Can AI agents shop freely online? Amazon's move against Meta raises questions
Amazon says Meta's Muse assistant violates its conditions of use by not identifying itself while shopping
Amazon has moved to block Meta's newly launched Muse AI assistant from shopping on Amazon.com, citing a violation of the site's conditions of use. According to GeekWire, Amazon says the agent fails to properly identify itself while browsing and making purchases on behalf of users, prompting the company to cut off its access.
This situation has been used as a precedent to a bigger issue that the tech industry will have to face – whether AI agents must get permission from the platform itself to represent the user in the use of websites and applications.
Amazon is very interested in how this issue will turn out, due to the importance of the data that it gets from its shopping experience.
In an unusual move, GeekWire's hosts put the question directly to Muse itself, asking the AI assistant to respond to Amazon's block.
The assistant gave what was said to be a surprisingly well-balanced view of the situation, recognising the legitimacy of the concerns regarding identification and website rules while nevertheless supporting the wider argument for the benefits of agentic shopping technology.
This case gets at a fundamental issue in the development of artificial intelligence technology: agents are being created to browse the Web and perform tasks on their own, yet the websites on which they must act were designed for people and not robots controlled by people.
Whether platforms such as Amazon decide to enforce the separation between the two through identification requirements, restricted access, or terms of service will be precedent-setting in determining the degree of freedom AI agents have online.
