Meta's AI agent Muse can no longer shop on Amazon. Since Sunday night, users who try to buy through it get a message that "continued access by an unauthorized AI agent violates Amazon's Conditions of Use, to which our customers have agreed."
GeekWire first spotted the block. Meta launched Muse in the United States on September 8, and buying things on a user's behalf is one of its headline features. Payments run through Stripe's Link service, and every purchase requires the user's approval.
An agent that acts
Meta describes Muse as an agent that carries out tasks rather than just answering questions. According to the company, it can send emails, book travel, fill out forms and keep working after the user closes the app. It runs on a model Meta calls Muse Spark, inside an isolated virtual machine, and Meta says it cannot see users' passwords or payment methods.
Why Amazon might say no
TechCrunch points to liability as one possible reason. If an agent places a wrong order, Amazon is the one handling the dispute with both the customer and the seller. Muse has one of the lower rates of made-up answers among AI models, the outlet notes, but that rate is still far from zero.
Amazon also has its own foundation models, AWS Nova, and its own inference platform, Bedrock. It has little reason to make life easier for a rival's agent unless it is required to.
Meta's response was not included in the TechCrunch report.
