Ninth Circuit Rules Agentic AI Browsers Don't Violate CFAA

Ninth Circuit Sets First Legal Boundary for Agentic AI Browser Tools

On August 4, 2026, the U.S. Court of Appeals for the Ninth Circuit issued the first federal circuit ruling on agentic AI and the Computer Fraud and Abuse Act (CFAA), vacating a preliminary injunction that had barred Perplexity's Comet Browser from accessing the Amazon Store.

What happened: Amazon sued Perplexity, arguing its AI assistant violated the CFAA by "accessing" Amazon's servers without authorization. A district court sided with Amazon. The Ninth Circuit reversed.

The court's core finding: Perplexity's servers never directly communicate with Amazon's servers. The chain is: the user's browser retrieves the page, the AI assistant analyzes it and sends screenshots to Perplexity's servers, and Perplexity's servers instruct the assistant on next steps. Under that architecture, it is the user who accesses Amazon's computers — not Perplexity.

Why the CFAA doesn't apply: The court emphasized that the CFAA is primarily a criminal anti-hacking statute. Expanding it to cover agentic AI browsing "would require a novel interpretation far afield from the statute's purpose." The rule of lenity also cut against finding liability, since doing so could expose ordinary users to criminal liability.

Key takeaways for AI practitioners:

• System architecture matters legally. If your agentic AI routes instructions through your own servers directly to third-party systems, liability exposure increases.
• The court condoned the Perplexity model: user and browser sit between your infrastructure and the target site.
• The ruling does not protect agentic AI from trespass to chattels, tortious interference, DMCA anti-circumvention, or contract claims.
• Operators should ask: who ultimately controls what the tool does, and do your servers communicate directly with third-party systems?

Read the full article on Wilson Sonsini