Amazon blocks Meta’s Muse from shopping on its platform

Amazon blocks Meta Muse from browsing and buying products on its platform, setting an early test for AI agents designed to complete online transactions for consumers. Amazon said the agent was unauthorized and that third-party applications making purchases for customers must operate openly and respect each service provider’s decision on whether to participate.

The move, reported around September 20 and 21, comes less than two weeks after Meta launched Muse on September 8. Meta describes Muse as a personal AI agent that can carry out multi-step tasks across services, including shopping, email, calendars, payments and dining. The company says the agent can connect through a public API when one exists or use a website through a browser when one does not.

For retailers, the dispute turns on permission as much as technical capability. An outside agent may act on a shopper’s instruction, but the retailer still controls the customer account, its conditions of use, access to its site and its checkout process. It also has an interest in retaining the direct shopping interaction, including the opportunity to present products and advertising in its own interface.

Why did Amazon block Meta’s Muse?

Amazon said it had not authorized Muse to access customer accounts, scrape data or process transactions, according to Axios. The company cited customer security and user-experience concerns and compared its position with food-delivery and travel-booking services that operate with participating merchants’ agreement.

Amazon also alleged that Muse did not identify itself while browsing and raised concerns about the handling of account credentials, GeekWire reported. Those points remain disputed. Meta had previously said Muse cannot see a user’s passwords or payment methods, and that credentials supplied by a user are placed in secure storage for the agent’s use.

Users attempting to shop with Muse on Amazon were shown a notice stating that continued access by an unauthorized AI agent breached Amazon’s Conditions of Use, according to GeekWire. Meta declined to comment to CNBC.

What does the dispute mean for AI shopping agents?

The episode indicates that broad automation may depend on commercial agreements rather than a consumer’s instruction alone. Amazon has previously acted against agentic tools associated with Perplexity, OpenAI and Google, CNBC reported. In its dispute with Perplexity, a preliminary injunction obtained by Amazon was later reversed by the Ninth Circuit, according to GeekWire. That report said the decision left potential claims based on contracts and terms of service available, rather than establishing a definitive right to exclude user-directed agents.

Other retailers have taken a different approach. Walmart said it was working with Google and OpenAI to enable discovery of Walmart products through their agentic shopping interfaces, while also developing its own assistant, Sparky, Axios reported. The contrasting approaches suggest retailers may decide individually whether outside agents can enter their shopping flows and on what terms.

Consumer acceptance is also unsettled. A Coveo survey cited by Axios found that 16% of shoppers were comfortable allowing an AI assistant both to find and purchase products. The Amazon-Meta clash therefore shows a developing commercial constraint on agentic commerce, rather than a settled model for how such purchases will work.

This story draws on original reporting from CNBC.