Analysis
Amazon has blocked Meta's Muse AI agent from completing purchases on Amazon.com, escalating a dispute over who controls the shopping experience when an AI agent acts on a customer's behalf, according to GeekWire and Forbes. By Sunday evening, Amazon was showing anyone attempting a Muse-driven checkout a popup reading: "Continued access by an unauthorized AI agent violates Amazon's Conditions of Use, to which our customers have agreed."
What Amazon Says Muse Is Doing
Amazon says it approached Meta before blocking the agent, asking the company to keep Amazon.com out of Muse's scope. Amazon's specific complaint is more than a competitive objection: it alleges Muse conceals its identity while navigating the site and appears to collect and retain customer credentials, potentially giving it access to account pages and order history without Amazon's knowledge or consent. Meta disputes the characterization, saying Muse has no visibility into passwords or payment methods and keeps any shared credentials in secure storage.
“## What Amazon Says Muse Is Doing Amazon says it approached Meta before blocking the agent, asking the company to keep Amazon.com out of Muse's scope.”
Not Amazon's First Agent Fight
Meta launched Muse earlier this month as a personal AI agent capable of sending emails, booking travel, filling out forms and making purchases on a user's behalf -- part of Meta's broader push to catch up in the agentic-AI race against OpenAI and xAI's Grok. This is at least the second time recently that Amazon has moved against a third-party shopping agent: the company also won a temporary court order blocking Perplexity's shopping bots. Two enforcement actions against two different AI agents in a short span shows Amazon treating agentic checkout access as a chokepoint it controls and intends to defend, not a one-off dispute specific to Meta.
The Agent Commerce Land Grab
The underlying fight is about who owns the customer relationship once an AI agent, rather than a human, is doing the browsing and buying. If agents like Muse can shop across any retailer's site on a user's behalf, Amazon's own recommendation engine, advertising business and first-party checkout data all become less valuable -- an AI agent that completes a purchase without ever rendering Amazon's product pages or ads to a human eyeball breaks the exact mechanics Amazon's retail media business depends on. That gives Amazon a direct commercial incentive to police agent access closely, on top of the security concerns it's citing publicly.
What The Dispute Misses
Meta's rebuttal -- that Muse never sees passwords or payment credentials directly -- hasn't been independently verified, and Amazon hasn't published technical evidence of the identity-concealment and credential-retention it's alleging either. Both companies have strong incentives to frame this dispute favorably: Amazon to protect its retail data moat, Meta to defend a flagship agent product launched just weeks ago. Neither account should be taken as the full picture until a neutral security researcher examines Muse's actual data-handling behavior.
What Comes Next
For any startup building an AI shopping or checkout agent, this is now a live case study: build too aggressively into a retailer's site without a negotiated data-sharing agreement, and you risk a public block days after launch, regardless of how large your parent company is. Whether Meta negotiates a formal access agreement with Amazon, routes Muse around Amazon entirely, or escalates into a legal fight over the Conditions of Use language is the next thing to track -- and whichever resolution comes first will likely become the template every other agent-commerce startup builds around.