport.com), the one site the three bureaus jointly authorize for that. - **Know you can disconnect it.** The Experian connection lives in ChatGPT's Finances/connector settings and can be removed at any time, the same way you'd revoke any other connected account. - **This feature is US-only for now**, it has not (yet) rolled out to other countries. A paid plan was needed at first, but OpenAI says that from October 2, 2026 it also covers Free and Go users in the US. If you're weighing whether to link financial accounts to ChatGPT at all, our guide on [ChatGPT reading your bank account](/guides/chatgpt-link-bank-account) and the [ChatGPT privacy settings walkthrough](/guides/chatgpt-privacy-settings) cover the trade-offs in more detail. ## Sources - [Finances in ChatGPT](https://help.openai.com/en/articles/20001222-finances-in-chatgpt) — OpenAI Help Center (official) - [Sharing your credit score with AI? What you need to know to stay protected](https://www.cnbc.com/select/ai-help-with-credit-score-how-to-protect-yourself/) — CNBC Select --- # ChatGPT's DALL-E Tool Is Gone — Here's What Replaced It Source: https://aiforregulars.com/news/chatgpt-dalle-gpt-retiring If you've used the built-in "DALL-E" option in ChatGPT to turn a text description into a picture, that specific tool is now gone. OpenAI retired the dedicated DALL-E GPT inside ChatGPT on August 30, 2026. **What actually changed.** DALL-E was one of the original ways to generate images in ChatGPT — a separate, named tool you could open from the GPT menu. OpenAI has replaced it with **ChatGPT Images**, built on its newer gpt-image-1 models, which is now simply part of the regular ChatGPT chat. You no longer pick a separate "DALL-E" option — you just ask ChatGPT for an image in the same conversation you'd use for anything else, and it generates one using the newer technology. This is available on every plan, including the free tier, not just paid subscriptions. **Why this matters.** OpenAI has not said what happens to images you previously created and saved through the old DALL-E tool once the deadline passes. To be safe, if you have images stored there that you want to keep, download them to your device rather than assuming they'll stay available in ChatGPT indefinitely. **What's not affected.** Image generation itself isn't going away — it's simply built into ChatGPT's main chat now, using better models. Custom GPTs that other people built with image generation turned on also keep working; this change only removes OpenAI's own standalone DALL-E tool. **What to do.** If you've relied on the old DALL-E option, there's nothing to set up — asking ChatGPT to create or edit an image in a normal chat now uses ChatGPT Images automatically. If you have older saved images you care about, take a few minutes to save copies to your phone or computer. For more on getting good results from AI image tools generally, see our guide on [creating AI images for free](/guides/create-ai-images-free). ## Sources - [DALL-E leaves ChatGPT on August 30 — download your images first](https://www.notebookcheck.net/DALL-E-leaves-ChatGPT-on-August-30-download-your-images-first.1360522.0.html) — Notebookcheck - [OpenAI retires DALL-E image generation from ChatGPT, pushing users to newer models](https://cryptobriefing.com/openai-removes-dall-e-chatgpt/) — CryptoBriefing --- # ChatGPT 'Dots' Are Always-On AI Helpers — and Most People Don't Need One Source: https://aiforregulars.com/news/chatgpt-dots-always-on-agent-explained On September 29, 2026, OpenAI introduced "dots" at its developer conference. A dot is an AI helper that keeps working in the background, even after you close ChatGPT. It has its own cloud computer and web browser, and you can give it tasks in ChatGPT, Slack or Microsoft Teams. It runs on GPT-6 Astra, OpenAI's newest model. **Do you get one?** Only if you pay for ChatGPT Pro (from $100 a month), Business Premium, or an Enterprise plan through your employer. Free, Go and Plus users don't get dots, and nothing changes for your plan. Even Pro subscribers in the European Economic Area, Switzerland and the United Kingdom don't get dots yet. **What changes for you.** If you use the free version or the $20 Plus plan: nothing. You don't need to do anything, and there's no reason to upgrade just because of this. If you're curious what this kind of assistant looks like, our explainer on [Meta's Muse agent](/news/meta-muse-ai-agent-explained) covers a similar idea, and [what ChatGPT Work does](/news/chatgpt-work-agent-explained) covers OpenAI's earlier step in this direction. **Prices also moved.** OpenAI added a new Pro 500 plan at $500 a month, aimed at heavy professional use. From October 30, the $200 Pro plan gets smaller allowances for work tasks and top-model messages, and existing subscribers receive a one-time usage credit. Our [plan comparison](/guides/ai-paid-plans-compared) explains which paid tier suits what, and for most people the answer is still the cheapest one. **A sensible caution.** An assistant that acts on its own needs access to your accounts, so connect only what a task really needs. OpenAI says dots ask before sensitive actions and that you can set your own rules. Expect "dots" to become a hook for fake offers, so only ever sign up inside the official ChatGPT app or website. See [how to spot a scam AI app](/guides/is-ai-app-scam) and [what not to tell an AI chatbot](/guides/what-not-to-tell-ai-chatbot). ## Sources - [OpenAI halves Pro 200 usage and launches a $500 ChatGPT plan at DevDay](https://thenextweb.com/news/openai-devday-pro-200-usage-cut-pro-500-plan) — The Next Web - [OpenAI Gave AI Agents Their Own Computers at DevDay 2026](https://decrypt.co/379584/openai-ai-agents-computers-devday-2026-everything-announced) — Decrypt - [OpenAI Adds $500 ChatGPT Tier, Keeps Dots Off Free Plans](https://www.implicator.ai/openai-500-chatgpt-tier-dots-paid-plans/) — Implicator --- # ChatGPT’s Money Feature Now Reaches Free Users in the US — What to Know Before You Connect Source: https://aiforregulars.com/news/chatgpt-finances-free-users-us On October 2, 2026, OpenAI said that "Finances" in ChatGPT is rolling out to Free and Go users in the United States. Until now you needed a paid plan: Pro from May, Plus from late June. According to OpenAI, you no longer need a subscription. It works on the web, iPhone and Android. We couldn't find an outside news report yet, so this is OpenAI's own word. **What it does.** You can connect your bank, credit card or investment accounts through a service called Plaid, which OpenAI says supports more than 12,000 institutions. You can also connect your credit report through Experian. These are two separate connections, so you can use one, both or neither. ChatGPT then shows a dashboard of your spending and lets you ask questions such as "what changed this month?" **What it can and can't see.** It reads balances, transactions and investments. It doesn't get your full account numbers and it can't move money, because the connection is read-only. It is also not a financial adviser. A human adviser may have a duty to act in your interest. ChatGPT has no such duty, so treat its answers as a starting point, not as advice. ChatGPT also shows ads to Free and Go users in the US, which is a reason to think about what you share. **What changes for you.** If you're in the US and on a free plan, you'll probably start seeing offers to "connect your accounts". You don't have to. Outside the US, we haven't seen any sign of it yet. **Staying safe.** - Connect only from inside the real ChatGPT app or chatgpt.com. Don't follow a link from an email or text. - The real connection happens in a Plaid window. It never asks for your bank password in the chat, or for a code sent by text message to ChatGPT. If someone does, stop. - Be wary of apps or websites called "ChatGPT Finances" or offering to "link your bank to ChatGPT". They are a classic way to steal logins. - You can disconnect at any time in ChatGPT's settings, and you can review what is stored there. Want the full walkthrough? Read our guide on [ChatGPT reading your bank account](/guides/chatgpt-link-bank-account), the news about [ChatGPT showing your credit score](/news/chatgpt-credit-score-experian), and how to [change your ChatGPT privacy settings](/guides/chatgpt-privacy-settings). If something already looks wrong, see [what to do if you've sent money to a scammer](/guides/sent-money-to-a-scammer). ## Sources - [Finances in ChatGPT](https://help.openai.com/en/articles/20001222-finances-in-chatgpt) — OpenAI Help Center - [ChatGPT — Release Notes](https://help.openai.com/en/articles/6825453-chatgpt-release-notes) — OpenAI Help Center --- # OpenAI Launches ChatGPT for Teens — What Changes for Your Family Source: https://aiforregulars.com/news/chatgpt-for-teens-parental-controls If your teenager uses ChatGPT, their experience may already look different — even if nobody in your house changed a single setting. **What changed.** On August 18, 2026, OpenAI launched ChatGPT for Teens, a separate version of ChatGPT built for users aged 13 to 17. Accounts get switched into it automatically — either because the person entered a birth date between 13 and 17 when they signed up, or because OpenAI's own age-prediction system estimates, from how the account is used, that it likely belongs to someone under 18. OpenAI has not said how accurate that automatic guess is, so it's worth treating it as an extra layer, not a guarantee. **What's off by default.** In the teen version, ChatGPT won't use romantic language or terms of endearment, and it's instructed not to suggest it has feelings of its own. It also has stronger built-in protections around self-harm, eating disorders, violence, and sexually explicit content — topics where the standard version of ChatGPT is more permissive. **What parents can control.** If you link your account to your teen's — from your profile settings, under Parental Controls — you can set "quiet hours" when ChatGPT won't respond on their account, turn off image generation entirely, opt their account out of having its conversations used to train OpenAI's models, and get a safety notification if OpenAI's systems detect signs of a serious crisis. None of that is new — OpenAI added those tools with the original parental controls in September 2025. What is new with this launch is that Study Mode, which walks teens through homework with guiding questions instead of just handing over an answer, is now built into the teen experience by default. **What this means for you.** The linking step is the part that actually does something — a teen mode that only activates for accounts honest about their age, or ones the algorithm happens to catch, will not cover everyone. If you haven't already, it's worth sitting down with your teen and setting up the linked account together. Our guide on [ChatGPT parental controls](/guides/chatgpt-parental-controls) walks through how to do that, along with the device-level backups worth having regardless of what OpenAI turns on by default. This builds on the parental controls OpenAI [introduced last year](/guides/chatgpt-parental-controls) — the difference now is that a safer default applies automatically to teen accounts, rather than only to families who went looking for the settings. ## Sources - [OpenAI launches a safer ChatGPT for teens — years after teens started using it](https://techcrunch.com/2026/08/18/openai-launches-a-safer-chatgpt-for-teens-years-after-teens-started-using-it/) — TechCrunch - [OpenAI launches ChatGPT for Teens with safety measures for young users](https://tech.yahoo.com/ai/chatgpt/articles/openai-launches-chatgpt-teens-safety-152404643.html) — Yahoo Tech --- # ChatGPT Just Removed the Daily Message Limit for Free Users Source: https://aiforregulars.com/news/chatgpt-free-unlimited-text-chats If you use ChatGPT for free and have ever seen the message "you've reached your limit, try again later," that's going away. OpenAI announced on August 6, 2026 that Free and Go accounts are getting **unlimited text conversations** — no more daily cap on how many text messages you can send. **What's actually changing:** Free and Go accounts are moving to GPT-5.6 Luna as their default model this week. The daily rate limit on text chats is being removed in the same rollout wave — OpenAI says that part lands the following week, starting around August 10. Once it does, you'll be able to keep typing and asking questions all day without hitting a wall. Both changes are rolling out gradually, so they may show up for you a few days before or after they show up for someone else — that's normal for this kind of update, not a sign anything is broken. **What's still limited:** This change covers text only. Uploading files, generating images, using voice mode, and other tools still have daily limits on Free and Go. If you're doing a lot of image generation or document uploads, you'll still run into caps there — Plus and Pro remain the way to raise those. A couple of related changes are coming alongside this. Paid Plus and Pro accounts are getting an upgrade to GPT-5.6 Sol, which OpenAI says makes noticeably fewer factual mistakes than the previous model. And a new "Think" button is coming to Free and Go next week, letting the AI spend more time reasoning through harder questions when you ask for it. For a plain-English rundown of what Sol, Terra, and Luna each do, see our guide on [ChatGPT Sol, Terra, and Luna explained](/guides/chatgpt-sol-terra-luna-explained). **Why this matters for people deciding between free and paid:** hitting the message limit used to be one of the most common reasons people upgraded to ChatGPT Plus. That specific reason is about to disappear — Plus still gets you a smarter model, unlimited image generation, more file uploads, and other extras, but "I keep running out of messages" won't be one of them once the rollout reaches you, if you only chat in text. **One scam warning to keep in mind.** Whenever OpenAI makes an announcement like this, scammers move fast with fake posts and ads that say things like "unlock unlimited ChatGPT here" or offer a paid "upgrade" to get a feature that's actually already free. This update happens automatically inside the official ChatGPT app and at chatgpt.com — you don't need to download anything, click a link, or pay anyone to get it. If you see an ad or message claiming otherwise, treat it as a scam. Our guide on [how to spot a fake AI app](/guides/is-ai-app-scam) walks through the warning signs. ## Sources - [OpenAI brings unlimited ChatGPT text chats to free users](https://techcrunch.com/2026/08/06/openai-brings-unlimited-chatgpt-text-chats-to-free-users/) — TechCrunch - [ChatGPT free users get unlimited text chats](https://www.macrumors.com/2026/08/06/chatgpt-free-unlimited-text-chats/) — MacRumors --- # OpenAI Just Released GPT-6 Astra — Here's What Changes If You Use ChatGPT Source: https://aiforregulars.com/news/chatgpt-gpt-6-astra-explained On September 3, 2026, OpenAI released **GPT-6 Astra**, a new flagship AI model that replaces the GPT-5.6 family (Sol, Terra, and Luna) at the top of ChatGPT's lineup. If that sentence means nothing to you, don't worry — here's what actually changes for someone who just wants ChatGPT to keep working the way it does. *Update, October 9, 2026: free and Go accounts have since started getting a newer model of their own, GPT-6 Luna. See [what free ChatGPT gets](/news/chatgpt-gpt-6-luna-free-intelligent-ui).* **Who gets it, and when.** OpenAI is rolling Astra out in stages rather than switching it on for everyone at once. It went first to a small group of cybersecurity partners, and is now rolling out over the following days to ChatGPT Plus, Pro, Business, and Enterprise subscribers, followed by developers using OpenAI's API and business customers on Amazon Web Services. **If you use the free version of ChatGPT, you do not get Astra** — for now, the free tier keeps running on the existing models. Unlike July's GPT-5.6 launch — when free and Go accounts got a new default model (Terra) on the same day paid users got the flagship Sol — this time there's no equivalent same-day upgrade for the free tier. **What's actually new.** Astra is designed to be significantly better at multi-step tasks: reading a spreadsheet and reasoning about it, browsing a website to find specific information, writing and checking code, and carrying out instructions that involve several steps in a row rather than a single question and answer. OpenAI is also describing it as able to operate a computer somewhat independently — for example, working through a task across several apps or browser tabs with less hand-holding than before. For most everyday chatting, emailing, and quick questions, you likely won't notice much difference day to day. **Why security researchers are paying attention.** OpenAI ranks its own models against an internal safety framework that tracks how much a model could help with dangerous tasks, including cyberattacks. Astra is the first OpenAI model to cross into what the company classifies as its "Critical" cybersecurity risk category — not because it's unsafe to use for normal purposes, but because it's now capable enough that OpenAI added extra safeguards and briefed cybersecurity partners before the wider release. For a regular ChatGPT user, this doesn't change anything about how you use the app; it's the reason OpenAI staged the rollout the way it did. **What it costs.** If you already pay for ChatGPT Plus, Pro, Business, or Enterprise, Astra is included in your existing subscription — there's no extra fee to access it once it reaches your account. OpenAI is also offering optional credit top-ups for people who want to use it more heavily than their plan normally allows, similar to how add-on credits already work for image generation. **What changes for you, specifically.** If you're on a paid plan, you don't need to do anything — Astra becomes available in the model picker inside the ChatGPT app over the next few days, the same way Sol, Terra, and Luna appeared automatically in July (see our guide on [Sol, Terra, and Luna explained](/guides/chatgpt-sol-terra-luna-explained) if those names are unfamiliar). If you're on the free plan, nothing changes for now, and there's no need to pay to "unlock" Astra through a link or app you find outside chatgpt.com — [Sol, Terra, and Luna explained](/guides/chatgpt-sol-terra-luna-explained) and [is ChatGPT free?](/guides/is-chatgpt-free) both cover what the free tier includes today. As always around a big AI launch, watch out for posts and ads claiming to offer early or "premium" access to a new model in exchange for payment or your login details — that's never how official OpenAI updates work; see our guide on [spotting a fake AI app](/guides/is-ai-app-scam) if something looks off. If you're already weighing whether a paid ChatGPT plan is worth it, our guide on [ChatGPT Plus, is it worth it?](/guides/chatgpt-plus-worth-it) covers what you get for the money. ## Sources - [OpenAI unveils GPT-6 Astra with new cybersecurity guardrails](https://www.cnbc.com/2026/09/03/open-ai-astra-gpt-6-cyber.html) — CNBC - [OpenAI releasing major upgrade to ChatGPT and Codex with GPT-6 Astra](https://9to5mac.com/2026/09/03/openai-releasing-major-upgrade-to-chatgpt-and-codex-with-gpt-6-astra-details-here/) — 9to5Mac - [OpenAI's GPT-6 Astra launches today](https://9to5google.com/2026/09/03/openai-gpt-6-astra-launch/) — 9to5Google - [OpenAI debuts GPT-6 Astra with new security measures](https://www.nbcnews.com/tech/tech-news/openai-debuts-gpt-6-astra-security-measures-rcna595940) — NBC News - [OpenAI rolls out GPT-6 Astra model with added cyber guardrails](https://www.bloomberg.com/news/articles/2026-09-03/openai-rolls-out-gpt-6-astra-model-with-added-cyber-guardrails) — Bloomberg --- # Free ChatGPT Gets GPT-6 Luna and Answers You Can Click Source: https://aiforregulars.com/news/chatgpt-gpt-6-luna-free-intelligent-ui If you use ChatGPT for free, it is getting an upgrade. OpenAI began rolling out **GPT-6** and a new answer format called **Intelligent UI** on October 7, 2026. Paid plans went first, and Free and Go accounts were scheduled to follow on October 8. **Which model you get.