A battle over the future of e-commerce has erupted between two Silicon Valley titans. Amazon.com Inc. blocked Meta Platform Inc.’s new artificial intelligence (AI) agent, Muse, from making purchases on its digital marketplace.
Beginning Sunday, users attempting to complete Amazon transactions via Muse were met with an explicit pop-up warning: “Continued access by an unauthorized AI agent violates Amazon’s Conditions of Use, to which our customers have agreed.”
The intervention marks a sharp escalation in an industry-wide power struggle over who dictates the online shopping experience when autonomous software acts on a consumer’s behalf.
The dispute centers on how Muse, which launched Sept. 8, interacts with third-party websites. Muse is designed to perform complex, multi-step actions such as booking travel, filling forms, and placing retail orders. To execute purchases on sites lacking a public interface, Muse navigates web pages using its own integrated browser, effectively acting as a proxy for the shopper.
Amazon argues that this setup allows Muse to browse the platform covertly without identifying itself as an automated system. According to the e-commerce giant, the agent appears to collect and store user login credentials, potentially granting Meta undisclosed access to sensitive account pages and order histories without Amazon’s explicit consent.
“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,” an Amazon spokesperson said, comparing the situation to food delivery apps or online travel agencies that operate via direct merchant agreements.
Amazon confirmed it requested that Meta voluntarily remove its marketplace from Muse’s scope but moved to block the software after Meta failed to comply.
Amazon said it is currently in direct discussions with Meta to resolve the conflict, though it declined to comment on potential legal action.
Meta has pushed back against privacy concerns. In launch materials, the company said Muse operates on secure cloud-based virtual machines overseen by a monitoring system called Sentinel. Meta maintains that Muse “has no visibility into people’s passwords or payment methods,” noting that any shared credentials are routed through encrypted storage so the system can utilize them without viewing them.
Amazon has a legitimate security argument, and a reasonable interest in knowing when an automated system is accessing customer accounts and initiating transactions, says Stephanie Walter, practice leader for AI Stack & Enterprise Application Development at HyperFRAME Research.
“The question is whether Amazon will create a fair path for third-party agents that meet those requirements,” she said. “Amazon’s Buy for Me already purchases products from other retailers’ websites, although Amazon says its agent identifies itself and allows retailers to opt out. If a competing agent meets comparable standards and Amazon still refuses access, the security argument starts to look more like a way to protect Amazon’s control of the customer relationship.”
“The industry needs standard ways to authenticate agents, limit their permissions, record their actions, and revoke access,” Walter said. “The long-term issue is whether the customer gets to choose an agent or the platform makes that choice for them.”
The clash highlights an awkward dynamic between the two companies, who remain key partners. Amazon goods have been directly purchasable on Facebook and Instagram since 2023, and Meta signed a major contract in April to run AI workloads on Amazon’s cloud infrastructure.
Yet the financial stakes surrounding control of the retail journey are massive.
Amazon generated over $68 billion in advertising revenue last year, a revenue stream tied directly to human shoppers manually scrolling through search results and viewing sponsored listings. Autonomous AI agents that bypass the traditional browsing interface threaten to disrupt that lucrative ad model.
Amazon has aggressively defended its ecosystem against outside automation. The company previously moved to restrict shopping agents from OpenAI and Google and engaged in legal action against Perplexity over its Comet browser.
While a federal appeals court recently ruled that consumers — not AI developers — are the entities that access sites under federal anti-hacking laws, the court left open breach-of-contract claims. Amazon’s new pop-up explicitly invokes its terms of service rather than anti-hacking statutes.
