AI Agents: Perplexity Wins First Amazon Battle

Amazon has just lost an important battle against Perplexity. On August 4, 2026, an American court of appeals indeed overturned the ban targeting Perplexity’s AI assistant on Amazon. A decision that could redefine the role of AI in global online commerce!

The American Judiciary Gives the Advantage to Perplexity’s AI: Here’s Why

The legal debacle between Amazon and Perplexity dates back to November 2025. The online retail giant accused the latter notably of accessing its customers’ accounts without authorization and violating the Computer Fraud and Abuse Act (CFAA). This is a US federal law against computer hacking.

In March 2026, California judge Maxine Chesney ruled in favor of Amazon by granting an injunction. This first ruling mentioned strong elements supporting a possible violation. She considered that Perplexity accessed password-protected accounts without authorization.

On August 4, 2026, a turnaround! The Ninth Circuit Court of Appeals overturned Judge Chesney’s decision by applying the rule of strict interpretation, favorable to the accused in case of doubt on a criminal law. Federal judges thus recognize that there is little case law on the liability of autonomous agents controlled by artificial intelligence.

And not only that! According to them, Amazon also failed to prove that Perplexity violated the CFAA. They consider indeed that it is the users themselves (not Perplexity’s AI) who access Amazon’s servers when using the Comet browser.

This Decision Marks a Decisive Turning Point for the Future of AI Agents in Online Commerce

The showdown between Amazon and Perplexity illustrates a broader tension between e-commerce platforms and new AI tools capable of autonomously acting on the web (notably through conversational AI assistants). Moreover, analysts agree on one point: the recent decision of the Ninth Circuit Court of Appeals goes beyond just the Perplexity case.

It indeed outlines a legal framework for all developers of agency AI. But above all, the federal judges’ interpretation allows a clear distinction between an autonomous AI agent and a malicious hacker.

The verdict explicitly cites the clarity rule: the CFAA remains first and foremost a criminal text. Extending it to a legitimate AI agent would therefore criminalize everyday behaviors. Which is absurd! Analysis: a user using Comet does not become an accomplice of hacking. A legal nuance that can only safeguard the digital ecosystem from authoritarian abuses.

Note that the Comet AI allows Amazon users to browse, log in, and make purchases using their own credentials. These are stored locally.

Despite This Victory, the Amazon VS Perplexity Case Is Far from Over

The case returns to district court. Amazon’s claims related to trademark and California law remain valid there. In this context, an Amazon spokesperson indicated the company is considering taking the case to the Supreme Court.

The court of appeals also specifies that its decision concerns only the preliminary injunction as well as the notion of access within the meaning of the CFAA. Other courts of appeals could therefore adopt a different reasoning. Amending the law by the US Congress is also an option to consider in order to better regulate AI algorithms.

Even the DeFi sector is closely watching this case. The fact is that AI agents are beginning to execute orders on crypto exchanges and manage digital wallets on behalf of users. This makes the question of liability of these artificial intelligence systems particularly important.

One thing is certain: Perplexity has just won a round that is sure to count in the history of agency AI. Above all, this ruling marks a step forward in regulating artificial intelligence applied to commerce. Enough to outline the contours of a future where the AI agent will be omnipresent.