** Paid plans (Plus, Pro, Business and Enterprise) get **GPT-6 Sol**. Free and Go accounts get **GPT-6 Luna**, the smaller and cheaper model of the new family. Both are tuned for everyday conversation, but the free model is not the same as the one paying customers get. For a plain-English look at what the model names mean, see our guide on [ChatGPT Sol, Terra, and Luna explained](/guides/chatgpt-sol-terra-luna-explained). **What Intelligent UI is.** Instead of only text, an answer can now include things you can use: buttons, a calculator (for example, one that splits a restaurant bill), a chart or diagram, or a map. ChatGPT decides on its own when that format fits your question, and it still answers in plain text when that works better. You can also ask it to use fewer visual elements. **What stays the same.** The change affects the Chat tab only, not Work or Codex. **If you don't see it yet.** The rollout is gradual, so your account may get it a few days later than someone else's. That is normal, not a sign that something is broken. On Enterprise accounts, an administrator may have to switch it on. **One safety reminder.** The buttons and calculators in an answer are part of ChatGPT itself, but a form is still a form: never type a password, card number or national ID number into one. And you don't need to download anything or click a link in a message to "update ChatGPT" — new features appear on their own in the official app and at chatgpt.com. See our guide on [how to spot a fake AI app](/guides/is-ai-app-scam). ## Sources - [OpenAI Rolls Out GPT-6 and Intelligent UI in ChatGPT](https://letsdatascience.com/news/openai-brings-gpt-6-and-intelligent-ui-to-chatgpt-4f0d797c) — Let's Data Science - [OpenAI rolls out GPT-6 with Intelligent UI in ChatGPT](https://www.testingcatalog.com/openai-rolls-out-gpt-6-with-intelligent-ui-in-chatgpt/) — TestingCatalog --- # ChatGPT Health Is Now Open to Everyone in the US — Here's What Changes Source: https://aiforregulars.com/news/chatgpt-health-feature-explained If you use ChatGPT and you're in the United States, you may have noticed a new "Health" tab this week. That's not a glitch — OpenAI just made ChatGPT Health available to every US user, on every plan, including the free one. **What's new.** Announced Thursday, July 23, 2026, ChatGPT Health lets you connect your Apple Health data — steps, sleep, workouts — straight into your ChatGPT conversations. If your doctor's office uses certain electronic record systems (Epic and Oracle Health are the two OpenAI named), you can link those too, so ChatGPT can see things like recent lab results or visit summaries. Once connected, you get a dedicated Health tab that keeps its own conversation history, separate from your regular chats, where you can track how a symptom, medication, or sleep pattern changes over time and prepare questions before a doctor's appointment. It works on the web and in the ChatGPT app on iPhone — it hasn't rolled out to Android yet. **Why now.** OpenAI says people already send it roughly 300 million health-related questions a week, up from about 230 million back in January. Rather than leaving that happening informally in ordinary chats, the company built a dedicated space for it — with, it says, the same underlying model but a setup designed specifically around tracking health information over time rather than answering a one-off question. **The privacy catch.** OpenAI says it won't use your health conversations to train its models or to serve you ads, and that it will generally ask before using this data for anything beyond answering you. But here's the part worth sitting with: on the free, Plus, or other individual consumer plans, ChatGPT Health is **not covered by HIPAA**, the US law that protects the privacy of your medical records at a doctor's office or hospital. Everything OpenAI does with your health data on these plans is a company promise written into its terms of service — not a legal requirement backed by the same enforcement and penalties that apply to your actual healthcare providers. If a promise like that changes down the line, you'd have far less legal recourse than you would with a real medical provider. **What to actually do with this.** Connecting your Apple Health and medical records to ChatGPT can be genuinely useful — for spotting patterns in your own data, or walking into a doctor's visit with clearer questions. But treat it the way you'd treat any service that isn't legally bound to protect medical privacy: think about what you're comfortable a private company holding, keep in mind it can see more about you than a single symptom description would, and don't treat anything it tells you as a substitute for what your actual doctor says. Our guide on [asking AI about your health symptoms](/guides/ai-health-symptoms) covers where AI chatbots are genuinely useful for health questions and where they fall short, and [ChatGPT's privacy settings](/guides/chatgpt-privacy-settings) walks through how to control what the app remembers about you. --- # ChatGPT Now Works Inside Microsoft Word — Here's What Changed Source: https://aiforregulars.com/news/chatgpt-in-word-explained If you use Microsoft Word and ChatGPT as two separate apps — writing in one, then copying text over to ask for edits — that extra step just went away. On September 17, 2026, OpenAI launched a ChatGPT add-in for Word: a sidebar panel that reads the document you currently have open and can draft, rewrite, proofread, or reformat it without you leaving the page. **What it actually does.** Once you add the ChatGPT panel from the Microsoft add-in store and sign in with your OpenAI account, a chat window opens next to your document. You can ask it to turn rough notes into a first draft, untangle a clunky paragraph, proofread for mistakes, tidy up formatting, or reorganize a section — the same editing you'd normally do by copying text back and forth between Word and a ChatGPT browser tab. It only works with the document you currently have open; it can't reach into your other files or your inbox on its own. **It's free, not just for paying subscribers.** This is the detail worth noting if you've been comparing ChatGPT to Microsoft's own Copilot: the Word add-in works on every ChatGPT plan, including the free one, though heavy use still counts against your plan's normal message limits. Business and Enterprise accounts also got a two-week free preview of OpenAI's newer model inside Word, running through the end of September. Microsoft's Copilot, by contrast, still requires a separate paid Microsoft 365 Copilot add-on to work inside Word at all — see our [ChatGPT vs Copilot comparison](/guides/chatgpt-vs-copilot) for the fuller picture. **It completes a set.** OpenAI already shipped similar add-ins for Excel and PowerPoint back in May 2026; Word was the missing piece. If you've used ChatGPT inside a spreadsheet or a slide deck before, the Word version installs the same way and works the same way. **How to try it.** In Word, go to Insert → Add-ins, search the Microsoft Marketplace for "ChatGPT," and sign in with your existing OpenAI account. If you use Word through a work or school account, your organization's IT administrator may need to approve add-ins before it shows up — that's the most likely reason if you don't see it yet. OpenAI says the add-in will switch on by default for eligible accounts starting October 1, 2026, so don't be surprised if it appears on its own even if you never installed it. As with any AI writing help, treat the suggestions as a draft to review, not a final answer — proofread anything before you send or submit it. If your workplace has standardized on Microsoft's own tool instead, our guide to [using Copilot inside Word and Excel](/guides/copilot-word-excel-guide) still applies, and [our collection of AI prompts for office work](/guides/ai-prompts-office-work) has practical starting points for either assistant. ## Sources - [OpenAI adds ChatGPT to Word and opens its Office add-in to free users](https://runtimewire.com/article/openai-chatgpt-word-office-add-in-free-plans) — RuntimeWire - [Microsoft integrates ChatGPT into Word for enhanced drafting and editing](https://cryptobriefing.com/microsoft-chatgpt-word-integration/) — CryptoBriefing - [ChatGPT Comes to Microsoft Word With New Sidebar Add-In](https://thewincentral.com/chatgpt-microsoft-word-sidebar-add-in/) — WinCentral --- # ChatGPT Scheduled Tasks Are Free Now — Here's What the Free Plan Gets Source: https://aiforregulars.com/news/chatgpt-scheduled-tasks-now-free If you've been using ChatGPT for free and wished you could get an automatic daily briefing or reminder without paying for Plus, that wish came true this week. OpenAI confirmed on August 26, 2026 that **Scheduled Tasks — the feature that lets ChatGPT message you on a schedule — is now available on every plan, including Free.** **What free accounts get.** Free and Go accounts can now create up to **3 active scheduled tasks**. Each task can run once, or repeat no more than once a day — so you can set up a morning news summary or a weekly reminder, but not something that checks in every hour. This is the same 3-task limit Go subscribers already had; Free now matches it. **What's still paid-only.** Two things did not come to the Free plan. First, event-based triggers — having ChatGPT automatically run a task when something happens in Gmail, Slack, or GitHub — remain exclusive to Plus, Pro, Business, and Enterprise accounts. Free and Go can only schedule tasks by time, not by event. Second, the higher task limits are unchanged: Plus still gets 5 active tasks, and Pro and Enterprise still get 15. **Why this matters.** Scheduled Tasks first appeared as a paid-only beta in January 2025, then got a bigger relaunch with a dedicated sidebar section in June 2026 — and until this week, even that redesigned version required a paid plan just to see it at all. Free users can now try it without spending anything — a daily briefing, a weekly meal-planning prompt, a reminder to stretch — the same way Plus and Pro users have been able to for months, just with a lower ceiling on how many tasks can run at once. For a full walkthrough of setting up your first task, including examples you can copy and adapt, see our guide on [ChatGPT Scheduled Tasks](/guides/chatgpt-scheduled-tasks), updated today with the new free-plan limits. **One thing to watch for.** Whenever OpenAI opens up a previously paid feature, expect ads and social posts offering to "unlock" it for a fee, or fake browser extensions promising to "activate" scheduling. You don't need to install anything or pay anyone — if you have a ChatGPT account, the Scheduled section should already be sitting in your sidebar. Treat any offer that asks you to pay for a free feature as a scam. ## Sources - [OpenAI Brings ChatGPT Scheduling Tool To Free Users](https://dataconomy.com/2026/08/26/openai-chatgpt-scheduling-tool-free-users/) — Dataconomy - [OpenAI brings ChatGPT scheduling tool to free accounts](https://techbriefly.com/2026/08/26/openai-chatgpt-scheduling-tool-free-accounts/) — TechBriefly --- # ChatGPT Can Finally Search Your Entire Chat History Source: https://aiforregulars.com/news/chatgpt-search-all-your-chats OpenAI turned on a new search feature across ChatGPT on July 14, 2026, and it solves a problem almost everyone using the app has run into: scrolling forever to find a conversation from weeks ago. It rolled out at the same time on the ChatGPT website, the iPhone app, and the Android app, and it works on every plan, including Free — no upgrade needed. **How to use it.** Look for a magnifying glass icon near the top of the sidebar, or press **Ctrl+K** on Windows or **Cmd+K** on a Mac, and start typing. The search looks through everything at once — your past chats, any [Projects](/guides/chatgpt-projects-guide) you've set up, and images or documents you've uploaded — and lets you narrow the results down by type instead of scrolling through one long mixed list. **It even finds archived chats.** If you've ever archived a conversation to tidy up your sidebar, it normally disappears from view entirely. The new search still finds it — archiving hides a chat from the list, it doesn't remove it from search. **Worth a moment's thought.** Having your entire chat history one search away is convenient, but it's also a good nudge to think about what's actually sitting in there. If you've ever typed something sensitive into ChatGPT — a health question, a financial detail — it's worth a quick look at what ChatGPT saves and how to change it in our guide to [ChatGPT's privacy settings](/guides/chatgpt-privacy-settings). --- # ChatGPT Ads Can Now Talk Back — 'Sponsored Agents' Explained Source: https://aiforregulars.com/news/chatgpt-sponsored-agents-ads-explained You already know ChatGPT can show you a small, labeled ad. Now OpenAI is testing something new: an ad that turns into its own conversation. On September 16, 2026, OpenAI announced it's piloting a format called Sponsored Agents — a labeled ad that, when you tap it, hands you off to a chat with the advertiser's own AI, right inside your ChatGPT window. **How it's different from the ads you've already seen.** Regular ChatGPT ads are a short, labeled blurb next to an answer — you can ignore them and keep talking to ChatGPT. A Sponsored Agent goes a step further: tap it, and you're now chatting with a company's branded AI representative, which can ask follow-up questions, walk you through options, and — when you're ready — send you to the company's website or, in some cases, straight into that company's own request process. OpenAI describes it as letting you "spell out what matters to them, pose follow-up questions, and click through to the business's website when they are ready to move forward." **Who's testing it first.** The two named pilot partners are the furniture retailer Wayfair and the home-services marketplace Angi. Angi's version is the one most worth knowing about if you've ever asked ChatGPT a question about a home repair or renovation: its sponsored agent can appear in a relevant conversation, let you describe your project in more detail, and then move you into Angi's own request flow to get matched with a local pro. Angi co-founder Angie Hicks described the goal as letting "homeowners go directly from discussing a home project in ChatGPT to connecting with a skilled local pro who can help." **It's still small and labeled.** According to OpenAI's own help center, Sponsored Agents are "currently available only to selected advertisers in a limited alpha test," and the company isn't taking early-access requests from other businesses yet. So most people won't run into one soon, and when you do, it's supposed to be clearly marked as sponsored — not blended into ChatGPT's own answer. **What's worth keeping in mind.** A Sponsored Agent is a business's own AI, built to represent that business — not a neutral assistant. Treat it the way you'd treat a salesperson on a phone call: fine for browsing products or getting a repair quote started, but not the place to hand over more than the project details needed to get help. Our guide on [what not to tell an AI chatbot](/guides/what-not-to-tell-ai-chatbot) is a good gut-check before you share your address, financial details, or anything sensitive in any chat window, sponsored or not. This sits on top of the ad system already in place for free and Go-tier ChatGPT accounts, which we covered when [ads went live in Europe](/news/chatgpt-ads-live-europe-2026); OpenAI hasn't said whether Sponsored Agents will follow the same free-tier-only rule. If you've been using ChatGPT to think through a home repair before calling anyone, our guide on [diagnosing home problems with AI](/guides/ai-home-repair-help) covers how to do that on your own terms — with or without a sponsored agent in the mix. ## Sources - [Sponsored Agents in ChatGPT Ads](https://help.openai.com/en/articles/20001524-sponsored-agents-in-chatgpt-ads) — OpenAI Help Center - [Angi Among First Brands to Pilot Sponsored Agents in ChatGPT](https://www.globenewswire.com/news-release/2026/09/16/3363209/0/en/angi-among-first-brands-to-pilot-sponsored-agents-in-chatgpt.html) — GlobeNewswire (Angi Inc.) - [OpenAI's Sponsored Agents Turn ChatGPT Into an Ad Platform Where Brands Are the Product](https://finance.yahoo.com/media-advertising/articles/openai-sponsored-agents-turn-chatgpt-155347973.html) — Yahoo Finance --- # ChatGPT Can Now ‘Try On’ Clothes on Your Photo — What to Know First Source: https://aiforregulars.com/news/chatgpt-virtual-try-on-clothes-explained On October 1, 2026, OpenAI added a "Try on" button to ChatGPT's shopping results. You upload a selfie or a full-body photo, and ChatGPT creates a picture of how a piece of clothing or an accessory might look on you. It also added "Favorites", a place to save products you like. **How it works.** When ChatGPT shows you clothes in its shopping results, a Try on button appears next to them. You can also upload a screenshot of an item from an online shop and ask ChatGPT to put it on you. Saved products go into the Library in the app, next to your try-on pictures. The pictures come from OpenAI's newer image model, ChatGPT Images 2.5. Reports say it works on the web and on phones, and OpenAI calls the launch global. We couldn't find which plans include it or whether every country gets it yet, so if you don't see the button, that's likely why. **What changes for you.** You can get a rough idea of colour and style before buying. That's all. The picture is generated, so it may not show the item or you accurately, and it can't tell you whether the size fits. Before you buy, check the seller, the size chart and the return policy. **Your photos.** ChatGPT keeps your reference photo so you don't have to upload it every time. To remove it, open Settings, then Personalization, then Reference photos. Please don't upload photos of other people, such as children or grandchildren, without their say-so. See [what not to tell an AI chatbot](/guides/what-not-to-tell-ai-chatbot) and [viral AI photo trends](/guides/viral-ai-photo-trends) for more on photo privacy. **A sensible caution.** ChatGPT's shopping suggestions have sometimes pointed to fake web shops, so look at the seller before you pay. And only upload your photo inside the official ChatGPT app or website, never to an unknown "try-on" site or app. Our guides on [shopping with AI safely](/guides/ai-shopping-agent-safety) and [how to spot a scam AI app](/guides/is-ai-app-scam) explain what to look for. ## Sources - [ChatGPT can now virtually try on clothes for you](https://techcrunch.com/2026/10/01/chatgpt-can-now-virtually-try-on-clothes-for-you/) — TechCrunch - [ChatGPT adds virtual try-ons for clothes and accessories](https://www.androidauthority.com/chatgpt-virtual-try-on-shopping-feature-3718219/) — Android Authority - [OpenAI's ChatGPT Adds Virtually Try On Clothing & Accessories](https://www.seroundtable.com/chatgpt-virtually-try-on-42215.html) — Search Engine Roundtable --- # ChatGPT Work: What OpenAI's New AI Agent Actually Does for You Source: https://aiforregulars.com/news/chatgpt-work-agent-explained If you use ChatGPT Plus and have noticed a new option that promises to "get the whole job done" instead of just answering one question at a time, that's ChatGPT Work. OpenAI launched it on July 9, 2026, alongside its newest model family, GPT-5.6. **What it actually does.