Product2 publishers3 min readPublished
Amazon shuts Meta's Muse out of its store with a terms-of-service popup
Amazon has cut Meta's Muse off from shopping its store, using the Conditions of Use its own customers accepted. Anyone shipping a buy-on-behalf feature inherits the merchant consent problem that comes with it.
The Product Desk · Product desk
What happened
- Amazon says it cut Meta's Muse agent off from shopping on Amazon.com for customers, after failing to get Meta to voluntarily exclude the store from the assistant's experience.
- Amazon's complaint is that Muse browses without identifying itself and appears to capture and store customer credentials, and that Meta never told Amazon the agent would access the store.
- Meta had said before the block that Muse has no visibility into people's passwords or payment methods.
- Amazon generated more than $68 billion in ad revenue last year, a business it says depends on people browsing its pages and seeing sponsored products.
Compiled by The Product DeskSomething wrong?How this is made
Why it matters
- constraint Merchant permission becomes a build dependency for agent features, because a retailer can revoke access overnight with a popup and nothing obliges it to warn the vendor first.
- decision Agent vendors have to pick a lane: browse anonymously inside the user's logged-in session, or identify the agent and negotiate store by store.
- precedent The contract theory works in production without a judge, so any retailer whose shoppers accepted terms can copy the approach at a fraction of litigation cost.
- contradiction Amazon describes credential capture and Meta describes credentials Muse uses without seeing them, and a team evaluating agent vendors has no way to check either account from outside the virtual machine.
A shopper who asked Muse to place an Amazon order on Sunday night got a notice instead: "Continued access by an unauthorized AI agent violates Amazon's Conditions of Use, to which our customers have agreed" [3]. Meta never signed that agreement. The shopper did.
The block is built on that sentence. Meta says Muse runs on a secure virtual machine with its own browser and checks with the user before sensitive actions like sending an email or making a purchase [11]. Muse does its shopping inside the user's session, as the user. In the Perplexity case, the Ninth Circuit held on Aug. 4 that the user and not the AI company was the party accessing Amazon's computers under federal anti-hacking law [15]. So the message Muse users see does not accuse anyone of hacking; it cites the Conditions of Use [17].
The timing is tight. Meta launched Muse on Sept. 8 as a personal agent for multi-step tasks, connected to email, calendar, payments, dining and shopping [9]. Two days later, on Sept. 10, the court denied Amazon's petition for rehearing, leaving contract and terms-of-service claims as the avenue still open to it [16][23].
On credentials, the two companies are describing different things. Amazon says Muse can reach account pages and order history when a customer prompts it [12], and that an agent which does not announce itself amounts to an undisclosed third party moving through customer accounts, processing transactions and handling sensitive data without Amazon's knowledge or consent [13]. Meta has said that credentials a user shares "go into secure storage, so Muse can use them without seeing them, including passwords a person types into the browser themselves" [7]. Amazon objects to the storing and to the session it cannot see into. GeekWire reported that Meta did not immediately respond to a request for comment on Sunday night [8].
Amazon's own comparisons are food delivery apps and the restaurants they take orders for, and online travel agencies and the airlines they book [14]. Both arrangements run on a contract between the two sides. Amazon's advertising revenue works out to more than $186 million a day [22], and an agent clicking through a reorder skips the sponsored results a shopper would scan.
These two already do business with each other. Amazon products have been purchasable inside Facebook and Instagram since 2023, and Meta signed a multibillion-dollar deal in April to run agentic AI workloads on Amazon's cloud [20]. Amazon said on Sunday it is in direct conversation with Meta, and declined to comment on whether it would take legal action [21]. Its spokesperson said: "Agentic third-party applications such as Muse have the same obligations, and we've requested that Meta remove Amazon from the experience" [5].
If you are building a buy-on-my-behalf feature, sort every destination on two questions: whose account the agent signs into, and whether the store has said yes in writing. Identified agent with an agreement is the delivery-app arrangement Amazon endorsed. Identified agent without one gets filtered by name, and you learn that during integration instead of after launch. An agreement plus an anonymous session is a support problem you volunteered for. An anonymous session with no agreement is where Muse was on Sunday night, and Amazon has run this play before: over the past year it sued Perplexity over the Comet browser and moved to block shopping agents from Google and OpenAI [19]. An Amazon spokesperson said: "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" [4].
What to watch
- A contract or terms-of-service complaint against Meta would test the one theory the Ninth Circuit left standing.
- Whether Meta removes Amazon from Muse's shopping scope, as Amazon's spokesperson requested, or leaves the popup in place.
- Whether Amazon publishes terms for identified agents, which would give vendors a route in that does not depend on the user's session.