Imagine asking your AI assistant to log into Amazon, find the cheapest air fryer, and buy it for you. Now imagine a federal judge calling you a hacker for it.
You probably haven’t heard of the lawsuit between Amazon and Perplexity AI currently sitting in the U.S. Court of Appeals for the Ninth Circuit. You probably don’t care about the intricacies of the Computer Fraud and Abuse Act (CFAA). But you need to care about this case, because it is about to determine whether the next decade of the internet belongs to you and your AI, or to the corporate giants trying to lock you out.
The surface-level story is that Amazon sued Perplexity over its new Comet browser, claiming the AI tool unlawfully accessed Amazon’s website. The TL;DR floating around legal circles is that Perplexity’s agents ignored Amazon’s robots.txt file. But that framing is a massive distraction.
This case has almost nothing to do with rogue bots scraping data. Perplexity’s Comet browser includes an AI Assistant that logs into websites using the user’s own username and password. You authorize it. You give it your credentials. You tell it your intent. Yet, Amazon claims this violates federal anti-hacking laws.
If a user deliberately hands their account and intent to an AI, is the website’s refusal to serve the AI a restriction on the tool, or a restriction on the user?
Amazon wants it both ways. They want human users to browse and buy, but they want to block AI agents from doing the exact same thing on our behalf. The problem? AI agents are rapidly becoming indistinguishable from human users when they browse with our exact session and intent.
The more seamlessly AI assists, the more it erases the legal boundary between ‘I visited the site’ and ‘my AI visited the site.’
This is the terrifying reality of the Amazon vs. Perplexity case. The court is being asked to decide who owns the delegation of access. If Amazon wins, they establish a legal precedent where using an automated assistant to navigate the web with your own credentials is treated as a cyberattack. Ordinary AI-assisted browsing could become a CFAA violation on any site that silently prohibits automation in their Terms of Service.
Think about how absurd that is. You give an AI your login. The AI does exactly what you would have done, just faster. And for that, you are legally a trespasser. It’s the digital equivalent of a store welcoming you inside, but calling the police because you brought a calculator to add up your grocery bill.
The internet is quietly being rezoned. What was once a public square is becoming a private mall where your AI assistant is banned by default.
We are standing at the edge of a digital feudalism where platforms claim ownership not just of their servers, but of the very method of interaction. If a website can dictate that you are only authorized to access it if you use your biological hands and a traditional browser, then you don’t own your digital identity—they do.
The Ninth Circuit’s decision will either protect the right to delegate our digital tasks or effectively ban the next generation of consumer AI assistants from the open web. We cannot let corporate monopolies criminalize convenience. If an AI acts under my authority, with my credentials, and my intent, it is me. Anything less is a digital hijacking of our autonomy.
FAQ
Q: Isn't Perplexity just scraping data and stealing content?
A: No, that's the distraction. Perplexity's Comet browser logs in using the user's own username and password. It isn't bypassing paywalls or stealing data; it is acting as a delegated assistant executing the user's direct intent with their own credentials.
Q: How does this affect my daily internet use?
A: If Amazon wins, any website that silently prohibits 'automation' in their Terms of Service can legally treat your AI assistant as a hacker under the CFAA. Using AI to shop, research, or navigate the web for you could become a federal crime.
Q: Shouldn't websites have the right to control who accesses their servers?
A: They have the right to protect their servers from malicious attacks, but they shouldn't have the right to dictate how a user interacts with their own account. If I authorize an AI to act for me, blocking the AI is a restriction on me, not the tool.