** Normally, you chat with ChatGPT back and forth: you ask something, it answers, you ask a follow-up. ChatGPT Work is different — you describe the *outcome* you want, and it works through the steps on its own, sometimes for a long stretch of time, and comes back with a finished result. That could mean turning a rough list of numbers into a formatted spreadsheet, building a slide deck from a set of notes, drafting a document, or pulling together a short webpage. It can keep working after you close the app, and it will check in with you before doing anything it considers a "consequential" action, like sending an email or making a purchase — you can also pause, redirect, or stop it at any point. **It works by connecting to your other apps.** To do this, ChatGPT Work needs access to things like your email, calendar, or file storage (for example, Gmail, Google Drive, or Slack), depending on which connections you set up. That is the trade-off worth understanding before you turn it on: the more accounts you connect, the more the agent can see and touch on your behalf, so it's worth connecting only the accounts you actually want it working in, and reviewing what permissions you're granting. **Who has it right now.** As of launch, ChatGPT Work is available immediately to Pro, Enterprise, and Edu subscribers on the web and mobile apps, with access rolling out to Plus and Business plans over the following days. In the ChatGPT desktop app, it's available on every plan already, including Free — the Free plan just doesn't have it on the web or mobile app yet. **Should you bother with it?** For most everyday tasks — asking a question, drafting an email, planning a trip — the regular chat window is still simpler and faster. ChatGPT Work is aimed at longer, multi-step jobs: the kind of task you'd normally block out an hour for. If that doesn't describe much of what you use ChatGPT for, there's no need to change anything. If you're curious whether ChatGPT Plus is worth paying for in the first place, our guide on [free vs. paid ChatGPT](/guides/chatgpt-plus-worth-it) breaks down what each tier includes. And if "agent" is a term you keep seeing without a clear explanation, [What Is an AI Agent?](/guides/what-is-ai-agent) covers the plain-English version. --- # Microsoft Is Shutting Down Copilot Group Chats, Podcasts, and Free Deep Research on August 18 Source: https://aiforregulars.com/news/copilot-podcasts-deep-research-retiring If you use the free Microsoft Copilot app, three features you may rely on are going away on **August 18, 2026**. Microsoft is merging its consumer Copilot app with the work-focused Microsoft 365 Copilot into one unified app, and as part of that merger it's cutting Group Chat, Podcasts, and free Deep Research from the free version. **Group Chat becomes 1:1 only.** If you've been using Copilot in a group conversation with other people, that group format disappears on August 18. Microsoft says existing group chats will be turned into individual chats, so your own messages stay visible in your history — but anything other participants shared or created inside the group, like their prompts or generated images, will no longer be accessible to you. If there's something from a group chat you want to keep, save it before the 18th. **Podcasts are shutting down completely.** Copilot's feature for turning documents or web pages into an AI-generated audio "podcast" is being removed, not just changed. After August 18 you won't be able to create new ones or open ones you already made, and any links you shared to a Copilot podcast will stop working. You can still download individual podcasts from your podcast library before the deadline — look for the download option in the podcast's menu. **Free Deep Research is being replaced by a paid tool.** Deep Research — the feature that builds a longer, cited report on a topic you ask about — is being retired for free users. Reports you already generated stay accessible and can still be exported to Word. But generating anything new this way will require **Microsoft 365 Premium**, a paid subscription (currently $19.99/month in the US) that includes a similar tool called Researcher. If you relied on free Deep Research for things like comparing products or planning a trip, that option ends on the 18th unless you upgrade. **What to do before August 18.** Download any podcasts you want to keep, export or save any Deep Research reports you still need, and copy anything important from a group chat that only exists in another participant's messages. None of this requires you to do anything with your account settings — it's a one-time cleanup before the changeover. If you're new to Copilot on Windows, our guide to [Microsoft Copilot on Windows 11](/guides/copilot-windows-guide) covers the basics that aren't affected by this change, and [too many AI subscriptions?](/guides/too-many-ai-subscriptions) can help you decide whether a paid Microsoft 365 Premium plan is worth it for your own use. ## Sources - [Updates to Copilot and the Microsoft Copilot app](https://support.microsoft.com/en-us/microsoft-365-copilot/learning/changes-microsoft-copilot-app) — Microsoft Support - [Microsoft kills off unsuccessful AI features while merging its separate Copilot apps](https://techcrunch.com/2026/08/13/microsoft-kills-off-unsuccessful-ai-features-while-merging-its-separate-copilot-apps/) — TechCrunch - [Microsoft is retiring Copilot Deep Research and Podcasts](https://www.thurrott.com/cloud/microsoft-365/329000/microsoft-is-retiring-copilot-deep-research-and-podcasts) — Thurrott.com --- # EU Pushes Its Toughest AI Rules Back to 2027 — What That Means for You Source: https://aiforregulars.com/news/eu-ai-act-deadline-delayed-2027 If you heard that August 2, 2026 was a big deadline for Europe's AI law, you heard right — but the ending changed. The toughest part of that deadline has been pushed back more than a year, to December 2, 2027. This isn't a rumor or a proposal still being debated: EU lawmakers finished the paperwork in July, and it's already law. Here's what was supposed to start on August 2, 2026. Under the EU AI Act, companies and public bodies that use "high-risk" AI systems — software that helps decide who gets a job interview, a loan, a school place, or how police and border agencies use AI — were meant to have finished risk assessments, documentation, and human oversight by that date. That's the part that's been delayed. The delay came through something called the "Digital Omnibus on AI," a formal amendment to the AI Act. The European Parliament and EU governments approved it in June 2026, it was published in the EU's Official Journal on July 24, and it became binding law on July 27 — a week before the original deadline. The new dates: December 2, 2027 for those standalone high-risk systems, and August 2, 2028 for AI built into regulated products like medical devices or cars. EU officials said the extra time is needed because the technical standards and testing tools companies need to comply weren't ready. You may be wondering how this squares with what you might have already noticed: chatbots on company websites suddenly labeling themselves as AI, or photos and videos flagged as AI-generated. That's a separate, earlier part of the same law — the transparency rules — and those did start on schedule on August 2, 2026, delay or no delay. Our [earlier article on that rule](/news/eu-ai-act-what-changes-for-you) covers it. This new delay only affects the stricter, behind-the-scenes rules for high-risk AI systems, not the disclosures you can actually see. So what changes for you personally, today? In practical terms, nothing appears and nothing disappears. No new button, warning, or disclosure shows up in your apps because of this news. What changed is a company's clock: businesses using AI to screen job applicants or score loan applications now have well over a year longer before the strictest EU oversight rules apply to them. If you want to understand what protections the AI Act does give you, and when they actually arrive, our guide on [your rights under the EU AI Act](/guides/eu-ai-act-your-rights) walks through it in plain language. ## Sources - [EU Digital Omnibus on AI Enters Into Force](https://www.hunton.com/privacy-and-cybersecurity-law-blog/eu-digital-omnibus-on-ai-enters-into-force) — Hunton Andrews Kurth - [Digital Omnibus on AI](https://www.europarl.europa.eu/legislative-train/package-digital-package/file-digital-omnibus-on-ai) — European Parliament, Legislative Train Schedule - [EU AI Act Omnibus Agreement — Postponed High-Risk Deadlines and Other Key Changes](https://www.gibsondunn.com/eu-ai-act-omnibus-agreement-postponed-high-risk-deadlines-and-other-key-changes/) — Gibson Dunn --- # New EU Rule: AI Chatbots Must Now Tell You They're Not Human Source: https://aiforregulars.com/news/eu-ai-act-what-changes-for-you Starting August 2, 2026, a new EU rule takes effect: any chatbot or AI assistant operating in the European Union must clearly tell you it's an AI system — not a human. At the same time, AI-generated content such as deepfakes, manipulated images, and synthetic audio or video must carry a label saying it was created by AI. This is part of the EU AI Act's transparency obligations, which are now coming into force for consumers. Some business-side requirements under the Act have been delayed, but the rules that directly affect what you see and hear as a regular user are moving ahead on schedule, according to coverage from InsidePrivacy (May 2026). **What you'll notice** If you're in Europe — or using a service that serves European users — you should start seeing clearer "I am an AI" disclosures when you open a chat on a company website, and labels on AI-generated images or videos. The goal is to make it harder for companies or bad actors to pass AI off as a real person without your knowledge. Customer service bots are a common place to watch. Companies have used AI chatbots for years without always making it obvious. That changes now. **The limits of a new law** Laws set rules for legitimate companies. Scammers and fraudsters operate outside the law — and will keep trying to impersonate real people using AI voices and fake images regardless. If you want to learn how to spot AI-generated content yourself, our guide on [how to tell if you're talking to AI](/guides/how-to-tell-if-text-is-ai) walks you through practical signals to look for. For scams specifically, see our guide on [fake customer service bots](/guides/fake-customer-service-scam). Enforcement also takes time. A law passed today doesn't mean every company complies tomorrow. Watch for the disclosures, but stay alert regardless. **Coming next: stricter protections** From December 2, 2026, the EU AI Act adds new prohibitions on non-consensual AI-generated intimate images — a step targeting deepfake abuse that has affected millions of people, mostly women. If you're outside the EU, similar rules are being debated in many countries. The EU tends to set the pace on tech regulation globally, and companies often apply new standards beyond their legal minimum. --- # The EU Just Classified ChatGPT as a Search Engine — Here's What That Means Source: https://aiforregulars.com/news/eu-chatgpt-search-engine-designation On August 31, 2026, the European Commission gave ChatGPT a label it has never given an AI chatbot before: a "Very Large Online Search Engine," or VLOSE, under the EU's Digital Services Act (DSA). It's the same legal category Google Search falls into — and it comes with the strictest oversight rules the EU has. **Why ChatGPT got this label.** The Digital Services Act already regulates the biggest online platforms and search engines, requiring them to manage risks and be more transparent once they reach 45 million average monthly users in the EU. OpenAI disclosed that ChatGPT's search feature — the part that looks things up on the live web to answer your questions — had about 159 million average monthly users in the EU over the six months ending March 2026, more than three times that threshold. Because ChatGPT can search the web and return results the way Google does, the Commission decided it counts as a search engine under the law, not just a chatbot. Reddit and Roblox were designated the same day as "Very Large Online Platforms" (VLOPs), a related category, after reporting 57.2 million and roughly 48 million EU users. **What this actually requires OpenAI to do.** Companies in this category have to run yearly risk assessments looking at things like illegal content, harm to minors, effects on people's mental and physical health, and risks to elections and public safety. They also have to submit to independent outside audits, give regulators and vetted researchers access to relevant data, publish transparency reports twice a year, and set up a clear channel for people to report illegal content so it actually gets acted on. Companies that don't comply can be fined up to 6% of their global annual revenue — the DSA has already levied close to €870 million in fines against various companies, including a €550 million penalty against AliExpress. **When it kicks in.** OpenAI now has four months from the Commission's formal notification to come into compliance — the Commission's own announcement points to a deadline around January 2027, though some outlets reporting on the story have described it as arriving anywhere from November 2026 to January 2027, since the exact notification date hasn't been made public. Either way, expect the practical changes to show up gradually over the next few months rather than overnight. **What changes for you, specifically.** Nothing about how ChatGPT answers your questions changes today — this is a regulatory step, not a new feature. But over the coming months, you can expect a few concrete things to show up for EU users: a published archive of any ads shown inside ChatGPT, regular public reports on how OpenAI is handling risks like misinformation or harm to minors, and — most usefully — an official way to flag content that seems illegal or harmful and expect OpenAI to actually respond to it. If you're used to reporting a scam message or harmful post on a big platform like Facebook or Google and having somewhere for that report to go, this brings ChatGPT closer to that same standard. This is a different law from the **EU AI Act**, which is a separate set of rules phasing in through 2026 and mostly covers how AI systems are built and disclosed (see our guide on [what the EU AI Act changes for you](/guides/eu-ai-act-your-rights)). The Digital Services Act, used here, is the same law that already governs Google, Facebook, and other large platforms — think of this less as a new AI-specific rule and more as the EU deciding that a wildly popular AI chatbot now has to play by the same safety rules as any other major search tool. ## Sources - [Commission designates ChatGPT, Reddit, Roblox under Digital Services Act](https://digital-strategy.ec.europa.eu/en/news/commission-designates-chatgpt-reddit-roblox-under-digital-services-act) — European Commission - [EU places ChatGPT, Reddit and Roblox under strictest digital safety rules](https://www.euronews.com/next/2026/08/31/eu-places-chatgpt-reddit-and-roblox-under-strictest-digital-safety-rules) — Euronews - [EU Now Regulates ChatGPT as a Very Large Online Search Engine](https://winbuzzer.com/2026/09/01/eu-designates-chatgpt-tougher-dsa-search-engine-scrutiny-xcxwbn/) — Winbuzzer --- # Meta's Own Oversight Board Says AI Account Bans Lack Due Process — Here's How to Appeal Source: https://aiforregulars.com/news/facebook-instagram-ai-account-ban-appeal If Facebook or Instagram ever locks you out of an account you've had for years — with no clear explanation and no person to call — you are not imagining things, and you are not alone. **What Meta's own watchdog found.** On June 5, 2026, the Oversight Board — the independent body Meta itself set up to review its content moderation — published its first ruling on an account disablement case. It upheld Meta's decision to permanently disable the specific Instagram account in question, which had posted threats and slurs. But the board used the case to call out much bigger problems: it found Meta's system for sorting violations into "severe" and "egregious" categories, which determines whether an account gets a warning or an instant permanent ban, to be "unclear and often contradictory." The board's most pointed recommendation was that people should be told whether and how AI was involved in reviewing their account and deciding any penalty against it — something Meta currently doesn't disclose. **Thousands say they were caught up in it.** A petition from the Canadian nonprofit People Over Platforms, titled "Hold Meta Accountable for Wrongful Account Disabling," has drawn tens of thousands of signatures over 2026 from people who say Facebook, Instagram, or Threads accounts — including long-running small business pages — were disabled without warning, often citing "severe" categories like child exploitation, with no human reachable to explain why. Local news outlets including CBS affiliates in Baltimore and Philadelphia have covered similar cases. Meta disputes that AI is the problem: a company spokesperson told Boston 25 News that its current moderation tools make 13% fewer mistakes than human reviewers did and catch 10% more genuine violations — but didn't dispute that appeals can be slow or unclear. **If it happens to you, here's what to actually do.** Log into the affected account (not a new one) and follow the "Request a review" or "Disagree with decision" prompt that appears on the disabled-account screen — for Instagram, that's usually under "Review details" in the app. You have **180 days** from the date your account was disabled to file this appeal; after that, Meta treats the account as permanently gone. Gather what you'll need before you start, since the form can't be saved partway through: the email or phone number tied to the account, and a government ID if asked. Some appeals now require a short video selfie, where you turn your head in different directions on camera so Meta can match your face to your profile photos — if that step fails, you can usually retry it. For a business Page, check Meta Business Suite's account status or support tools rather than the personal account flow. Meta says most reviews are completed within about 48 hours, though there's no guarantee, and a meaningful share of first appeals get denied — a second appeal is generally looked at by a more senior reviewer, so it's worth filing again rather than giving up after one rejection. --- # FBI Warns: Fake Police and Officials Now Use AI and Video Calls Source: https://aiforregulars.com/news/fbi-ai-fake-police-video-call-warning On September 17, 2026, the FBI issued a public warning about scammers who pretend to be police officers or government officials. Between January 2025 and July 2026, its Internet Crime Complaint Center received nearly 61,000 complaints about this kind of fraud, with losses of more than $1.6 billion. The FBI says criminals now use technology such as artificial intelligence, including video calls, to make the impersonation more convincing. **The one rule that protects you.** Police and government agencies will never phone or text you to demand payment, according to the FBI. If a caller insists you must pay right now to avoid arrest, a fine or a lost license, it is a scam, however official they look or sound. Hang up. **What the stories sound like.