Perplexity Amazon Ninth Circuit Ruling

There's now a federal court ruling on who's liable when your AI agent shops for you, and it's the opening argument in a much bigger case: who's responsible when any autonomous system acts on your behalf.

On Tuesday the Ninth Circuit vacated Amazon's injunction against Perplexity's Comet shopping agent, ruling that it's users, not the AI company, who "access" a website under the Computer Fraud and Abuse Act. Amazon had argued Comet violated the 1986 anti-hacking law by scraping product data and completing purchases through customers' own accounts. The court disagreed, applying a legal principle called the rule of lenity to construe the statute's ambiguity against liability.

Buried in the opinion is the real headline. The panel wrote there is "little to no existing caselaw" on how to assign responsibility for AI agents under the CFAA. Courts are building AI agent law in real time, and this is the first federal appeals precedent to do it.

For any business building or deploying agents that act on behalf of customers, the ruling itself matters less than the line it draws: between the agent and the person directing it, under one of the broadest anti-hacking statutes on the books. That line will be tested again immediately. The case returns to district court, and Amazon's trademark and state-law claims survive.

Enterprises racing to ship agentic features shouldn't read this as a green light. Read it as the first data point in a legal framework still being written one case at a time.

If your company's AI agent completes an action a human user requested, who should bear legal responsibility, the agent's operator or the end user?