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 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, walks through those questions in more detail.