** The FBI's figures show a few common scripts: - **Missed jury duty or a court date.** About 6,800 complaints and nearly $36 million lost. The caller says a warrant is out for you unless you pay a "fine." - **A problem with a medical license.** About 3,300 complaints and more than $37 million lost, aimed at doctors and other medical professionals. - **A "foreign official."** About 1,800 complaints and more than $140 million lost. This is the most elaborate version, and reports say it often targets international students and immigrant communities. According to The Register, some callers appear on video in a uniform or in a set made to look like a government office. - **A license renewal.** About 500 complaints and a smaller $348,000 lost. **Is the caller a deepfake?** Not necessarily. The FBI says AI is *used to support* the scheme, but the real trick is pressure and fear, not perfect technology. A uniform or an official-looking room on screen proves nothing. Judge the request, not the picture. **What to do if someone calls** 1. Don't pay, don't give personal details, and don't stay on the line. Never pay with prepaid cards, cryptocurrency, bank wires or cash at crypto kiosks. Real agencies don't ask for these. 2. Ignore the phone number or website the caller gives you. Look up the agency's official number yourself and call back. 3. Don't be rushed or told to keep it secret. Secrecy and urgency are the scammer's tools. Talk to a family member or friend. 4. Check web addresses carefully. Fake sites copy real names with small changes. **If you already paid or shared details.** Contact your bank or card company right away, tell your local police, and in the United States file a report at [ic3.gov](https://www.ic3.gov). Include the caller's details, how you paid and any messages. Outside the US, report to your local police. For more on spotting fake voices and faces, see [how to spot an AI voice on a phone call](/guides/detect-ai-voice-phone-call) and [how to spot a deepfake video](/guides/how-to-spot-deepfake-video). If the caller claims to be a relative, read [the grandparent scam](/guides/grandparent-scam-ai). ## Sources - [Law Enforcement Impersonation Fraud Alert (I-091726-PSA)](https://www.ic3.gov/PSA/2026/PSA260917) — FBI Internet Crime Complaint Center - [FBI: fake cop and government impersonation scams cost victims $1.6B](https://www.theregister.com/cyber-crime/2026/09/18/fbi-fake-cop-and-government-impersonation-scams-cost-victims-16b/5297499) — The Register --- # Google's Gemini App Just Launched on Windows — Here's What It Does Source: https://aiforregulars.com/news/gemini-app-windows-launch On September 10, 2026, Google released a dedicated Gemini app for Windows — a free, official download for both Windows 10 and Windows 11, available at **gemini.google/desktop**. It follows a similar app Google launched for Mac computers back in April 2026. **What makes it different from using Gemini in a browser.** Instead of opening a browser tab, you press **Alt+Space** anywhere on your PC, and Gemini pops up as an overlay on top of whatever you're already doing — a document, an email, a spreadsheet. You don't have to switch windows or lose your place to ask it a question or hand it a task. **What it can see and connect to.** Google's own announcement confirms the app connects to Gmail and Google Drive, so you can ask it things about your emails or files without digging through them yourself. Coverage from other outlets also mentions Google Docs and Calendar as connected apps. **What it can do beyond chatting.** The app includes a feature called "Gemini Spark" for longer, multi-step research or planning tasks — the kind of thing that would normally take you several searches and tabs to piece together. It can also generate images (using a tool Google calls "Nano Banana") and video ("Gemini Omni") directly inside the app, without needing a separate website. **What it costs.** The basic app is free to download and use. Some of the more advanced features require a paid Google AI subscription (Pro or Ultra) — if you already pay for one, it carries over to the desktop app automatically. **Where it works.** The basic app is available in every country and language where Gemini already works. The Gemini Spark research feature is the exception — it isn't available yet in the EU, the UK, Switzerland, or Nigeria, even for Pro or Ultra subscribers there. **A safety note.** Because this is a genuinely new, popular download, expect copycat sites and fake installers to show up in search results and ads. Only download the Gemini app from **gemini.google/desktop** — never from a link in an email, a pop-up ad, or a third-party download site, no matter how official it looks. ## Sources - [Gemini app now on Windows](https://blog.google/innovation-and-ai/products/gemini-app/gemini-app-now-on-windows/) — Google - [Google launches a dedicated Gemini app for Windows](https://9to5google.com/2026/09/10/gemini-windows-app/) — 9to5Google - [Google launches a dedicated Gemini app for Windows with Alt+Space shortcut](https://chromeunboxed.com/google-launches-a-dedicated-gemini-app-for-windows-with-alt-space-shortcut/) — Chrome Unboxed --- # Google Gemini Can Now Clone Your Face and Voice Into AI Videos Source: https://aiforregulars.com/news/gemini-avatar-ai-clone-yourself Google has been rolling out a feature called Avatar inside the Gemini app that lets you create an AI clone of your own face and voice, then use it to star in AI-generated videos. It started reaching paid subscribers more broadly in early June 2026, and if you use Gemini, you'll likely run into it sooner or later — either your own version or, eventually, a friend's. **How it works.** Avatar creation starts inside the Gemini app: tap "Add Files" in the message box and choose Avatar, then scan the QR code that appears using your phone. From there, you'll be asked to look into the camera, turn your head a few different ways, and read some numbers aloud so the system can map your face and voice. The whole thing takes a few minutes. Afterward, you can bring your avatar into a conversation by typing "@me" or your name, and Gemini's video-generation model — called Omni — can produce short videos where you appear to be speaking, using your actual face and voice. **Who can use it.** Avatar requires a paid Google AI plan (Plus, Pro, or Ultra) — it isn't available on the free tier. You also need to be 18 or older, and you have to be the one physically present during setup; you can't create an avatar of someone else from a photo. **The safety guardrail worth knowing about.** Every video Gemini generates with Avatar carries an invisible SynthID watermark baked into the file. It doesn't change what the video looks like, but it lets tools inside Chrome and Google Search confirm a clip was AI-generated — useful if you ever need to check whether a video that looks like you (or someone else) is real. **What's not here yet.** Some device teardowns of the Gemini app in early July 2026 turned up unreleased text referencing avatar sharing — the ability to send your AI avatar to someone else so they could use it to generate their own videos. That feature hasn't launched and Google hasn't confirmed a date; app teardowns regularly surface things that change or never ship, so treat it as a preview of a possible direction, not an announcement. For now, the practical takeaway is simpler: if a video shows up that looks and sounds exactly like someone you know, "AI clone" tools like this are a big part of why that's now possible with nothing more than a phone camera and a few minutes. Our guide on [how to tell if a video call is a deepfake](/guides/verify-video-call-deepfake) covers ways to check when it matters. --- # Gemini Is Coming to Every Student in Google Classroom — What Parents Should Know Source: https://aiforregulars.com/news/gemini-classroom-expands-all-ages If your child's school uses Google Classroom, you may be about to see a new "Gemini" tab appear inside it — even if your child is nowhere near 18. Google announced on August 10, 2026 that Gemini in Classroom is expanding to students of **all ages**, from elementary school through college. Until now, it was only available to higher-education students who were 18 or older. **What's actually changing.** The rollout starts on the web this week and reaches phones and tablets on August 17. Once it's live, eligible students can open a Gemini tab inside Classroom and turn their own class materials — a reading, a set of notes, a study guide the teacher posted — into flashcards or practice quizzes tailored to that specific assignment. Google calls these "contextualized starter prompts": instead of a blank chat box, students pick a class and an assignment first, and Gemini builds the study aid around it. **Your school has to turn it on — it doesn't switch on by itself.** This is the part worth knowing before you worry: students only get access if their school's IT administrator has already enabled Gemini in Classroom, Gemini, and Gemini Notebook for that account. If your school's admin hasn't turned those on, nothing changes for your child on August 10 or 17 — the tab simply won't appear. Schools can also set up a separate group for students under 18 and switch access off for just that group, independent of older students or staff. **What happens to your child's data.** Google's Workspace for Education terms — which cover Classroom, the Gemini app, and NotebookLM as "Core Services" — state that student data in these tools is not used to train Google's AI models, is not reviewed by people to improve those models, and is not used to build advertising profiles. That protection isn't new to this rollout; it's the same enterprise-grade data agreement that has applied to Google's education tools since 2025, and it extends to how Gemini in Classroom is being rolled out now. **What to actually do about it.** You don't need to do anything for this rollout to reach your child — it depends entirely on a setting your school's IT department controls, not one in your home. If you want to know whether it's turned on, the most direct step is to ask your child's school or district technology office. If you'd rather understand what a Gemini tab your child can already see does day to day, our guide to [Gemini in Gmail and Google Docs](/guides/gemini-google-docs-gmail) covers the same assistant in a non-school setting, and [parental controls across ChatGPT, Gemini, and Copilot](/guides/ai-parental-controls-all-apps) walks through the settings you do control at home. ## Sources - [Gemini in Google Classroom is expanding to users of all ages](https://workspaceupdates.googleblog.com/2026/08/gemini-in-google-classroom-is-expanding-to-users-of-all-ages-with-contextualized-Gemini-starter-prompts-for-students.html) — Google Workspace Updates - [Google is officially bringing Gemini in Classroom to students of all ages](https://chromeunboxed.com/google-is-officially-bringing-gemini-in-classroom-to-students-of-all-ages/) — Chrome Unboxed - [Privacy & Security FAQs](https://edu.google.com/intl/ALL_us/our-values/privacy-security/frequently-asked-questions/) — Google for Education --- # Free Gemini Drops to One Model on October 9 — What Changes and Do You Have to Pay? Source: https://aiforregulars.com/news/gemini-free-plan-flash-lite-oct-9 Gemini is still free. But from **October 9, 2026**, tomorrow, Google is narrowing what you get without a paid plan. If you use Gemini on a personal Google account and don't pay for a Google AI plan, you will lose the "Flash" and "Pro" models and be left with the lightest one, "Flash-Lite". **Do you have to pay?** No. You can keep using Gemini for free and nothing needs to be done on your side. Flash-Lite is simply the model you'll be using. Google hasn't published exact numbers for how many questions you get, and neither have the news sites that covered the change, so we won't guess. **What changes, plan by plan** - **No paid plan (free):** Flash-Lite only. Flash and Pro stop being available on October 9. - **AI Plus ($4.99 a month):** you keep Flash-Lite and Flash but lose Pro. Google says AI Plus subscribers will get an email with the date that applies to them, so watch your inbox. - **AI Pro ($19.99 a month):** you lose nothing, and you gain "Deep Think", a setting for harder problems that was previously limited to the most expensive plans. - **Top plans (AI Ultra):** all three models stay available. **A new setting: effort level.** Google's help page also describes a choice of "low", "medium" or "high" effort for each model you can use. Higher effort gives more thorough answers but uses up more of your allowance, so for everyday questions "low" or "medium" will usually do. Reports say this arrives in the Gemini app during October, and Google hasn't given an exact date. **What you might notice.** Answers from Flash-Lite may be shorter or less detailed on tricky questions, such as long documents or multi-step problems. For simple things like drafting a message, explaining a word or planning a meal, you may notice little difference. If a free answer seems thin, try asking again with more detail. Our [guide to writing prompts](/guides/how-to-write-prompts) shows how. **What we couldn't confirm.** Google's help page says the changes apply when you use a personal account. It doesn't say anything about work or school accounts, and none of the reports mention differences between countries. If you use Gemini through work or school, or live outside the US, check the model menu in your Gemini app after October 9. **One safety tip.** Scammers like a moment of change. Google won't ask you for card details by email or text to "keep" your models. If you want a paid plan, go to the Gemini app or your Google account yourself rather than tapping a link in a message. See [what not to tell an AI chatbot](/guides/what-not-to-tell-ai-chatbot) and [how the free chatbots compare](/guides/free-ai-chatbots-compared) if you want to look at alternatives. ## Sources - [Changes to Gemini model access and limits](https://support.google.com/gemini/answer/17004136) — Google Gemini Help - [Gemini limiting what models free & AI Plus users can access, more](https://9to5google.com/2026/10/03/gemini-model-limits-oct-26/) — 9to5Google - [Google Caps Free Gemini at Flash-Lite Starting October 9](https://aiweekly.co/alerts/google-caps-free-gemini-at-flash-lite-starting-october-9) — AI Weekly --- # Google Cuts Gemini's Model Choices for Free and AI Plus Users on October 9 Source: https://aiforregulars.com/news/gemini-free-plus-model-access-cut-october-9 From October 9, 2026, Google is narrowing which Gemini models you can pick on a personal account. Google announced it on its own help page for the Gemini app. **What changes, by plan:** - **Free:** only the lightest model, called Flash-Lite. The more capable Flash and Pro models are no longer included. - **Google AI Plus** (the lower-priced paid plan): you lose Pro and keep Flash-Lite and Flash. Google says AI Plus subscribers get an email explaining the date the change reaches their account. - **Google AI Pro and AI Ultra:** you keep Flash-Lite, Flash and Pro. Google's page also says each model has low, medium and high effort levels, and higher levels use up your limit faster. Google's page doesn't say why it made the change, and it describes personal accounts only. **What it means for you.** If you use free Gemini, you may notice shorter or less thorough answers, especially on harder questions such as a confusing form or a long letter. That isn't a fault in the app; it's the lighter model. For anything important, such as health, money or legal paperwork, double-check the answer with an official source or a professional. Our guide to [what not to tell an AI chatbot](/guides/what-not-to-tell-ai-chatbot) still applies. **Should you pay?** Don't upgrade in a hurry. If you only subscribed to AI Plus to get Pro, that reason is gone, so check what you actually use before the next renewal. You can also compare options: free ChatGPT now offers [unlimited text chats](/news/chatgpt-free-unlimited-text-chats), and our [free chatbots comparison](/guides/free-ai-chatbots-compared) and [paid plans comparison](/guides/ai-paid-plans-compared) cover the rest. **One safety tip.** News like this attracts scammers. Only change or buy a plan inside the official Gemini app or on gemini.google.com, never through a link in an email, text or ad. Be wary of any message that pressures you to "act now" or asks for card or login details to "keep your access". Not sure the rollout reaches you on the same day, or in your country? Google's page doesn't say, so check the help page or your own app. ## Sources - [Gemini model access changes](https://support.google.com/gemini/answer/17004136) — Google Gemini Help - [Google is limiting free and AI Plus users' access to Gemini models](https://www.androidauthority.com/google-gemini-free-ai-plus-usage-limits-3719012/) — Android Authority - [Google limits free and budget users' access to newer Gemini models](https://mobilesyrup.com/2026/10/05/google-limit-free-budget-user-access-newer-gemini-models) — MobileSyrup --- # Google's Voice Search Just Got a Big Upgrade — Here's What Changed Source: https://aiforregulars.com/news/gemini-search-live-voice-upgrade On September 15, 2026, Google released two new voice AI models — **Gemini 3.8 Live** and **Gemini 3.8 Live Extended Thinking** — and put them to work inside products many people already use for free: **Search Live** (the voice-and-camera mode in the Google app) and the **Gemini Live** assistant. **Where you'll actually meet it.** Gemini 3.8 Live is the version now running Search Live — the feature you get to by tapping the "Live" icon under the search box in the Google app, then talking or pointing your camera at something instead of typing. The more powerful Extended Thinking version is going into the Gemini Live app for everyone, plus Gmail and Keep for anyone with a paid Google AI subscription, and Docs specifically for people on the pricier Google AI Pro or Ultra plans. **What's actually new when you talk to it.** Two things stand out for everyday use. First, the model can recognize and switch between 97 languages in the middle of a single conversation — so if you start a question in English and slip into Czech, German, or Spanish partway through, it follows along instead of getting confused. Second, it can look at what your camera sees in close to real time and keep searching or working in the background while it keeps talking to you, instead of going quiet while it "thinks." **What Extended Thinking adds.** For harder, multi-step questions — planning something with several steps, or comparing a few options — the Extended Thinking model can reason through the problem and talk at the same time, rather than pausing for a long silence before answering. **What it costs.** Search Live's new voice model is free to anyone using the Google app, the same as before, and so is Extended Thinking inside the Gemini Live app itself. Getting it in Gmail and Keep needs any paid Google AI subscription; getting it in Docs needs the pricier Pro or Ultra tier specifically — if you already pay for one of those, there's nothing extra to turn on. **Worth knowing.