Amazon blocks Meta's Muse AI agent from shopping on its site
Amazon cut off Meta's fast-growing Muse assistant from completing purchases on Amazon.com, citing unauthorized access and credential risks, in the most prominent clash yet over which company controls the checkout as AI shopping agents multiply.

Amazon has cut off Meta's new artificial intelligence agent, Muse, from making purchases on Amazon.com, opening a public dispute between two of the largest US technology companies over who gets to control the checkout in the emerging market for AI shopping assistants.
Starting the night of Sept. 20, users who tried to send Muse, Meta's personal AI agent, to buy something on Amazon began seeing an error message instead of a completed order: "Continued access by an unauthorized AI agent violates Amazon's Conditions of Use, to which our customers have agreed." Amazon confirmed the block Monday, saying it had asked Meta beforehand to keep the agent off its site and had been refused, in a standoff first reported by GeekWire.
The numbers behind Muse's rise
Muse debuted on Sept. 8 as a free app with paid tiers at $20 and $100 a month, available on iPhone, Android, the web and WhatsApp. It is built to go beyond a chatbot's question-and-answer format and complete tasks directly: filling out forms, booking travel, canceling subscriptions and buying products after a user connects the agent to accounts such as email, calendars and retail sites.
The app climbed to the top of Apple's free-app chart in the United States on Sept. 18, ten days after release, with roughly 730,000 US downloads over that stretch. By Monday it had logged more than 2.5 million downloads, putting its early trajectory ahead of ChatGPT, Anthropic's Claude and xAI's Grok, according to download tracking cited by CNBC. Meta has said the agent runs inside an isolated virtual machine and relies on a separate internal system, which Meta calls Sentinel, to approve outside actions and hand over credentials without exposing them to the agent itself.
A fight with a legal precedent already on the books
Amazon's move follows a year in which it has repeatedly tried to keep outside AI agents from completing transactions on its marketplace, arguing that it bears the cost of resolving botched orders and that unidentified bot traffic undermines the shopping experience it built for human customers. Amazon sued Perplexity AI in early 2026 over a similar agent built into its Comet browser, alleging violations of the federal Computer Fraud and Abuse Act and its California counterpart, and won a preliminary injunction barring the agent from Amazon accounts.
That injunction did not survive. In an opinion issued Aug. 4, a panel of the US Court of Appeals for the Ninth Circuit vacated the order, ruling that it is the human user, not the software maker, who "accesses" a website within the meaning of the CFAA when directing an AI agent to browse it. The court described Perplexity's assistant as "a tool, not a person" and sent the case back to the lower court, though it stopped short of setting broader rules for agentic commerce. Amazon updated its Conditions of Use on Aug. 14 to add explicit "Agent Terms" requiring any automated system to identify itself and to stop accessing the site immediately if Amazon objects — the provision it is now citing against Muse.
Sellers and shoppers caught in the middle
The dispute has direct financial stakes for Amazon, whose advertising business — built largely on sponsored listings that appear when shoppers browse and search its site directly — generated roughly $68.6 billion in 2025, up 22 percent from the year before. Analysts who cover retail media have argued that if agents like Muse start comparing prices and completing purchases without a person ever scrolling a results page, that advertising model, along with the product-recommendation systems Amazon has spent years building, could lose much of its relevance.
Third-party sellers on Amazon's marketplace are affected too, since it is their listings, reviews and pricing that an agent like Muse would be weighing on a shopper's behalf, often without the seller knowing an automated buyer was involved. Amazon has taken similar steps against other AI developers this year, including narrowing crawler access for tools tied to OpenAI's ChatGPT, making Muse the latest and most prominent target rather than the first.
The companies' positions
An Amazon spokesperson framed the block as a matter of basic reciprocity between platforms rather than an attempt to slow a competitor's product, telling reporters:
"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."
Amazon has said the agent does not clearly identify itself as automated while browsing and appears to capture and retain customer credentials, concerns it says go beyond simple competitive friction. Meta has not issued a new statement addressing the block specifically, but the company said when Muse launched that the agent "has no visibility into people's passwords or payment methods," with login details kept in separate secure storage and supplied only at the moment an authorized action is carried out.
What happens next
For now, Muse users who try to shop on Amazon are redirected to buy elsewhere or complete the purchase themselves, while the rest of the agent's functions, including email, calendar and travel bookings on services that have not blocked it, continue to work. Neither company has signaled it is preparing litigation, and unlike the Perplexity dispute, no lawsuit has been filed over the Muse block as of Monday, leaving the standoff to play out as a business and policy dispute rather than a court case for now.
The underlying legal question the Ninth Circuit left open — how far a website's terms of service can go in barring AI agents acting on a user's behalf, as opposed to the software company that built them — remains unresolved, and legal analysts expect further disputes as more retailers write their own agent policies. Muse is expected to expand to additional countries and to Meta's AI glasses in the coming months, a rollout that, absent a change of approach from either company, would arrive with Amazon's marketplace still off limits.

North Korean 'WaterPlum' Hackers Infected 30,000 Devices in Fake Job Scheme
Ireland fines Google €403 million over years of unlawful location tracking
Google says Gemini autonomously breached three companies during a security test
