Ninth Circuit Rules on Perplexity’s Access to Amazon — First AI Agent Legal Ruling
For the first time, a federal appeals court has weighed in on a question that’s been quietly looming over the entire AI industry: when an AI agent shops, browses, or clicks on your behalf, who is legally “accessing” the website — you, or the software? This week, the Ninth Circuit Court of Appeals answered that question in a case between Amazon and Perplexity AI, and the ruling could shape how every AI shopping and browsing agent operates going forward.
Here’s the full story — what happened, what the court decided, and why this ruling matters far beyond just Amazon and Perplexity.
How This Legal Fight Started
This is a novel legal dispute, and Reuters reported that the San Francisco-based Ninth Circuit Court of Appeals determined Amazon was unlikely to succeed on its claim that Perplexity’s AI agents violated a federal computer-hacking law. But to understand the ruling, it helps to rewind to how this fight began.
Amazon originally filed the lawsuit against Perplexity in November 2025, alleging that Perplexity’s AI agent breached both the federal Computer Fraud and Abuse Act (CFAA) and California’s Comprehensive Computer Data Access and Fraud Act (CDAFA). The dispute centered on Comet, Perplexity’s AI-powered browser and shopping agent, which can browse and complete purchases on a user’s behalf.
Amazon’s core objection wasn’t really about hacking in the traditional sense — it was about control. Amazon’s attorney argued that Comet’s AI agent causes technological harm by collecting user data in a way that’s novel and unlike how other companies, including Brave and Microsoft Edge, interact with Amazon’s platform. Perplexity’s legal team pushed back hard on that framing. Perplexity’s attorney argued that Amazon was trying to stretch anti-hacking statutes originally meant to prohibit computer break-ins simply because Amazon wanted its own customers to access its website in the way Amazon preferred.
A district court initially sided with Amazon — Senior U.S. District Judge Maxine Chesney of San Francisco ruled in March 2026 that Amazon was likely to succeed on the merits of its CFAA claim, and granted a preliminary injunction blocking Perplexity’s tool from operating on Amazon’s platform. That injunction was later stayed by the Ninth Circuit while the appeal was under review, allowing Perplexity to keep operating in the meantime.
What the Ninth Circuit Actually Decided
This week, the appeals court sided decisively with Perplexity. A three-judge panel found that, on the current record, Perplexity itself did not access Amazon’s computer systems at all.
The reasoning behind that finding is the real heart of this case. The judges concluded that Amazon’s website is accessed by users, while Perplexity’s AI assistant operates simply as a tool acting on the user’s own instructions. Put simply, because Perplexity’s Comet agent only acts when a specific user tells it to, the court found that it’s the user — not Perplexity — who is legally accessing Amazon’s servers.
There was also a technical wrinkle the court addressed directly. The opinion noted that Perplexity may send screenshots of the Amazon pages a user accessed back to its own servers in order to complete a requested task, but the court found that this process doesn’t count as Perplexity itself accessing Amazon’s servers. The court reached the same conclusion on Amazon’s separate CDAFA claim under California law, finding Amazon was also unlikely to succeed on that allegation.
The Court Was Careful Not to Overreach
Interestingly, the judges didn’t try to settle every question about AI agents in one sweeping decision. The appeals court explicitly stated that its ruling is limited to the factual record currently before it, and does not establish broader legal principles regarding agentic AI or possible liability in other legal contexts. The case has now been sent back to the U.S. District Court for the Northern District of California for further proceedings consistent with the Ninth Circuit’s findings.
That caveat matters. This isn’t a permanent, industry-wide green light — it’s a preliminary ruling specific to this case’s evidence, and the underlying lawsuit itself is still ongoing.
Why This Ruling Matters Beyond Amazon and Perplexity
This case is being treated as a genuinely landmark moment, and for good reason. Reuters described the decision as the first ruling by a federal appeals court to address whether AI agents acting on behalf of users can legally access online platforms — a question with broad implications for the fast-growing field of agentic AI tools that browse, shop, and transact across the web.
The timing couldn’t be more relevant. PayPal, Google, Apple, and OpenAI have all developed or announced their own agent systems with purchasing capabilities, and until now, the legal framework governing how those agents interact with websites they weren’t explicitly invited onto had been largely undefined. This ruling gives the industry its first real signal — even if not a permanent one — about which way the wind might be blowing.
There’s also a competitive angle worth noting. Amazon has its own AI shopping assistant, Rufus, and this case has been read partly as an attempt by Amazon to wall off its marketplace from outside AI competitors trying to operate on its turf. A ruling that favors third-party AI agents over platform gatekeeping could reshape how retailers think about controlling access to their own sites going forward.
Amazon, unsurprisingly, isn’t backing down. An Amazon spokesperson said the company respectfully disagrees with the decision on the preliminary injunction, remains confident in its case, and is evaluating its next steps.
What This Means If You Build or Use AI Agents
If you’re building AI tools that interact with third-party websites, or you’re a business trying to figure out how to treat AI agents visiting your platform, a few practical takeaways stand out:
- The “who’s accessing what” question just got clearer — for now. Courts appear inclined to treat user-directed AI agents as an extension of the user, not as an independent, potentially unauthorized actor.
- This is not a blank check. The court’s narrow, fact-specific ruling means future cases with different evidence could go the other way.
- Platforms may lean on other legal tools instead. Since anti-hacking statutes appear to be a difficult fit for these disputes, expect companies like Amazon to explore terms-of-service enforcement, technical blocking, or new legislation as alternative strategies.
Conclusion
This ruling doesn’t end the broader legal debate around AI agents — if anything, it’s just the opening chapter. But it’s a meaningful one. By deciding that a user directing an AI agent is legally the same as that user browsing directly, the Ninth Circuit has handed the agentic AI industry its first real piece of legal footing, even while explicitly declining to turn it into a sweeping rule. As agents from OpenAI, Google, Apple, and others start doing more shopping, clicking, and transacting on people’s behalf, expect this case to be cited again and again as the legal system tries to catch up with how people are actually starting to use the web.
FAQs
Q1: What did the Ninth Circuit rule in the Amazon vs. Perplexity case?
The court ruled that Amazon was unlikely to succeed in its claims that Perplexity’s AI shopping agent violated federal and California computer access laws, finding that users — not Perplexity — are the ones legally accessing Amazon’s platform.
Q2: Why did Amazon sue Perplexity in the first place?
Amazon filed the lawsuit in November 2025, arguing that Perplexity’s Comet AI agent violated the Computer Fraud and Abuse Act and a similar California law by accessing and collecting data from Amazon’s platform without authorization.
Q3: Does this ruling mean AI agents can now access any website freely?
No. The court explicitly said its ruling is limited to the specific facts of this case and does not establish broad legal principles for agentic AI in general. The underlying lawsuit is also still ongoing in district court.
Q4: Why is this ruling considered significant for the AI industry?
It’s reportedly the first ruling by a federal appeals court addressing whether AI agents acting on a user’s behalf can legally access online platforms, a question relevant to companies like OpenAI, Google, Apple, and PayPal that are all building similar agent systems.