** Google is rolling this out gradually, as it usually does with new AI features, so it may take a little longer to reach some countries or languages than others — if the Live icon still behaves like before, that's likely why, not a sign anything is broken on your end. And as always, a genuinely new, talked-about AI feature is exactly the kind of thing scammers copy — download or update the Google app only through the Play Store or App Store, never through a link in an email or ad promising an "exclusive new Gemini voice." If you're still finding your way around Gemini on your phone, our [guide to Gemini on Android](/guides/gemini-android-guide) covers the basics, and [AI Search vs Google](/guides/ai-search-vs-google) helps you decide when voice search is the right tool versus typing a normal search. ## Sources - [Introducing Gemini 3.8 Live and 3.8 Live Extended Thinking](https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-8-live-gemini-3-8-live-extended-thinking/) — Google - [Google Launches Gemini 3.8 Live and Extended Thinking Voice Models](https://www.unite.ai/google-launches-gemini-3-8-live-and-extended-thinking-voice-models/) — Unite.AI - [Google Announces Gemini 3.8 Live and 3.8 Live Extended Thinking](https://www.thurrott.com/a-i/google-gemini-a-i/341685/google-announces-gemini-3-8-live-and-3-8-live-extended-thinking) — Thurrott --- # Chrome Now Has a Built-In Gemini Side Panel — Here's What It Does Source: https://aiforregulars.com/news/gemini-side-panel-chrome-explained If a new icon has appeared in the top-right corner of your Chrome browser and clicking it opens an AI chat panel down the side of the screen, that's Gemini in Chrome. Google has been rolling this out gradually since late January 2026, and it keeps reaching more people — most recently, users in the UK started seeing it in mid-July. **What it actually does.** Gemini used to open in a floating window over the page. Now it lives in a side panel that stays open next to whatever tab you're viewing. Click the "Ask Gemini" icon and it can summarize the page you're on, answer questions about it, or — if you choose to share more than one tab — compare information across up to 10 open tabs at once, which is handy for things like comparing prices or reading several reviews side by side. It also connects to other Google apps: it can draft an email and hand it to Gmail as a ready-to-send compose window, pull up relevant results in Maps, check your Calendar, or pull key points from a YouTube video's transcript. People on a paid Google AI Pro or Ultra plan also get "Auto Browse," a feature that lets Gemini carry out a multi-step task on your behalf across tabs, such as researching several options and filling in details for you. **What it can see.** When you share a tab with Gemini — the tab you're currently viewing is shared by default once you open the panel — it reads that page's content and web address, and does the same for any extra tabs you add. Google says that data is logged temporarily to your account, but doesn't show up in your regular Gemini Apps Activity, and isn't used to train its models without permission. You can stop sharing a specific tab at any time by clicking the "X" next to its name in the panel, and you're not required to share anything to keep using Chrome normally. **How to turn it off.** If you'd rather not have it around, open Chrome's settings, search for "Gemini," and you'll find a "Gemini in Chrome" section — usually under an "AI innovations" heading — where you can toggle the feature off entirely, or leave it on but turn off individual pieces like page-content sharing. Turning the whole thing off removes both the icon and the side panel; you can always switch it back on the same way later. **The bottom line.** This isn't a scam or a bug — it's Google's real, ongoing rollout of AI features into Chrome, and it's optional. If you find it useful for comparing tabs or drafting emails, it's worth trying. If you'd rather your browser stayed as it was, turning it off takes under a minute. For a closer look at what Gemini can do elsewhere in Google's apps, see our guide to [Gemini in Google Docs and Gmail](/guides/gemini-google-docs-gmail). --- # Google Is Replacing Gemini 'Gems' With 'Skills' — Free for Everyone 18+ Source: https://aiforregulars.com/news/gemini-skills-replace-gems-free On September 30, 2026, Google began rolling out "Skills" in Gemini. A Skill is a set of saved instructions for something you do again and again, such as writing in your own style or planning meals around your dietary rules. It replaces what Google called a "Gem". Instead of opening a separate Gem, you type "/" in a normal Gemini chat and pick the Skill you want. **Do you get it?** Yes, if you're 18 or older and use a personal Google account. Skills don't need a paid Google AI plan. The rollout is gradual, so some free accounts still only show Gems for now. If yours does, that's normal and you don't need to do anything. **What happens to your Gems?** Gemini converts existing Gems into Skills automatically, so nothing you built is lost. On personal Google accounts, Gems stop working in November 2026. Work and business Google Workspace accounts follow in March 2027, and school accounts in June 2027. Google says sharing a Skill through a link and adding files from Google Drive will arrive "over the coming weeks". **What changes in practice.** The idea is the same: you write your instructions once and reuse them. You just reach them with "/" in the chat box instead of a side menu. Because the rollout is phased, the buttons and menus may look different on your account than in our guides. If something doesn't match, look for the "/" option or a section called Skills. **If you've never used Gems.** Now is a good time to try. Pick one thing you keep re-explaining, like your writing style or your family's food rules, and save it as a Skill. Our [Gems guide](/guides/gemini-gems-guide) walks through the same idea step by step (the steps will change slightly as Skills roll out), and [how to write better prompts](/guides/how-to-write-prompts) helps you word the instructions. **One safety tip.** Saved instructions are stored in your account and can be shared, so never put passwords, card numbers or ID numbers in them. Google won't email or text you asking to "verify" your Gems or Skills, so treat any message like that as a scam. See [what not to tell an AI chatbot](/guides/what-not-to-tell-ai-chatbot). ## Sources - [Google details Gems to skills migration, including free access](https://9to5google.com/2026/09/30/gemini-skills-free/) — 9to5Google - [Gemini Skills expand to free accounts](https://www.androidauthority.com/gemini-skills-expand-free-accounts-3716627/) — Android Authority - [Let skills in Gemini tackle your most repetitive tasks](https://blog.google/products-and-platforms/products/gemini/automate-tasks-with-skills/) — Google --- # Google Made Gemini's Visible AI Watermark Optional — Here's Why That Matters Source: https://aiforregulars.com/news/gemini-watermark-toggle-explained If you've gotten into the habit of looking for the small sparkle icon in the corner of an image to tell whether it came from Google's AI, that habit just got less reliable. Google has added a setting that lets people turn that visible mark off. **What changed.** Gemini (and Google's video/music tool, Flow) now has a "Media watermark" switch in settings — On or Off. With it off, images made with Gemini's image tool, videos made with its video tool, and music made with its music tool no longer carry the small visible AI mark in the corner. You'll find it in the web version of Gemini by clicking the gear icon in the bottom left, then "Media watermark." Google has said it plans to bring the same option to Search. **What did NOT change.** Turning the switch off only removes the mark you can *see*. Every file still carries an invisible digital watermark (called SynthID) and metadata (called C2PA) embedded in the file itself, regardless of the setting. Tools built to check for that hidden marker — like Google Search's "About this image" feature — can still identify the content as AI-generated even with the visible mark switched off. **One exception.** In places where the law requires a visible AI label — China is the main example — Gemini will keep the visible watermark on no matter what the user's setting says. **Why this matters for you.** Until now, "no sparkle in the corner" was a reasonable, if imperfect, sign that an image probably wasn't from Gemini. That's no longer true — anyone can now choose to hide it. If you're trying to judge whether a photo, video, or song you're looking at is real, a missing visible watermark tells you nothing on its own. Our guide on [how to spot AI-generated photos](/guides/spot-ai-generated-photos) covers what still works: a reverse image search and checking for that invisible Content Credentials (C2PA) label at a site like verify.contentauthenticity.org, both of which look past whatever visible mark a creator chose to show or hide. ## Sources - [Google will now allow users to remove visible watermark from its AI generations](https://techcrunch.com/2026/08/14/google-will-now-allow-users-to-remove-visible-watermark-from-its-ai-generations/) — TechCrunch - [Google now lets you nix visible Gemini image watermarks](https://www.pcworld.com/article/3213409/google-now-lets-you-nix-visible-gemini-image-watermarks.html) — PCWorld - [Gemini now lets you turn off the visible watermark on your AI creations](https://www.techradar.com/ai-platforms-assistants/gemini/gemini-now-lets-you-turn-off-the-visible-watermark-on-your-ai-creations-heres-how-to-do-it-and-how-your-content-is-still-flagged-as-ai) — TechRadar - [Google Gemini's visible watermarks are now optional](https://dataconomy.com/2026/08/17/google-gemini-visible-watermarks-optional/) — Dataconomy --- # Google Is Giving College Students a Free Year of Its Paid AI Plan Source: https://aiforregulars.com/news/google-ai-pro-free-year-college-students If you have a child in college — or you're taking classes yourself — there's a deal worth knowing about. Google is offering verified college students a full year of its paid AI plan for free. In the US, that means **Google AI Pro**, which normally costs about $20 a month (roughly $200 a year). Students in more than 140 other countries get **Google AI Plus**, a lower-priced tier. Both include higher usage limits on the Gemini app, Gemini built into Gmail and Google Docs, and extra Google storage. One change to keep in mind: from October 9, 2026, Google AI Plus no longer includes the Pro model (it keeps Flash-Lite and Flash), while AI Pro keeps all three ([what changed](/news/gemini-free-plus-model-access-cut-october-9)). To sign up, a student needs a personal Google account (not a school-issued one) and must verify their student status. Google also asks for a payment card at sign-up, even though the first year is free. The offer can be claimed through December 31, 2026. Here's the part to pay attention to: **after 12 months, the plan automatically switches to a paid subscription** unless the student cancels first. This is a normal way free trials work, but it's easy to forget about a year from now. If your student signs up, it's worth adding a reminder a few weeks before the year is up, so nobody gets an unexpected charge. If you want to understand what Gemini actually does day to day, our guide to [Gemini on Android](/guides/gemini-android-guide) and our comparison of [ChatGPT vs. Gemini](/guides/chatgpt-vs-gemini) are good places to start. For more tools worth trying this school year, see our guide to [the best AI study tools for students](/guides/back-to-school-ai-study-tools-2026). ## Sources - [College students get 12 months of Google AI free](https://blog.google/innovation-and-ai/products/gemini-app/student-offer-google-ai/) — Google - [Google is offering college students a year of its AI Pro plan for free](https://www.engadget.com/2240161/google-is-offering-college-students-a-year-of-ai-pro-for-free/) — Engadget - [Google is giving students a free year of Gemini Pro — plus, new AI study features](https://www.tomsguide.com/ai/google-gemini/google-is-giving-students-a-free-year-of-gemini-pro-plus-new-ai-study-features) — Tom's Guide --- # Google Assistant Is Leaving Android for Good — Gemini Takes Over, No Way Back Source: https://aiforregulars.com/news/google-assistant-gemini-mandatory-switch If you have an Android phone and still say "Hey Google" out of habit, get ready for a different voice to answer. Starting September 4, 2026, Google began switching Android phones, tablets, Wear OS watches, and paired earbuds or Android Auto over to Gemini as the one and only assistant — and this time, there is no setting to switch back to the old Google Assistant. **What's actually changing.** Google Assistant, the voice helper that has answered "Hey Google" on Android phones for years, is being retired on personal devices in favor of Gemini, Google's newer AI assistant. This has been coming for a while — Google announced the plan and then delayed it once after complaints that Gemini felt slower and less predictable than the assistant it was replacing — but this week the mandatory switch actually started rolling out. **Why "mandatory" matters here.** Until now, Android let you choose: keep the classic Google Assistant, or opt into Gemini yourself. That choice is disappearing. In an email sent directly to users, Google was blunt about it: "You will no longer be able to use — or switch back to — the legacy Google Assistant on your Android phone, tablet, or paired accessories." Once your device gets the update, saying "Hey Google" or holding down the power button brings up Gemini instead — permanently. **It won't happen to everyone at once.** Google is rolling this out gradually, over a period of weeks rather than switching every device on the same day. If your phone hasn't changed yet, that doesn't mean it has been skipped — it likely just has not received the update yet. Google has not published an exact date for when the rollout reaches every device, so expect it to arrive at different times for different people. **What stays the same, for now.** Two places keep the old Google Assistant, at least for the moment: cars with Android Automotive built directly into the dashboard (not your phone mirrored through Android Auto), and Google's smart speakers and Google TV devices. If you mainly talk to Google through a Nest speaker at home, nothing changes yet. **What this actually feels like day to day.** Basic things — setting a timer, sending a text, making a call — will still work. What changes is the personality answering you: Gemini talks more like a chatbot having a conversation, while the old Assistant was built around quick, single commands. If a request that used to just work now gets a longer or different-sounding answer, that is the new assistant, not a broken phone. Our [guide to using Gemini on Android](/guides/gemini-android-guide) walks through what it can do and how to get comfortable with it. **One more thing worth knowing.** If you get an email from Google about this change, it is real — Google is notifying users directly as the switch reaches their devices. That said, this is exactly the kind of "your service is changing, click here" message that scammers like to copy. Stick to checking assistant settings directly on your phone rather than clicking links in an email you were not expecting, and see our guide on [spotting a fake AI app](/guides/is-ai-app-scam) if anything about a message like this feels off. ## Sources - [It's official: Gemini will completely replace the Google Assistant on phones next month](https://chromeunboxed.com/its-official-gemini-will-completely-replace-the-google-assistant-on-phones-next-month/) — Chrome Unboxed - [Google's Gemini Will Start Replacing Google Assistant Today](https://www.sitepronews.com/2026/08/31/googles-gemini-will-start-replacing-google-assistant-today/) — SitePro News --- # Google's $68 Million Assistant Privacy Settlement: How to Claim Your Payout by August 27 Source: https://aiforregulars.com/news/google-assistant-privacy-settlement-payout Google has agreed to pay $68 million to settle a class action lawsuit over Google Assistant, the voice helper built into Google Home speakers, Nest smart displays, and Pixel phones. The claim window is open now, and the deadline is firm: **August 27, 2026**. **What Google is accused of.** The case, *In re Google Assistant Privacy Litigation* (Case No. 5:19-cv-04286, U.S. District Court for the Northern District of California), alleges that Google Assistant sometimes activated and recorded audio without anyone actually saying "Hey Google" or "OK Google" — a mistake the industry calls a "false accept" — and that some of those recordings were reviewed by outside contractors to help improve the software. Google denies any wrongdoing and settled without admitting fault. **Who can file a claim.** Two groups qualify, and you don't need to belong to both. You're likely eligible if, between May 18, 2016, and March 19, 2026, you or your minor child either bought a Google-made device — a Google Home, Home Mini, Home Max, Nest Hub, Nest Hub Max, or Pixel phone — or used any Google Assistant-enabled device (including third-party smart speakers with Assistant built in) and believe a conversation was picked up without the wake word being said. Both groups must also confirm they live in the US or a US territory. **What it actually pays.** This isn't a life-changing sum. Early estimates from settlement trackers put payouts for people who bought an eligible device (up to three per claimant) at roughly $18 to $56 per device, while people claiming only under the privacy class, without a purchase, are estimated at around $2 to $10. The exact amount depends on how many people file valid claims, since the $68 million fund is split after legal fees and costs — more claims means a smaller share for everyone. **How to file.** Go to the official settlement site, GoogleAssistantPrivacyLitigation.com, and complete the claim form online (a mail-in form is also available). If you received a notice by email, it included a Unique ID and PIN that speeds up verification — but you don't need one to file; the site has a separate option for people without a notice code. Purchaser claims may ask for supporting documentation like a receipt or order confirmation, while privacy-only claims can generally be filed on your own attestation, without proof. It takes a few minutes either way. **Don't wait until the last week.** The deadline to file a claim, or to opt out of the settlement entirely, is August 27, 2026 — more than a month before the court's final approval hearing on October 1, 2026. If you're unsure whether your household's smart speaker has ever picked up a conversation by mistake, [how Alexa's wake word actually works](/guides/alexa-always-listening-privacy) explains the same basic mechanism most voice assistants share, including how to check and delete your own voice history. --- # Google Opens Pre-Orders for 'Googlebook' — Its First Gemini AI Laptops Source: https://aiforregulars.com/news/google-googlebook-gemini-laptop-launch If you've been shopping for a new laptop, there's a new option on the shelf. On September 21, 2026, Google opened pre-orders for **Googlebook** — a new laptop category built around its Gemini AI, made together with five hardware partners: Acer, Asus, Dell, HP, and Lenovo. The five launch models run from $899 to $1,299, with shipping starting October 4, 2026, through the Google Store, Best Buy, and other retailers. **You don't need one.** This is the part worth reading first: Gemini already works, for free, on the computer or phone you already own. Googlebook doesn't unlock some AI feature you're otherwise missing — it's a new choice for people who happen to be buying a brand-new laptop anyway, not a reason to replace a computer that's working fine. **What's actually new is two hardware touches.