Amazon gets temporary relief from Perplexity’s AI

Amazon has won a temporary court order to block AI startup Perplexity’s shopping bots from its online store.

A San Francisco federal court this week put a hold on Perplexity’s Comet browser agent from making purchases on behalf of people shopping on Amazon’s marketplace. Comet is an AI-powered browser that comes with a chatbot that can view and interact using an Amazon customer’s account. Amazon sued Perplexity in November 2025, alleging “unauthorized use” of Perplexity’s Comet web browser’s “agentic AI functionality to access password protected sections of its website.”

While agentic AI has changed the way people search for items online, it has not managed to generate the same kind of buzz when it comes to online shopping. The discrepancy matters because it brings the spotlight back to how human connection remains essential in retail, even as AI agents handle more of the shopping experience.

Specifically, the San Francisco court order bars Perplexity from accessing customer data of Amazon’s “password protected” member accounts, and directs the company to destroy the Amazon data it previously collected. The ruling is stayed for one week to allow Perplexity to appeal.

If upheld, this ruling could potentially restrict AI agent permissions and set a precedent around AI commerce. As agentic capabilities start to take real shape with big announcements from Google and Walmart, AI agents cropping up seemingly everywhere has become a talking point in retail circles.

The injunction holds while both companies battle in court, a Perplexity spokesperson told Bloomberg, on March 10, “It will continue to fight for the right of internet users to choose whatever AI they want.” 

That same day, Amazon told Bloomberg blocking Perplexity’s “unauthorized access” is an “important step” and the company will continue to defend its case in court.

“At the same time that expectation for digital adoption will grow and come to fruition, the craving of human input and human interaction won’t stop either,” Kassi Socha, Gartner’s senior director analyst of marketing, previously told Retail Brew.

“This ruling is a classic example of the ecosystem’s concern around transparency in the age of AI,” Gilit Saporta, fraud researcher and expert at DoubleVerify, a cybersecurity company for digital media, wrote in an email. “The primary challenge in maintaining trust when facing sophisticated, AI-driven traffic is that AI agents are often evasive, mimic human behavior, and are able to operate at endless scale.”