<a href="https://www.google.com/preferences/source?q=finance.biggo.com” rel=”nofollow noopener” target=”_blank”>Add preferred source
Amazon has blocked Meta’s Muse AI assistant from shopping on its marketplace, citing unauthorized access, failure to disclose its agent identity, and account security risks. The block took effect Sunday, with users seeing a warning that continued access by an unauthorized AI agent violates Amazon’s conditions of use. Amazon said it previously asked Meta to exclude its site from Muse but received no response. The dispute follows Amazon’s ongoing legal fight with Perplexity over its Comet browser, where a federal appeals court overturned an injunction in August. Amazon generated more than $68 billion in advertising revenue in 2025, and agentic shopping threatens that model by bypassing its pages. Meta launched Muse on September 8, and it quickly topped Apple’s U.S. App Store chart. The confrontation underscores the unresolved rules governing AI agents in e-commerce.
Key Elements
Amazon has cut off Meta’s new Muse AI assistant from shopping on its marketplace, escalating a broader conflict over whether third-party agents should be allowed to act on behalf of consumers inside e-commerce platforms.
Users who attempt to make purchases through Muse on Amazon now encounter a warning stating that “continued access by an unauthorized AI agent” violates the retailer’s conditions of use. The block took effect Sunday evening which first reported the move
Amazon said it had previously asked Meta to voluntarily exclude its marketplace from Muse’s reach but received no response. The company is now in direct communication with Meta, though it declined to say whether legal action is under consideration.
The confrontation is notable because the two companies maintain extensive commercial ties. Amazon products have been purchasable directly within Facebook and Instagram since 2023, and in April Meta signed a multibillion-dollar agreement to run agentic AI workloads on Amazon’s cloud infrastructure.
Why Amazon Pulled the Plug
Amazon’s objections center on authorization, identity disclosure, and account security. The company said Meta did not provide advance notice that Muse would access its platform, and that the agent does not identify itself while browsing. From Amazon’s perspective, this amounts to an unauthorized third party entering user accounts, processing transactions, and touching associated data.
The e-commerce giant also pointed out that Muse can access account pages and order history when prompted by a user. “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,” an Amazon spokesperson said.
Meta has pushed back on the security concerns. The company maintains that Muse cannot see users’ passwords or payment methods, and that any credentials shared by users go into secure storage where the agent can use them without viewing them directly.
The product design sits at the heart of the dispute. Muse can connect to services that offer public APIs through user credentials. For services without APIs, the agent is designed to operate “like a user through a browser.” Amazon’s block targets that second access method.
A Legal Battle With Mixed Results
Amazon has previously tried to use the courts to keep AI agents off its platform, with mixed outcomes. The company sued Perplexity over its Comet browser and won a preliminary injunction in March. But on August 4, the Ninth Circuit Court of Appeals overturned that ruling, finding that under federal anti-hacking law, the person accessing Amazon’s computer systems is the user, not the AI company. The court declined Amazon’s request for reconsideration on September 10.
That ruling still leaves room for claims based on contracts and terms of service. The warning displayed to Muse users does not invoke hacking statutes but instead points directly to Amazon’s conditions of use, signaling that the fight is shifting toward whether AI agents have platform authorization.
The Commercial Stakes
Amazon’s hard line on agentic shopping is tied to its business model. The company generated more than $68 billion in advertising revenue in 2025. If AI agents handle search, price comparison, and checkout on their own, users may never land on Amazon’s pages, threatening the display advertising and recommendation systems that drive that revenue.
Over the past year, Amazon has moved to restrict external AI agents beyond the Perplexity litigation, including efforts to block shopping agents from Google and OpenAI. At the same time, the company is building its own agentic shopping tools. Its “Alexa for Shopping” feature, launched in May, includes a “Buy for Me” capability that can purchase items from external brand websites on a user’s behalf. Amazon emphasizes that the feature discloses its identity and allows brands to opt out.
The emerging question is whether platforms can refuse third-party AI agents entry into their transaction systems, and whether they can maintain control over the customer relationship and transaction entry points.
Muse’s Rapid Rise
Meta launched Muse on September 8 as a general-purpose personal AI agent capable of executing multi-step tasks across email, calendars, payments, dining, and shopping. It is available on iOS, Android, muse.ai, and WhatsApp.
The app gained traction quickly, reaching the top of Apple’s U.S. App Store free-app chart within a week, ahead of ChatGPT. Meta says Muse runs in a secure virtual machine with a dedicated browser and requires user confirmation before sensitive actions such as sending emails or completing purchases. A monitoring agent called Sentinel reviews outgoing content.
Scott Galloway, an NYU Stern professor and entrepreneur, argued on the Prof G Markets podcast that Meta’s years of accumulated user data could give Muse an advantage over rivals building from scratch. He described Meta as “the largest first-party data miner in history” and said the personal context Meta has gathered through Facebook, Instagram, and WhatsApp could make Muse more useful than competitors that would need years to build similar context.
Galloway also suggested Meta may not need to win the race for the most advanced AI model. Polymarket traders give Meta roughly a 3% chance of having the world’s best model by the end of 2026, far behind Anthropic at around 72%. The argument is that consumers may care less about model rankings than about what an agent can actually accomplish for them.
That usefulness, however, depends on other companies allowing Muse to operate on their platforms. Amazon’s block demonstrates that even a widely adopted personal agent cannot assume the entire internet is open for business.
The conflict also highlights a trust gap. An Oppenheimer survey found that only 8% of U.S. consumers would trust Meta with their passwords, compared with 30% for Google and 16% for ChatGPT. Meta shares were trading around $711 on Monday morning, up roughly 16% since Muse launched.
As AI agents move from answering questions to taking real-world actions, the rules governing agentic commerce remain unsettled. Who controls the agent, who owns the customer relationship, which services agents may access, and how transaction value gets distributed are all questions that will shape the next phase of online retail.
Once added, BigGo Finance appears first in Google Search Top Stories, so you get the broadest, most up-to-the-minute, and most comprehensive global financial news first.