** The first is called Magic Pointer: instead of copying text or taking a screenshot to ask Gemini about something on your screen, you wiggle your cursor over it and a Gemini panel opens with that content already in view. The second is Glowbar, a thin light strip on the laptop's lid that glows to show battery status and lets you know when Gemini is actively listening or responding — similar in spirit to the light some smart speakers use. **What's bundled matters more than the hardware.** Every Googlebook comes with 12 months of Google AI Pro included at no extra cost — that's Google's $19.99-a-month plan (see our [comparison of ChatGPT Plus, Google AI Pro, and Claude Pro](/guides/ai-paid-plans-compared)), which normally runs close to $240 a year on its own. Buyers also get 3 months of YouTube Premium and 3 months of Adobe Photoshop thrown in. **Who Google is actually building this for.** The "50 million" figure attached to this launch in press coverage refers to Chromebooks already in schools, used by students and teachers — not the general public. Google says those existing Chromebooks will keep being supported, but has said many will eventually be eligible to move to the Googlebook experience, which reporters read as a long-term push to bring Gemini into education. If you're buying for yourself rather than for a classroom, that detail doesn't really apply to you — what matters is the question below. **Is it worth it?** Do the math on your own situation rather than the marketing pitch. If you're already planning to buy a new laptop soon and you live inside Google's world — Gmail, Google Docs, Google Drive — the free year of AI Pro is real money back, especially if you'd have paid for extra Google storage anyway. If your current laptop still does what you need, there's no urgency: nothing about Gemini is being taken away from the device you already have. ## Sources - [Pre-order Googlebook](https://blog.google/products-and-platforms/devices/googlebook/pre-order-googlebook/) — Google Blog - [The built-in intelligence of Googlebook](https://blog.google/products-and-platforms/devices/googlebook/googlebook-built-in-intelligence/) — Google Blog - [Googlebook pre-orders go live with 5 new laptops, from $899 to $1,299](https://www.tomsguide.com/computing/laptops/googlebook-pre-orders-go-live-with-5-new-laptops-from-usd899-to-usd1-299-heres-what-theyre-like-to-use) — Tom's Guide - [Google opens pre-orders for partner-built Googlebook laptops](https://www.unite.ai/google-opens-pre-orders-for-partner-built-googlebook-laptops/) — Unite.AI - [Google's $899 Googlebook is a bet that you'll buy a new laptop for Gemini](https://techcrunch.com/2026/09/21/googles-899-googlebook-is-a-bet-that-youll-buy-a-new-laptop-for-gemini/) — TechCrunch --- # ChatGPT Just Got Three New Models — Here's What Changes for You Source: https://aiforregulars.com/news/gpt-5-6-sol-terra-luna OpenAI released a new family of AI models today called GPT-5.6, and it comes with three options instead of one. They're named Sol, Terra, and Luna — after the sun, the earth, and the moon — which is a shorthand for how powerful each one is. **Sol** is the flagship of this release. It's built for tasks that take serious thinking: complex research, writing long reports, coding projects, and multi-step problems where the AI has to plan ahead. If you've heard the phrase "deep reasoning," this is the model that does that. There's even a stronger version called Sol Ultra for the most demanding work. Most everyday users won't need this one for routine tasks. (Update, September 2026: OpenAI has since released a newer flagship, GPT-6 Astra — see our news on [what GPT-6 Astra changes](/news/chatgpt-gpt-6-astra-explained).) **Terra** is the middle option. OpenAI describes it as balanced — roughly as capable as what ChatGPT used for last year, but at half the running cost. The most important thing for regular users: **Terra is now the default model for Free and Go accounts**. That means if you use ChatGPT without paying, you're getting a better AI starting today, automatically, with no action required on your part. **Luna** is the smallest and fastest of the three. It replies almost instantly and costs the least to run. It's well suited for simple questions, quick lookups, and back-and-forth chat where speed matters more than depth. If you're on a paid plan (Plus, Pro, Business, or Enterprise), you get to choose. A model picker in the chat interface lets you switch between Sol, Terra, and Luna, and you can also adjust how much thinking time the model takes. A setting called "max reasoning effort" lets Sol go deeper on hard problems — useful when you want thorough answers and don't mind waiting a few extra seconds. Want a plain-English explanation of what these three models mean in practice? Read our guide on [ChatGPT Sol, Terra, and Luna explained](/guides/chatgpt-sol-terra-luna-explained). OpenAI also replaced the old Advanced Voice Mode with something called **GPT-Live-1**. The biggest practical difference is that it works like a real phone call: you and the AI can speak at the same time, and it handles interruptions naturally instead of waiting for you to finish. There's also built-in live translation — it can interpret between languages in real time as you speak, which opens up some genuinely useful possibilities for travel or calling a family member who speaks a different language. Free accounts get the smaller GPT-Live-1 mini; paid accounts get the full version. Read more in our guide to [ChatGPT voice mode](/guides/chatgpt-voice-mode). **The one thing you need to do right now: be careful about fake "upgrade" links.** Every time OpenAI makes a big announcement, scammers flood social media with posts and ads that say things like "Download GPT-5.6 now — limited time!" or "Upgrade your ChatGPT to Sol here." These links lead to fake websites that either charge you money for something that's free, steal your login details, or install software you don't want on your device. The real update happens automatically inside the official ChatGPT app and at chatgpt.com. You do not need to download anything new. You do not need to pay anyone outside of OpenAI's own website to access the new models. If a link or post is telling you otherwise, it's a scam. Our guide on [how to spot fake AI apps](/guides/is-ai-app-scam) walks you through the warning signs. The short version: if you use ChatGPT for free, it got better today without you lifting a finger. If you pay for it, you now have more control over which model handles your requests. And the voice feature works more like a real conversation. Everything else stays the same — same app, same login, same website. --- # iOS 27's Public Beta Is Here — Should You Install It for the New Siri? Source: https://aiforregulars.com/news/ios-27-public-beta-new-siri Apple opened the public beta of iOS 27 on July 13–14, 2026 — as always, free to join without a developer account. The main reason anyone's bothering: for the first time, the rebuilt Siri Apple showed off in June is available outside the closed developer beta, for anyone willing to try unfinished software. **How to get it.** Sign up for free through the Apple Beta Software Program, then install iOS 27 Public Beta from **Settings > General > Software Update** on your iPhone. No developer account and no fee required — just some time and a supported device. **Back up your phone first.** A public beta is, by definition, not finished software. Apps can crash, battery life can suffer, and features can behave unpredictably. Before you install it, back up your iPhone to iCloud or your computer so you can restore it if something goes wrong. **Not every iPhone gets the new Siri.** Installing the beta doesn't automatically unlock the rebuilt Siri. It requires an iPhone 15 Pro or 15 Pro Max or newer with the chip Apple's new AI features depend on. Older iPhones can install the general iOS 27 beta but won't see the Siri changes. The new Siri also isn't rolling out at all for Apple Accounts registered in the European Union — Apple says EU rules (the Digital Markets Act) are delaying it there. Our guide to [what's actually new in Siri](/guides/siri-ai-whats-new) covers what's different and which devices qualify in more detail. **Even if you qualify, it's not instant.** Most people who install the beta on a supported iPhone and open Siri land on a waitlist rather than getting the new features right away. Treat the public beta as a chance to get in line early, not a guaranteed preview. **If you'd rather not risk it.** The finished version of iOS 27, including the new Siri, is now out: Apple released it on September 14, 2026 as a standard update for everyone. There was no downside to waiting — the same features, without the risk of unfinished beta software on your main phone. See [our update on the launch](/news/ios-27-siri-ai-live-eu-blocked) for who actually gets the new Siri now, including the EU exclusion this beta already hinted at. --- # The New Siri Is Out Now With iOS 27 — Unless You're in the EU Source: https://aiforregulars.com/news/ios-27-siri-ai-live-eu-blocked Apple released iOS 27 on September 14, 2026, alongside iPadOS, macOS, watchOS, and visionOS 27. The headline feature is the one we covered back in July when its public beta opened: a rebuilt Siri that can carry out multi-step tasks, see what your camera is pointed at, and remember past conversations. It's no longer a beta — it's now the version of Siri arriving on ordinary iPhones through a standard software update. Our guide to [what's actually new in Siri](/guides/siri-ai-whats-new) covers the features themselves in detail; this is about who gets them, starting now. **Which iPhones qualify.** You need an iPhone 15 Pro, 15 Pro Max, or newer. If you install iOS 27 on an older supported iPhone, you'll get the general update but not the rebuilt Siri — it needs the chip that only those models have. **Having the right iPhone doesn't mean instant access.** Even on a qualifying device, most people who update and open Siri are landing on a waitlist rather than getting the new features immediately. Apple is rolling access out gradually rather than switching it on for everyone at once. If you update and nothing looks different yet, that's expected — not a sign something went wrong. **If you're in the European Union, the new Siri isn't coming with this update at all.** Apple confirmed it's withholding the rebuilt Siri from iOS 27 and iPadOS 27 for EU-registered Apple Accounts, citing an unresolved dispute over the bloc's Digital Markets Act. Apple says it proposed ways to comply that would let it ship the feature; those proposals haven't been accepted, and there's no date for when EU iPhone and iPad users will get it. Notably, the same rebuilt Siri **is** shipping in the EU on Mac and Vision Pro with macOS and visionOS 27 — the restriction covers iPhone, iPad, and Apple Watch, tied to how those devices are regulated under the DMA. If you're in the EU, updating your iPhone to iOS 27 is still worth doing for the other changes, but don't expect the new Siri to show up. **Language matters too, everywhere.** Even where the new Siri is available, it launches in English first. Apple has said French, Japanese, Korean, Portuguese, and Spanish support will follow in October. **How to update.** If you're on a qualifying iPhone outside the EU, go to **Settings > General > Software Update** to install iOS 27. There's no separate download or sign-up needed now that the beta period is over — just the normal update process. ## Sources - [Apple releases iOS 27 with all-new Siri](https://www.cnbc.com/2026/09/14/apple-releases-ios-27-redesigned-siri-ai.html) — CNBC - [iOS 27 Features Available Tomorrow](https://www.macrumors.com/2026/09/13/ios-27-features-available-tomorrow/) — MacRumors - [iOS 27 With Siri AI Will Be Available on September 14](https://www.engadget.com/2254005/ios-27-with-siri-ai-will-be-available-on-september-14/) — Engadget - [iOS 27 Launches Today: Siri AI Requires Waitlist, Not Just a Compatible iPhone](https://www.techtimes.com/articles/327504/20260914/ios-27-launches-today-siri-ai-requires-waitlist-not-just-compatible-iphone.htm) — Tech Times - [Due to the DMA, Siri AI is delayed in the EU for iOS 27 and iPadOS 27](https://www.apple.com/newsroom/2026/06/due-to-dma-siri-ai-delayed-in-eu-for-ios-27-and-ipados-27/) — Apple Newsroom --- # Meta AI Will Now Warn Parents If a Teen Talks About Self-Harm Source: https://aiforregulars.com/news/meta-ai-self-harm-parent-alerts Meta announced on July 16, 2026 that Meta AI — the chatbot built into Instagram and Facebook — will now alert parents when it detects a teenager talking about self-harm or suicide in a conversation with the AI. **How it works.** A dedicated AI classifier scans chats with Meta AI for signs of self-harm or suicidal distress. Flagged conversations get a human review before any alert goes out, and Meta says that when the signal isn't clear-cut, it errs on the side of notifying a parent anyway. **What you'll see — and what you won't.** If your teen's account is flagged, you'll get a notification letting you know there's cause for concern, along with crisis resources and tips for starting a conversation. You won't see the exact wording of what your teen wrote — Meta keeps that private. **Who has it right now.** The alerts are live today for parents using Instagram's Parental Supervision tool in the US, UK, Australia, and Canada. Meta plans a global rollout by the end of 2026. If you haven't turned on Parental Supervision for your teen's Instagram account, this is a good reason to check. **Worth knowing.** Human review means there's a delay between a flagged chat and any alert reaching you, and no detection system catches everything. Treat this as an extra layer, not a substitute for talking with your teen directly. If you're in the US and worried about a child in crisis, call or text **988** (Suicide and Crisis Lifeline) any time, day or night. For step-by-step help turning on Instagram's parental tools, see our guide to [parental controls for AI apps](/guides/ai-parental-controls-all-apps). Want to know what else Meta AI can do on Facebook and Instagram? Check out [our Meta AI guide](/guides/meta-ai-facebook-instagram). --- # Meta Launched an AI Agent That Shops and Pays With Your Money — Here's What to Know Source: https://aiforregulars.com/news/meta-muse-ai-agent-explained On September 8, 2026, Meta launched Muse — its first "personal AI agent," a step beyond the chatbots you're used to. Instead of just answering questions, Muse goes out and does things for you: it opens a web browser, fills out forms, sends emails, books travel, and completes purchases using a payment card you've added. It's available now in the US through WhatsApp, Meta's own app on iOS and Android, and the web at muse.ai. **What Muse can actually do.** Point it at a goal — "find a birthday gift for my sister" or "book a table for four on Friday" — and it browses the web, compares options, and can finish the task itself, including checking out at a store. It can turn things you've saved elsewhere, like an Instagram recipe, into a grocery list, or suggest dinner ideas based on dietary restrictions you've told it about. Unlike a regular chat, it keeps working after you close the app and comes back to you when it's done — or when it needs your say on something. **How it handles your money.** Muse doesn't use your card number directly. Purchases currently run through Link by Stripe, which generates a one-time card number for each transaction — Meta says Shop Pay and 1Password support are coming later. Meta also says it's the first AI agent covered by Link's purchase protections — the same coverage you'd get for a damaged item, a price drop, or a return, had you bought it yourself. You do need to add a payment method to use the shopping features at all. **What it costs.** Basic use is free. Two paid tiers unlock heavier use: Power at $20 a month and Maximum at $100 a month. **The safety controls Meta built in.** Muse runs inside its own isolated "Secure VM," a separate virtual computer walled off from other users, and a second system Meta calls a "Sentinel agent" reviews and approves everything it does on the internet before it happens. You choose which of your apps Muse can reach and how much access to give each one, and it's supposed to ask for your approval before sending an email or making a purchase rather than doing it silently. This "pause and ask" pattern is also something independent developers build into their own agents — services like [Impri](https://impri.dev) exist specifically to add a human-approval step before an agent is allowed to act (disclosure: built by the author of this site). Meta says none of this activity is shared with its advertising systems. **Why this is worth pausing on.** A tool that can browse, negotiate, and spend money on its own is a real convenience — and a real change from anything you've given app permissions to before. Before you connect a card, review exactly which apps and how much spending Muse can access, start with small, low-stakes tasks while you get a feel for it, and check your card statement afterward the way you would with any new payment method. It's currently US-only, so if you're outside the US, you won't see it in WhatsApp or the Meta app yet. ## Sources - [Introducing Muse: The World's First Personal AI Agent Built for Everyone](https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/) — Meta - [Meta debuts its Muse AI agent. Will consumers trust it?](https://techcrunch.com/2026/09/08/meta-debuts-its-muse-ai-agent-will-consumers-trust-it/) — TechCrunch - [Meta puts an AI agent that buys things inside WhatsApp, US only](https://ppc.land/meta-puts-an-ai-agent-that-buys-things-inside-whatsapp-us-only/) — PPC Land --- # Meta Launched a Paid AI Subscription Called Meta One — Here's What's Still Free Source: https://aiforregulars.com/news/meta-one-ai-subscription-explained If you've noticed a new "Meta One" mention popping up in Facebook, Instagram, or WhatsApp settings, it's not a bug and it's not a hidden fee — it's a new paid tier Meta rolled out on September 15, 2026 for people who want more out of its AI tools. **What Meta One actually is.** It's a subscription that spans Facebook, Instagram, WhatsApp, and Meta AI, sold in two consumer bundles: **Core** at $7.99 a month and **Premium** at $19.99 a month. Paying gets you higher usage limits on the AI features that are otherwise capped — more image and video generation using Meta's Muse models, and more use of in-app tools like Restyle on Instagram. Premium gives you noticeably more headroom than Core. Meta says the plans are rolling out globally, though exact pricing and which features are included can vary by country and app. **What stays free.** This is the part worth underlining: Meta has said directly that "the core experience across our apps and Meta AI has always been free, and that's not changing." You don't need Meta One to use Facebook, Instagram, or WhatsApp, or to chat with Meta AI. What you're paying for is a higher ceiling on the AI-heavy extras — think of it like a free app with an optional pass for people who generate a lot of AI images or video and keep hitting a daily limit. **Who this is actually for.** If you occasionally ask Meta AI a question or generate the odd photo edit, you likely won't notice a difference — you're using less than the free tier allows. Meta One mainly matters to people who create a lot of AI content regularly and keep running into "you've reached your limit" messages. There's also a separate set of business and creator plans (roughly $15 to $499 a month) built around a "Meta Business Agent" that answers customer questions and books appointments automatically — that's aimed at businesses running a Facebook or Instagram page, not at everyday personal use. **Not to be confused with Muse.