Reagan-Era Hacker Law Meets AI Scrapers: Amazon v. Perplexity

Picture this: it's 1986. Reagan's in the White House, Ferris Bueller is about to hit theaters, and Congress passes the Computer Fraud and Abuse Act to stop shadowy hackers from breaking into government mainframes. Fast-forward nearly four decades, and that same law is being weaponized in a turf war between Amazon — the $1.8 trillion everything store — and Perplexity AI, the buzzy answer-engine startup that's been vacuuming up the internet since its 2022 launch.

The U.S. Appeals Court is now weighing whether the CFAA — a statute written before the World Wide Web even existed — can meaningfully police how AI companies scrape, crawl, and ingest data from websites that don't want to be scraped. Spoiler: it can't. Not really. And the implications stretch far beyond these two combatants.

Here's the setup. Amazon, which operates one of the most aggressively defended data fortresses on the planet — AWS alone generated over $100 billion in revenue last year — has been going after Perplexity for allegedly accessing Amazon's servers and data in ways that violate the CFAA's "exceeding authorized access" clause. Perplexity, valued at roughly $9 billion after its December 2024 funding round, has built its entire business on doing what every AI company does: crawling the web, ingesting content, and using it to power its AI-powered answer engine. The difference? Perplexity does it more transparently and more aggressively than most, and it's been catching heat from publishers, Reddit, Forbes, and now a tech titan that really, really doesn't like anyone touching its stuff without permission.

The core legal question is almost philosophical: what does "unauthorized access" even mean in 2025? The CFAA was designed for a world where hackers used stolen passwords to break into DOS-era systems. Nobody in 1986 imagined that a search startup backed by Jeff Bezos himself (yes, Bezos Expeditions invested in Perplexity's 2023 round — the irony is suffocating) would build a business model around sending bots to read publicly available web pages. And yet here we are, watching a Reagan-era statute get stretched over AI scraping like a cheap latex glove.

The appeals court knows this is absurd. During oral arguments, judges reportedly struggled with the same question every tech lawyer has been asking since OpenAI started hoovering up the internet for GPT training data: if a website is publicly accessible, but the owner says "don't scrape me," does ignoring that request constitute a federal crime? The CFAA says maybe. Common sense says absolutely not.

This matters enormously because the CFAA carries both civil and criminal penalties. If the court sides with Amazon's expansive reading, it would effectively give every website owner a veto over whether AI companies — or any automated tool — can access their pages. That's not just a problem for Perplexity. It's a problem for Google, which crawls billions of pages daily. For OpenAI's GPTBot, for Anthropic's Claude crawlers, for every SEO tool, price comparison site, and academic research project that's ever run a script against a public URL. The internet as we know it — the open, indexable, machine-readable internet — would fundamentally break.

Amazon's position is, predictably, corporate hypocrisy at its finest. Amazon Web Services scrapes, indexes, and analyzes more data than perhaps any entity on Earth. Amazon's own Alexa team spent years crawling the web to train voice models. Amazon's product recommendation engine is built on behavioral data harvesting that would make a surveillance state blush. But when a nimble AI startup — one that's eating into Amazon's search and shopping referral traffic — does the same thing to Amazon's servers? Suddenly it's a federal case.

The subtext here is competition, not law. Perplexity has been positioning itself as the anti-Google: a conversational answer engine that gives you information directly instead of sending you through a maze of sponsored links. Amazon, which has been quietly building its own AI capabilities — from the $4 billion Anthropic investment to the underwhelming Rufus shopping assistant launched February 2024 — sees Perplexity as a threat to the commerce-search pipeline that feeds its ad business, which pulled in $56.2 billion in 2024. When a startup threatens your ad revenue, you don't just compete. You litigate.

Perplexity isn't blameless either. The company has faced repeated accusations of scraping content without attribution, ignoring robots.txt directives, and republishing summaries that sometimes look suspiciously like lightly edited copies of original reporting. Forbes called them out in June 2024 for plagiarizing its content. Wired documented Perplexity's crawler ignoring exclusion protocols. The startup has tried to clean up its act — launching a revenue-sharing Publishers Program, adding citations, and claiming its new Perplexity Pages feature properly attributes sources — but the reputation for being a "scrape first, ask forgiveness later" operation has stuck.

But here's the thing: bad behavior isn't the same as federal crime. Perplexity might be annoying. It might be aggressive. It might be playing fast and loose with content licensing norms that the entire AI industry has decided to pretend don't exist. But the CFAA — a law that was supposed to stop people from hacking into NORAD — is not the right tool for settling a business dispute between two multi-billion-dollar companies arguing about web scraping.

The appeals court's decision, expected in the coming months, will set a precedent that ripples across the entire AI landscape. If they narrowly construe the CFAA — following the Supreme Court's 2021 Van Buren decision, which limited the law's reach — it's a win for the open internet and for AI companies that depend on public data access. If they side with Amazon's broad interpretation, we're looking at a future where every website becomes a walled garden, every scraper is a potential felon, and the only companies that can afford to train AI models are the ones big enough to buy their way into every data licensing deal on the planet.

That future looks like exactly five companies controlling all AI development. Which, let's be honest, is probably the future Amazon's lawyers are aiming for.

The real scandal isn't that Perplexity scraped Amazon's data. It's that in 2025, we're still using a law written for dial-up modems to decide the future of artificial intelligence. Congress could fix this. They won't. The EU's AI Act, which took effect August 2024, at least attempts to create modern rules for modern technology. The U.S.? We're out here prosecuting AI companies under a statute that predates Windows 1.0.

This case isn't about justice. It's about power. Amazon has it. Perplexity wants it. And the courts are stuck playing referee with a rulebook that doesn't even mention the word "internet." Place your bets accordingly.