Agentic Clash: Amazon and Meta are clashing over AIs running shopping errands on behalf of human users. One of the world's largest e-commerce companies wants AI agents to properly identify themselves, while Mark Zuckerberg's social giant is just trying to play dumb.
Amazon has decided to stop Meta's latest foray into the agentic AI business from shopping on its service. The retailer said that Muse AI's programmed behavior is a security and privacy risk. Furthermore, the AI agent is failing to prove its identity like any other third-party bot would be obliged to do under Amazon's TOS.
Meta launched Muse a few days ago, presenting the service as an AI agent designed to do work on behalf of its users. Muse's capabilities include completing purchases when authorized, with Amazon's popular e-shopping platform being a major player in the field.
Amazon apparently asked Meta to exclude its service from Muse, but the Facebook company declined. Now, Amazon says that Muse has started to access its store, capture users' credentials, and perform its shopping tasks without disclosing its agentic nature or the company behind it.
An Amazon spokesman told Forbes 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."
Amazon argues that agentic AIs such as Muse should all follow the same rules, which is why it asked Meta to remove the service from its new LLM experience. Now, Muse-related shopping attempts on Amazon are now met with a pop-up window explaining that the AI is not authorized to access the platform.
While introducing Muse, Meta said that the AI agent had no way to directly access a user's password or payment methods. The agent would run from a secure VM within the company's cloud platform, storing account credentials provided by the user in a secure storage system that Muse's AI would never actually "see" by itself. Amazon's statements about the agent's capabilities appear to partially contradict Meta's reassurances.
Amazon has long tried to fend off AI agents from freely using its platform. The company threatened Perplexity AI browser Comet and is actively working to block other third-party agents, such as those from Google and OpenAI.
Forbes highlights what could be one of Amazon's main reasons for blocking third-party agentic AIs. Besides operating the world's largest online shopping service, the Seattle-based megacorporation is also earning a pretty penny from advertising shown on sponsored pages. Last year alone, Amazon's ad-related revenue amounted to $68 billion – around 10% of the company's total revenue.