** Meta also launched an AI shopping agent called Muse earlier this month — see our [explainer on Meta's Muse AI agent](/news/meta-muse-ai-agent-explained). Muse is a separate product with its own pricing (Power at $20/month, Maximum at $100/month) that browses the web and makes purchases for you. Meta One is about usage limits on AI tools already inside Facebook, Instagram, and WhatsApp; Muse is a shopping assistant. They're easy to mix up because both launched in September 2026, but they do different things and are billed separately. **Why this is worth pausing on.** Any time a company rolls out a new paid tier, it's also a moment scammers exploit. Expect fake "Your Meta One subscription payment failed" or "confirm your Meta One billing" messages by email or text in the coming weeks — these follow the exact same pattern as fake Netflix or Apple billing alerts. Manage or check any Meta One subscription only inside the app itself (Settings → Meta One or Accounts Center), never through a link in an email or text. Our guide on [spotting AI phishing emails](/guides/spot-ai-phishing-emails) covers the warning signs, and if you're already juggling a few AI subscriptions, [our guide to auditing them](/guides/too-many-ai-subscriptions) can help you decide what's actually worth paying for. ## Sources - [Introducing Meta One: A New Way to Subscribe Across Our Apps](https://about.fb.com/news/2026/09/introducing-meta-one-subscription-service-more-features-ai/) — Meta - [Meta One brings paid AI tools to Instagram and WhatsApp](https://dataconomy.com/2026/09/16/meta-one-brings-paid-ai-tools-to-instagram-and-whatsapp/) — Dataconomy - [Meta Is Hoping You'll Pay for More AI Features on Instagram](https://gizmodo.com/meta-is-hoping-youll-pay-for-more-ai-features-on-instagram-2000812018) — Gizmodo --- # Minnesota Just Banned AI 'Nudification' Apps — Here's What That Means Source: https://aiforregulars.com/news/minnesota-ai-nudification-law-explained If you've read about "nudification" apps — AI tools that take an ordinary photo of a real person and generate a fake nude or sexualized image of them — there's a new law worth knowing about. On August 1, 2026, Minnesota became the first U.S. state to ban this specific kind of AI tool outright. The law, known as HF 1606, doesn't target the person who types a request into one of these tools. Instead, it goes after the companies and websites that offer the tool in the first place — banning any online service that lets someone turn a real photo into a convincing fake nude image of an identifiable person without that person's consent. Violators face civil penalties of up to $500,000 per violation, pursued by the state Attorney General. Separately, the law also gives victims themselves the right to sue an app's owner directly for damages. The law hasn't gone unchallenged. Elon Musk's AI company, xAI, whose Grok chatbot can generate and edit images, sued to block it just before it took effect, arguing it's written so broadly that it violates free speech. A federal judge, Donovan Frank, denied xAI's request for an emergency pause on July 31 — his reasoning was that xAI had waited nearly three months after the law was signed before asking for emergency relief, which undercut its claim that it needed the court to act immediately. That let the law take effect on schedule. A full hearing on whether to block the law for longer took place on August 19; as of this writing, the judge hasn't yet issued his decision. For now, the law stands, and Minnesota residents targeted by this kind of fake image have both the Attorney General's office and their own right to sue behind them, on top of ordinary platform reporting. If you or someone you know discovers a fake AI image or video like this — in Minnesota or anywhere else — our guide on [what to do if someone made a deepfake of you](/guides/deepfake-of-me-what-to-do) covers the practical steps: saving evidence, getting it taken down, and involving police if it's being used to threaten or extort you. ## Sources - [Judge denies request by Elon Musk's xAI to pause Minnesota nudification ban](https://www.nbcnews.com/tech/elon-musk/judge-denies-request-elon-musks-xai-block-mn-nudification-ban-rcna589993) — NBC News - [Request by Elon Musk's xAI to temporarily halt Minnesota nudification technology ban denied by federal judge](https://www.cbsnews.com/minnesota/news/restraining-order-denied-ai-nudification-ban-law/) — CBS News Minnesota - [Minnesota Bans Nudification Technology in First-of-Its-Kind AI Law](https://www.multistate.ai/updates/vol-103-minnesota-nudification-technology-ban) — Multistate.ai - [Minnesota Deepfake Laws: AI Images, Voice Cloning & Penalties](https://www.recordinglaw.com/us-laws/deepfake-laws/minnesota-deepfake-laws/) — Recording Law --- # WhatsApp's Meta AI Gets an Incognito Mode — Here's What It Actually Hides Source: https://aiforregulars.com/news/whatsapp-ai-private-mode WhatsApp's built-in AI assistant just got a privacy upgrade. Meta launched Incognito Chat for Meta AI conversations in WhatsApp on May 13, 2026, and it changes how your AI chats are handled in a meaningful way. **What's actually new.** When you turn on Incognito Chat before talking to Meta AI, those messages are processed in a Trusted Execution Environment (TEE) — a secure, isolated part of the server that Meta says even its own engineers can't access. The AI chats also disappear by default when the conversation ends, leaving no history stored on Meta's servers. **Why this matters.** Normally, when you chat with Meta AI in WhatsApp, those messages don't have the same protections as your regular WhatsApp conversations. Your regular messages to friends and family have always been end-to-end encrypted — Incognito Chat doesn't change that. What it does is bring your Meta AI chats closer to that standard for the first time. So if you've been hesitant to ask Meta AI about something personal — a health question, a financial worry, something you'd rather not have stored — this mode gives you a more private way to do it. **What it does not do.** Even with Incognito Chat on, you're still talking to an AI, and the conversation is still processed by computers to generate a response. "Private from Meta" is not the same as "completely private." It is not the right place for passwords, login details, or sensitive medical records. Read [what not to tell an AI chatbot](/guides/what-not-to-tell-ai-chatbot) for a clear rundown of where to draw the line. **How to use it.** Look for the incognito option before starting a Meta AI chat in WhatsApp. The exact location in the app may vary depending on your version and region. Meta has also announced a "Side Chat" feature — a separate space for AI conversations — planned for later in 2026. If you're new to Meta AI in WhatsApp, our guide to [Meta AI in WhatsApp](/guides/meta-ai-whatsapp) explains the basics. --- # Check if a charity is legitimate before you donate — it takes two minutes Source: https://aiforregulars.com/tips/ai-charity-scam-check When a hurricane hits, a wildfire breaks out, or a flood makes the news, scammers are ready within hours. They spin up convincing charity websites and donation campaigns — and today, AI tools make fake appeals look every bit as polished as the real thing. AARP and consumer-protection agencies consistently warn that charity fraud spikes after every major disaster. The good news: two quick checks will protect your money and make sure your generosity actually reaches people who need it. **Step 1: Look the charity up in an official registry** Before you give, search for the organization on Charity Navigator (charitynavigator.org, free) or the IRS Tax-Exempt Organization Search — also called IRS TEOS — at irs.gov/charities-non-profits. Both are free. If the charity isn't listed, that's a serious red flag. Legitimate organizations are registered, and the registries show how they spend donations. **Step 2: Paste the appeal into ChatGPT and ask for a second opinion** If you received an email, social media post, or website text asking for a donation, copy it and paste it into ChatGPT or any AI chatbot. Try a prompt like this: > "Here is a donation appeal I received. What red flags do you see? Does this charity appear in official registries?" The AI will flag pressure tactics, vague language about where the money goes, unusual payment requests, and names that sound like well-known charities but aren't quite right. **The red flags to know** Watch out for these warning signs, with or without AI help: - Pressure to donate **right now** before an offer expires - Requests for payment by gift card, wire transfer, or cryptocurrency — legitimate charities don't ask for these - An organization name nearly identical to a famous charity (for example, "American Red Cross Relief Fund" instead of "American Red Cross") - Donation links arriving in unsolicited texts, emails, or social media messages **How to pay safely** If the charity checks out, go directly to the charity's official website — type the address yourself rather than clicking a link. Pay by credit or debit card on the official site. Never follow a payment link from an unsolicited message, even if it looks genuine. The same pattern of manufactured urgency that charity scammers use shows up in phishing emails too — [our guide to spotting AI phishing emails](/guides/spot-ai-phishing-emails) walks through how to recognize it. **One important caveat** AI is a useful first filter — it can spot the language patterns and pressure tactics that real charities don't use. But it cannot confirm that an organization is actually legitimate. Only the official registries can do that. Use AI to flag what looks suspicious, then use Charity Navigator or IRS TEOS to verify. When you're unsure whether a specific claim is true, the same fact-checking approach from [spotting AI-generated misinformation](/guides/ai-fake-news-fact-check) applies here too. Two minutes of checking is a small price for knowing your donation actually helps. --- # Packing a bag? Let AI make the checklist Source: https://aiforregulars.com/tips/ai-packing-list-for-trip Most of us have arrived somewhere and realised the charger, the medication or the umbrella is still on the kitchen table. A chatbot such as ChatGPT or Gemini can make you a checklist built around your trip, not a generic one. **Open ChatGPT or Gemini and write something like:** > "I am going to Lisbon for 4 nights in late October with my husband. We will walk a lot and have one nice dinner out. Make me a packing checklist grouped by clothes, toiletries, documents and chargers, and keep it short." Change the details to match your trip: where, how many nights, who is coming, and what you will be doing. Then add a follow-up if you like: "I only have a carry-on bag, what can I leave out?" or "Add things for a toddler." Tick items off on your phone as you pack. **One thing to watch:** The chatbot may not know the real weather for your dates, so check the forecast yourself and tell it. Documents, tickets and medication are the things you cannot buy on arrival, so double-check those against your own booking confirmations. Next trip, spend half a minute on the list before you start packing. It is the cheapest way to avoid a forgotten charger. --- # Ask AI what to cook tonight using only what's already in your kitchen Source: https://aiforregulars.com/tips/ai-recipe-from-what-you-have Most nights, the hardest part of dinner isn't cooking — it's deciding what to make. Open the fridge, stare at half a bag of spinach, three eggs, and last night's rice, and it's easier to just order takeout or make another store run. AI is good at exactly this kind of puzzle: turning a random list of ingredients into an actual meal. **Tell ChatGPT or Gemini what you already have, and ask it to build a meal around it:** > "I have [list your ingredients — e.g. spinach, three eggs, half an onion, cooked rice, cheddar cheese]. Suggest one simple dinner I can make mostly with these, using only basic pantry staples (salt, oil, pepper) for anything else. Keep it under 30 minutes and walk me through it step by step." You can be as specific as you like — mention how many people you're feeding, any allergies, or that you only have 15 minutes to cook. If the first suggestion doesn't sound appealing, just ask for a different one; it costs nothing to try again. This works just as well for lunch, a school-week breakfast, or figuring out what to do with produce that's about to go bad. **One thing to watch:** AI recipe suggestions are a starting point, not a food-safety check — use your own judgment about anything that smells off or has sat in the fridge too long, and double-check any ingredient swap yourself if you're cooking for someone with a serious allergy. Next time you're staring into the fridge wondering what to make, skip the guesswork — describe what's in there and let AI figure out dinner. --- # Ask AI to decode your energy bill — you might be on the wrong rate plan Source: https://aiforregulars.com/tips/ai-understand-energy-bill Energy bills are designed by utilities, not by people who read them. Behind the total due sits a small maze: tiered pricing tiers, peak-hour charges, distribution fees, fuel adjustments, and sometimes a "customer charge" you pay whether you use any electricity or not. Most people just pay the number at the bottom and never know if they're overpaying. Here's a quick way to find out. Type — or photograph — your bill and bring it into ChatGPT or Gemini. Then ask: > "Here is my electric bill. Explain each line item in plain English, and tell me whether my usage pattern suggests I should ask my utility about a different rate plan (like time-of-use or budget billing)." The AI will walk through every charge and flag anything worth questioning. It can also look at your usage numbers and tell you whether your habits — running the dishwasher after dinner, charging a car overnight, or heavy air-conditioning in summer — suggest a different plan might save you money. **What might come up:** - **Time-of-use (TOU) plans** charge a lower rate during off-peak hours (typically late night and early morning) and a higher rate during peak demand hours (usually late afternoon and evening). If you can shift laundry, dishwasher cycles, or EV charging to off-peak windows, these plans can cut your bill by 20–30%. - **Budget billing** smooths out seasonal spikes by averaging your annual usage into equal monthly payments — useful if you dread the summer air-conditioning bill or the winter heating spike. - **Supplier switching** is available in some regions where the energy market is deregulated. Your utility still delivers the power, but you can choose who you buy it from — and rates can vary meaningfully. If you already use AI to [understand contracts or legal documents](/guides/ai-explain-contracts), the same skill applies here: it's good at turning official jargon into plain English. **What to watch for:** AI reads what you give it — it doesn't have access to your utility's current rate schedules, promotions, or local eligibility rules. Use the AI's explanation to know what questions to ask, then confirm the details on your utility's website or by phone. If you're uploading a photo of your bill, black out your name, address, and account number first — they're not needed for rate analysis, and there's no reason to share them. For a fuller picture of where your money goes each month, [AI budgeting tools](/guides/best-ai-budgeting-apps) can pull in your bills automatically and spot patterns over time. --- # Ask AI to help you appeal a denied insurance claim Source: https://aiforregulars.com/tips/appeal-denied-insurance-claim You get a letter or an email: "Your claim has been denied" — for health, car, home, or travel insurance — full of policy-code jargon and no clear next step. Most denial letters do explain what to do if you disagree, but the wording is confusing enough that a lot of people just give up and pay out of pocket. A chatbot can decode the reason in a minute and help you draft a response the same day. **Paste the denial letter into ChatGPT or Gemini and ask:** > "This is a letter denying my insurance claim. Explain in plain language why they say they're denying it, and tell me whether the letter mentions how to appeal or ask for reconsideration. Then draft a short, polite letter appealing the decision, referencing the claim number, and asking them to review it again." The more you paste in — the full letter, your policy number, any receipts or notes you already have — the more useful and specific the draft will be. **One thing to watch:** AI is only reading the letter you gave it, not your full policy or the exact rules in your country or state, so it can miss a hard deadline or a required form. Before you send anything, check the letter (or call the insurer) for the actual appeal deadline — some give you as little as 30 days — and treat the AI draft as a strong starting point, not a guaranteed win. Next time a "claim denied" letter lands in your mailbox, don't file it away — paste it into AI first and see whether a two-minute appeal is worth sending before you accept the answer as final. --- # Ask ChatGPT or Gemini to identify a mystery plant or bug from a photo Source: https://aiforregulars.com/tips/ask-ai-identify-plant-bug You're pulling weeds and spot something you don't recognize. Or a strange bug is crawling along your patio and you want to know if it bites. You could download a dedicated plant or insect identification app — or you could just use the chatbot already on your phone. **Open ChatGPT or Gemini, take or upload a photo of the plant or bug, and ask:** > "What plant/bug is this? Is it safe to touch or should I avoid it? And if it's a weed, what's the easiest way to get rid of it?" You'll usually get an answer in seconds: the likely species, whether it's harmless or something to be careful around (stinging nettle, poison ivy, a wasp rather than a bee), and a next step — pull it out, leave it alone, or call pest control. This works well for garden mysteries, but it's just as handy on a walk with the kids when they find an interesting leaf, or when you spot mold or a stain on a wall and want a first guess at what it is before you decide whether to worry. **One thing to watch:** treat the answer as a strong first guess, not a certain diagnosis — a bad photo (blurry, poor lighting, odd angle) can lead the AI astray, and some toxic plants closely resemble harmless ones. Never eat a plant, let a child handle one, or assume a bite or sting is harmless based on a photo ID alone; when it actually matters, double check with a local nursery, extension office, or poison control. Next time something growing or crawling in your yard makes you go "what is that?", your phone already has the answer. --- # Turn on ChatGPT's Lockdown Mode before discussing anything sensitive Source: https://aiforregulars.com/tips/chatgpt-lockdown-mode ChatGPT quietly added a setting called Lockdown Mode in June 2026. It's designed for moments when you want a conversation to stay contained — nothing going in, nothing coming out beyond plain text. When you turn it on, ChatGPT disables live web access, image generation and viewing, Deep Research, Agent Mode, and file downloads. The conversation becomes strictly text-to-text. **When it's worth switching on** Think about conversations you'd rather keep quiet: working through a health concern with your symptoms written out, talking through a legal situation, doing the maths on a family financial decision, or