Amazon blocks Meta’s Muse from shopping on its platform
Amazon has stopped Meta’s Muse AI agent from shopping on Amazon, exposing a split over how retailers will handle AI-led purchases.
By Jordan Bell · Startups & Deals Reporter
· 3 min read
Amazon blocks Meta’s Muse from browsing and buying products on its platform, a move that limits a prominent new AI shopping tool less than two weeks after its launch. For investors watching AI commerce, the dispute shows that an agent’s usefulness may depend on whether retailers allow it into their sites and checkout systems.
Amazon told users that continued access by an unauthorized AI agent violates its Conditions of Use, according to GeekWire. The company said third-party apps that purchase from other businesses on customers’ behalf should operate openly and respect a provider’s decision on whether to participate.
Muse is Meta’s personal AI agent, launched Sept. 8 to perform multistep tasks such as completing forms, booking services and shopping. Meta says Muse can use a browser to access a service when that service does not provide a public application programming interface, or API, which is a formal software connection between services.
The immediate consequence is narrow but clear: Muse users cannot use the agent to shop on Amazon. The reports do not say that Muse has been shut out of shopping across the broader web.
Why did Amazon block Meta’s Muse?
Amazon said it had not authorized Muse to access customer accounts, collect data or process transactions, according to Axios. It also raised customer-security and user-experience concerns.
GeekWire reported that Amazon alleged Meta did not tell it Muse would access Amazon, that the agent did not identify itself while browsing, and that it appeared to capture and store customer credentials. Those are Amazon’s allegations, not independently established findings.
Meta has offered a different description of its safeguards. The company previously said Muse cannot see users’ passwords or payment methods, and that credentials users provide are placed in secure storage so the agent can use them without viewing them, GeekWire reported.
What does the dispute mean for AI shopping?
The clash is an early test of agentic commerce, where software does more than recommend products and instead carries out steps toward a purchase. Amazon’s position is that participation should be optional for the retailer. That could leave consumers with different shopping options depending on which sites agree to work with a given agent.
Amazon has taken similar positions toward other outside tools. CNBC reported that it sued Perplexity and blocked agentic tools from OpenAI and Google. In the Perplexity dispute, the Ninth Circuit’s August decision left possible claims based on contracts and terms of service, GeekWire reported. That case did not settle the broader legal questions around AI agents accessing retail sites.
Other companies have described a consent-based approach. OpenAI says merchants using its Agentic Commerce Protocol can accept or decline orders while remaining responsible for fulfillment, returns, customer support and communication. Walmart said it plans to let customers shop Walmart through ChatGPT using OpenAI’s Instant Checkout, while also developing its own AI agent, Sparky.
Those examples show different approaches rather than a uniform retail policy. Consumer adoption is also unsettled: Coveo’s 2026 Commerce Relevance Report found that 16% of surveyed shoppers were comfortable letting an AI assistant both find and buy products for them, according to Axios.
For now, Amazon’s block demonstrates the practical constraint on browser-acting agents: a retailer can decide not to take part. Similar conflicts could arise in travel, reservations and other online services as AI tools attempt to complete more tasks across the web.
This story draws on original reporting from CNBC.
