A new kind of turf war is under way between the companies that sell things online and the companies that want AI assistants to buy things for us. Amazon has blocked Meta’s Muse, a personal AI agent that launched earlier this month, from shopping on Amazon’s website for its users, according to reports by GeekWire, The Verge and Engadget.
What people are seeing
From Sunday 20 September, people asking Muse to shop on Amazon started seeing a pop-up saying that continued access by an unauthorised AI agent violates Amazon’s Conditions of Use, which its customers have agreed to. Amazon told GeekWire that it first asked Meta to exclude Amazon’s site voluntarily. In a statement, an Amazon spokesperson said that third-party applications that offer to make purchases on behalf of customers from other businesses should operate openly and respect service providers’ decisions about whether to take part.
According to GeekWire, Meta had not told Amazon that Muse would use its store. Amazon is also reported to be frustrated that Muse does not identify itself when it browses the site, and it believes Muse poses privacy and security risks, including access to and retention of customers’ information.
What Muse is
Meta launched Muse as a “secure, private personal AI agent” that proactively helps with people’s goals and suggests ideas. Engadget reports that it can fill out forms, send emails and shop using a generated single-use card from Link by Stripe. Its Mac app can also connect to the Messages, Calendar and Notes apps, The Verge reported.
Meta rejects the security concerns. In its launch announcement, the company said Muse has no visibility into people’s passwords or payment methods, and that credentials a person shares go into secure storage so that Muse can use them without seeing them. The Verge’s reviewer used Muse to buy tank tops on Amazon during testing while other products were in the basket.
A privacy question that came up last week
Separately, The Verge reported that a user, Jason Aten, posted screenshots suggesting Muse had referred to the content of his messages even though he says he had not given it access. Meta’s David Singleton replied that the feature is opt-in, that Muse does not watch notifications on a Mac and only syncs Messages data after a user enables it, and that Muse had given an incorrect explanation of its own workings. He apologised for that answer and said Meta is working to improve how Muse describes its own functions.
Amazon’s wider strategy
The Verge described this as Amazon’s latest effort to fend off rival agentic AI services and keep customers shopping directly on its platform. Amazon sued the AI search company Perplexity in November last year to keep its store out of Perplexity’s Comet shopping experience, though a judge sided with Perplexity in August. The Verge also noted that since July, Amazon’s order-confirmation emails have listed fewer specifics, omitting item names and product images, which limits the information that outside AI services can pull from them.
GeekWire framed the standoff as part of a larger industry fight over who controls the shopping experience and the customer relationship in agentic commerce.
What it means for shoppers
For now, the practical effect is narrow: people who use Muse can’t use it to buy from Amazon. But the questions behind it will come up again for every AI assistant that acts on our behalf. Should a store be allowed to refuse an agent that a customer has chosen to use? What must an agent disclose about itself? And who is responsible if something goes wrong when an agent holds a payment method? Neither company has said the dispute is settled; Engadget and The Verge said they had asked Meta for comment.