describing a personal conflict in detail. These are the moments where you'd prefer no web searches are running in the background and no files leave the chat. > "I need to talk through a sensitive family situation. Before we start, I've turned on Lockdown Mode so this stays text-only." Lockdown Mode is also useful if you're sharing information on behalf of someone else — a family member's medical situation or a neighbour's legal problem — and you want extra reassurance that the conversation isn't pulling in outside data. **What Lockdown Mode does NOT cover** This is the important part: Lockdown Mode does not delete your chat history, turn off ChatGPT's memory, or change OpenAI's data practices. Your normal privacy settings still apply exactly as before. If you haven't reviewed what ChatGPT remembers about you or whether your chats are saved, those are separate steps — the [ChatGPT privacy settings guide](/guides/chatgpt-privacy-settings) walks you through them. And Lockdown Mode doesn't change the golden rule: some things don't belong in any chatbot. Bank login details, full identity numbers, passwords — these shouldn't go in regardless of mode. The guide on [what not to tell an AI chatbot](/guides/what-not-to-tell-ai-chatbot) has the full list of what to hold back. **How to find it** Go to Settings → Safety and security → Advanced security → Lockdown Mode — it rolled out on June 4, 2026 as an optional security feature. Remember to turn it off again for everyday use; with it on, you lose image support, web browsing, and the AI's research tools, so it slows you down when you don't need the extra caution. One setting, used at the right moments, makes your most sensitive conversations meaningfully more contained. But it's a layer, not a guarantee — pair it with good habits and a quick look at your privacy settings. --- # Ask AI to check a job listing for scam red flags before you apply Source: https://aiforregulars.com/tips/check-job-listing-for-scam-signs Job scams have gotten harder to spot — fake postings copy real company logos, "recruiters" message you first on LinkedIn or WhatsApp, and the pitch always sounds reasonable at first. If a listing or a message about a job feels slightly off, don't just shrug it away — spend two minutes running it past AI first. **Copy the full job posting or the recruiter's message, paste it into ChatGPT or Gemini, and ask it to check for scam red flags:** > "Here's a job posting/message I received: [paste the text]. Look at it like a job-scam checker and tell me if it has any common red flags — such as asking for money or bank details, pressure to decide fast, vague company details, or a request to move the conversation off official channels. List what looks suspicious and what looks normal." Ask it to also flag if the pay seems unusually high for the work described, or if the company name doesn't match what a quick search would normally show. AI can't confirm whether a company is real, but it's good at spotting the patterns scammers reuse — the "equipment deposit," the interview that's a single chat message, the deadline of "reply within the hour." **One thing to watch:** never send bank details, ID or passport numbers, or a payment to "start" a job, no matter how convincing the message looks — legitimate employers don't ask candidates for money. And treat the AI's read as a first filter, not a final verdict: if it says a posting looks fine, still check the company independently — its real website, its LinkedIn page, or a search for the recruiter's name — before sharing anything personal. Next time a job offer lands that feels a little too eager, take two minutes to have AI check it for red flags before you reply — cheap insurance against handing your details to a scammer. --- # Ask AI to sanity-check a repair quote before you approve it Source: https://aiforregulars.com/tips/check-repair-quote-is-fair You get a call or a text: "Your car needs a new alternator, that'll be [amount]" or "the furnace repair is [amount] for the part and labor." You have no way to know if that's a fair price, and questioning a mechanic or technician to their face feels awkward when you don't know the first thing about the part they're quoting. A chatbot can give you a reasonable range and a second opinion before you say yes. **Tell ChatGPT or Gemini what you were quoted, and ask:** > "I was told my [car make/model or appliance] needs [the repair], and the quote is [amount] for parts and labor. Is that price in a normal range for this kind of repair? Does this repair actually match the problem I'm having, which is [describe the symptom]? What questions should I ask the shop before I approve it?" The more detail you give — make and model, the exact part, your city or region — the more useful the answer. If the AI says the price looks well above typical, or that the recommended fix doesn't match your symptom, treat that as a reason to ask more questions or get a second quote, not proof the shop is dishonest. **One thing to watch:** AI price ranges come from general information, not a live look at your actual car or furnace, so treat the answer as a starting point for a conversation, not a verdict. For anything with real safety implications — brakes, gas lines, structural repairs — always get a second opinion from a real technician too, not just from AI. Next time a repair estimate lands in your inbox or a mechanic hands you a clipboard, run the numbers past AI first — it takes two minutes and might save you from an inflated bill or a repair you didn't actually need. --- # Ask AI to check a suspicious text message for scam signs before you click anything Source: https://aiforregulars.com/tips/check-suspicious-text-for-scam A text lands on your phone: “Your package couldn’t be delivered — confirm your address here,” or “Unusual activity on your account, verify now,” or “You’ve won a prize — claim it today.” These messages are built to make you act fast, before you have time to think it through. Scammers count on that rush. AI won’t click anything for you, but it’s very good at spotting the pressure tactics and odd details that give a scam text away. **Copy the message — or type it out — and ask ChatGPT or Gemini:** > "Here’s a text message I received: [paste the full text]. Does this look like a scam? Point out anything specific that makes you suspicious or reassured, and tell me what I should do next." You’ll usually get a clear answer plus the reasoning behind it — a mismatched sender number, urgency that doesn’t fit how that company normally contacts you, a link that doesn’t match its real website. You can also paste in a screenshot if that’s easier than retyping the text. Try it on the next few odd messages you get, even ones that seem obviously fine — it’s a useful habit to build before the message that really matters shows up. **One thing to watch:** never click a link in a suspicious text, even just to “check” it — if the message claims to be your bank, a delivery company, or a government agency, contact them directly through their official app or a phone number you look up yourself. Treat AI’s read as a second opinion, not a guarantee: when in doubt, don’t click and don’t reply. Next time your phone buzzes with something that feels a little too urgent, paste it into AI first — thirty seconds now beats untangling a scam later. --- # Ask AI to check the tone of a message before you hit send Source: https://aiforregulars.com/tips/check-text-tone-before-sending We've all typed a reply while annoyed, hit send, and regretted it five minutes later. Or sent something meant kindly that came out sounding cold. Before your next tricky text, email, or WhatsApp message, try this: **Copy what you've written, paste it into ChatGPT or Gemini, and ask:** > "How does this message come across? Could it sound rude, cold, or confusing to the person reading it? Suggest a friendlier version that still says what I mean." You'll get two useful things back: an honest read on the tone, and a rewritten version you can send as-is or adjust until it sounds like you. This is worth doing for the messages that matter most — a reply to your boss when you're frustrated, a text to a family member after a disagreement, a note to a teacher, or a firm reply to a company that's let you down. It only takes a few seconds longer than typing the message itself, and it's much cheaper than smoothing things over afterwards. **One thing to watch:** AI reads only the words on the page — it doesn't know your history with the person, your usual sense of humour, or how sarcasm lands between the two of you. Treat its read as a second opinion, not the final word, especially with people who'd notice if a message suddenly didn't sound like you. Next time you're hovering over "send" with a knot in your stomach, this thirty-second check might save you an awkward phone call later. --- # Ask AI to translate your lab results into plain English before your next appointment Source: https://aiforregulars.com/tips/explain-medical-results-before-doctor-visit After a blood test, an X-ray, or a doctor's visit, you often get a summary full of terms like "mildly elevated," "within reference range," or "recommend follow-up" — without much explanation of what any of it actually means for you. Portals hand you the numbers; they rarely explain them in plain language. You can fix that in two minutes. Copy the text of your lab result or after-visit summary — or take a photo of the printed page — and paste it into ChatGPT or Gemini with a prompt like this: > "Here are my lab results / after-visit summary. Explain in plain English what each term and result means, what's normal versus not, and suggest three questions I should ask my doctor about anything unclear or concerning." You'll get a clear translation of the jargon and a short list of specific questions to bring to your next appointment — much more useful than sitting in the exam room trying to remember what you wanted to ask. **Before you paste anything:** black out or delete your name, date of birth, and any account or medical record numbers first. You don't need to include them for the AI to explain the terms, and there's no reason to share them. **One important caveat:** AI is good at translating medical language into everyday words and helping you prepare better questions — it is not a doctor and cannot diagnose you or tell you what to do next. Treat its explanation as a way to walk into your appointment better informed, never as a replacement for your doctor's judgment. If anything looks urgent or you feel unwell, contact your doctor or urgent care directly instead of waiting. Two minutes of translating jargon into questions turns a confusing portal message into an appointment where you actually get answers. --- # Ask AI to write your script before you call to negotiate a bill Source: https://aiforregulars.com/tips/negotiate-lower-bill-script Most people dread the call to lower a phone, internet, cable, or insurance bill — so they just don't make it, and keep overpaying every month. Companies count on that. The call itself is the hard part, not the outcome: reps usually have discounts they're allowed to give, you just have to ask the right way. **Tell ChatGPT or Gemini what you're paying and what you want, and ask it to write you a short call script:** > "I pay [amount] a month for [phone/internet/insurance] with [company]. I've been a customer for [X years] and haven't had a discount in a while. Write me a short script for calling to ask for a lower rate — what to say to open the call, what to ask for, and two things to say if the first person tells me no." Read the script over a couple of times before you dial. You don't need to follow it word for word — just have the shape of the conversation in your head so you're not caught off guard. Ask the AI to also suggest a specific number to ask for, and a line to use if you're offered a "downgrade" instead of a discount. **One thing to watch:** the AI doesn't know your account history or your provider's current offers, so treat any numbers it suggests as a starting point, not a guarantee — and never share your account password, PIN, or full card number with the AI while drafting this. If the first person says no, asking to speak to the "retention" or "loyalty" department (and mentioning you're considering cancelling) often gets you a different answer. Next time a renewal notice or a bill increase lands, spend two minutes getting AI to write your script first — a five-minute call with the right words could save you real money every month, not just once. --- # Ask AI to run a mock job interview with you before the real one Source: https://aiforregulars.com/tips/practice-job-interview-with-ai You've got an interview coming up and the nerves are already creeping in. Reading a list of "common interview questions" online only gets you so far — you need to actually say your answers out loud, the way you will in the room. A chatbot can be your practice partner, any time of day, for free. **Open ChatGPT or Gemini (its voice mode works especially well for this) and ask:** > "I have a job interview for [role] at [type of company] on [date]. Act as the interviewer and ask me one question at a time. After each answer I give, tell me briefly what worked and what I could improve, then ask the next question." Answer out loud if you can — typing works too, but speaking builds the kind of confidence you'll actually need in the room. Go through five or six questions and you'll start noticing your own patterns: rambling, forgetting to give examples, or burying the good part of your answer at the end. This works whether it's your first interview or your fifteenth — and it's just as useful for a nerve-wracking phone screening or a performance review. Once you're feeling steadier, ask it to play a deliberately tough interviewer, so the real one feels easy by comparison. **One thing to watch:** the AI doesn't know your real work history unless you tell it, so its feedback is only as good as the details you give it — mention your actual experience and achievements, not just the job title. And treat its questions as practice, not a script: memorizing word-for-word answers will make you sound stiff when the real interviewer asks something slightly different. Next time an interview lands on your calendar, run through it with AI first — a few minutes of practice can be the difference between freezing up and walking in ready. --- # Ask AI to translate any official letter into plain English Source: https://aiforregulars.com/tips/rewrite-official-letter The next time you receive a letter from the government, your bank, your insurer, or any organisation that loves long sentences and small print — try this before you call anyone or worry: **Copy the whole text, paste it into ChatGPT or Gemini, and type:** > "Please explain what this letter is asking me to do, in plain English. What action (if any) do I need to take, and by when?" That's it. In most cases you'll get a clear two-paragraph summary: here's what happened, here's what you need to do, here's the deadline. This works equally well for rental agreements, medical reports, council notices, and T&Cs you'd normally skip entirely. **One thing to watch:** AI won't know your personal circumstances. If the letter is about something serious — a legal dispute, a large sum of money, a health decision — use the summary to understand the basics, then confirm the key details with the relevant professional. For most everyday letters, though, you'll have your answer in under a minute. --- # Ask AI to summarize a contract before you sign it Source: https://aiforregulars.com/tips/summarize-contract-before-signing You're about to sign something — a lease, a gym membership, a phone contract, a freelance agreement — and it comes with pages of small print you're expected to just accept. Most people skim it, spot nothing alarming, and sign anyway, because reading legal language carefully takes real effort. A chatbot can do that reading for you in under a minute. **Copy the full text of the document (or take clear photos of each page and upload them) into ChatGPT or Gemini, and ask:** > "Read this contract and summarize it in plain language. Tell me: what am I agreeing to pay and when, how do I cancel it, what happens if I'm late or want to leave early, and is there anything unusual or one-sided in here compared to a normal contract like this?" Ask follow-up questions about anything that's still not clear — "what does 'auto-renewal' mean here exactly?" or "how much notice do I have to give to cancel?" The AI won't get impatient, and you can ask the same question three different ways until the answer actually makes sense to you. **One thing to watch:** treat this as a way to spot what to ask about, not as legal advice. For anything with serious money or long-term consequences attached — a home purchase, a business contract, a divorce settlement — have an actual lawyer look it over too. AI is good at flagging the clause you'd have skimmed past; it's not a substitute for professional advice when the stakes are high. Next time a contract lands in your inbox or on your doormat, paste it into AI before you sign — two minutes of reading could save you from a surprise fee or a commitment you didn't mean to make. --- # Too long to read? Ask AI for the short version Source: https://aiforregulars.com/tips/summarize-long-email-or-chat A long email from the council, a school newsletter, or a family group chat with 80 unread messages: somewhere in there is the one thing you actually need to know. Chatbots such as ChatGPT and Gemini are good at finding it for you. **Copy the text, paste it into ChatGPT or Gemini, and ask:** > "Summarize this in three short sentences. Then list anything I need to do, with dates, and tell me whether I need to reply." You can keep going with follow-up questions: "What time does it start?" or "Write a short, friendly reply saying I can come." Always read the original for anything with a deadline or a payment. **One thing to watch:** A summary can miss a detail or get a date wrong. Check dates, amounts, and names against the original before you act. Also, do not paste passwords, bank card numbers, or other people's private details into a chatbot. Next time a message looks like homework, paste it in and ask for the gist. It takes half a minute and saves you the reading. --- # Photograph a menu or sign and let AI translate it Source: https://aiforregulars.com/tips/translate-photo-menu-sign-with-ai You are on holiday, or helping a relative, and you are holding a menu, a train notice, or a medicine box in a language you cannot read. Typing it into a translator letter by letter is slow. Chatbots such as ChatGPT and Gemini can read a photo directly, so you can skip the typing. **Open the ChatGPT or Gemini app, tap the camera or photo button, take a picture, and ask:** > "This is a photo of a [menu / sign / label]. Translate it into English, and explain anything that is not obvious. If it is a menu, tell me which dishes contain meat, fish, or nuts." Because the chatbot can answer follow-up questions, you can keep going: "Which of these is the cheapest vegetarian dish?" or "Is this platform for my train?" Hold the phone steady, in good light, and make sure the whole text is in the frame. **One thing to watch:** A translation can be wrong, especially with handwriting or blurry photos. For allergies, medicine doses, and legal or safety notices, treat the answer as a first look and confirm with a pharmacist, a member of staff, or the official source. Next time you meet a page of words you cannot read, take a photo and ask. It takes half a minute and saves a lot of guessing.