Amazon has sued Perplexity AI, accusing the startup of using its autonomous browser agent, Comet, to make purchases on behalf of users without authorization.
The e-commerce giant alleges that Comet violated its site rules and committed computer fraud by concealing when an AI, not a human, was behind the transaction.
According to Bloomberg, the case, filed in a San Francisco federal court, seeks to block Perplexity from allowing Comet to shop on Amazon, marking one of the first legal tests of how far so-called agentic AI can act on behalf of real users.
Inside Amazon’s lawsuit
Amazon says Comet masked itself as a standard Chrome browser and kept transacting even after it was blocked. That conduct, the company argues, flouts its platform rules and runs afoul of federal computer-misuse laws.
The retailer also contends the setup “degrades the shopping experience” and creates privacy risks because customers weren’t told an automated system was acting in their name.
Amazon spokesperson Lara Hendrickson said, in an emailed statement to Bloomberg, that third-party assistants must “operate openly” and respect platform boundaries.
The company is asking the court for an order stopping Comet-enabled purchases on its site, casting the dispute as a fight over transparency and control.
A ‘bullying tactic’?
Perplexity hit back, calling Amazon’s lawsuit “a bully tactic” meant to scare off smaller competitors. In a blog post, the startup said Amazon was trying to block users from choosing their own AI assistants and protect a business model built on ads, not customer choice.
The AI company argued that Comet doesn’t scrape or mine data but acts as a “user agent” — an AI performing only the tasks a person directs it to do. “Your AI assistant must be indistinguishable from you,” Perplexity wrote, insisting that users should have the right to let an assistant act in their place.
It accused Amazon of eliminating user rights to protect its advertising revenue, citing CEO Andy Jassy’s comments about the value of sponsored listings. “They’re more interested in serving you ads,” the post read, “than letting you decide how to shop.”
Perplexity closed its response with defiance: “The future of AI, like all technology, is for people.” The company said it would not back down from offering tools that empower users, framing the case as a fight over who controls the next generation of the web.
The first real test for autonomous agents
The case is the first time a court will weigh how far an AI agent can act on a person’s behalf, and at what point autonomy turns into overreach.
Bloomberg noted that the dispute could set a precedent for how far agentic AI can go in performing tasks for users.
If Amazon wins, platforms could gain greater power to restrict how outside AI tools access their systems. For companies building autonomous tools, it could redefine what it means for a machine to act independently.
From allies to adversaries
Despite the courtroom clash, Amazon and Perplexity remain deeply linked. The startup runs much of its infrastructure on Amazon Web Services and has committed hundreds of millions of dollars to the platform. Amazon-founder Jeff Bezos is even one of its investors, a reminder of how tightly connected today’s AI players are behind the scenes.
Amazon, meanwhile, has been developing its own AI shopping tools, including Rufus, an in-app assistant that recommends products, and Buy For Me, a feature still in testing that can place orders directly with brands. Those projects hint at why the company is drawing firmer lines around who gets to automate its marketplace.
Perplexity is also facing a separate legal challenge, this time from Reddit, which has accused the company of pulling content from its platform without permission.