US Court Overturns Amazon's Block on Perplexity's AI Shopping Assistant

A Federal First on AI Agent Legality

The decision from the San Francisco-based appeals court is the first time a US federal appeals court has weighed in on the question of whether AI agents may legally access online platforms on behalf of users. According to Handelsblatt, the ruling overturns a preliminary injunction that had prevented Perplexity from deploying its AI shopping assistant on Amazon's marketplace. The case sits at the intersection of long-standing US computer-crime law and a new generation of autonomous tools that browse and transact online with minimal human supervision, and the court used the opportunity to draw an early line.

The significance lies less in the immediate parties and more in the precedent. As Handelsblatt notes, the decision is "likely to have far-reaching consequences for the growing industry of AI tools" that perform internet tasks with little human oversight. For platforms and AI developers across the Atlantic — including German e-commerce players experimenting with their own agentic tools — the ruling provides an early signal of how US judges will read decades-old computer-misuse statutes against a very modern product.

The "User, Not Bot" Doctrine

The court grounded its reasoning in a clean conceptual distinction: it was not Perplexity itself, but the users of the AI agent, who accessed the Amazon platform. In the judges' view, that distinction forecloses a claim of unauthorized access under the relevant US computer-crime statute, because the access was initiated by a human principal rather than by the AI company as a trespasser. The ruling frames the AI agent less as a separate actor and more as a tool wielded by the user, in the same way a browser or a script extension is.

That framing matters because it sidesteps one of the harder doctrinal questions in the case — whether the scraping and interaction patterns of an autonomous assistant should be treated like those of a traditional bot. Amazon had framed Perplexity's tool as a violator of platform terms and US computer-misuse law; the appeals court, by contrast, located the legal actor in the end user. According to the article, Perplexity said it "will continue to fight for the right of internet users to use the AI of their choice," signalling the company will press the same argument in any further proceedings.

A Crack in the Platform Walled Gardens

Amazon publicly disagreed with the decision but said it remained confident in the further proceedings and was reviewing its next steps, according to Handelsblatt. That posture leaves open the prospect of further litigation, including a possible appeal or a renewed fight on different legal grounds. For now, however, the practical effect is that Perplexity's assistant can operate on Amazon's marketplace while the underlying case continues.

For the broader AI industry, the ruling loosens — at least temporarily — a key lever that platforms have used to keep autonomous agents off their properties. If the doctrine holds, platforms will have to rely more heavily on terms-of-service enforcement and technical countermeasures rather than criminal-style computer-misuse claims. That shifts the burden of gatekeeping from courts back to the platforms themselves, a dynamic that will resonate well beyond the Amazon–Perplexity dispute and into every online marketplace that must decide how to treat a bot acting in a shopper's name.