Theranos whistleblower John Carreyrou sues OpenAI, Google, and Meta in high-stakes AI copyright case
These big tech companies are about to find out what happens when creators demand full statutory damages instead of settling.
John Carreyrou, the Pulitzer winner who exposed Theranos, filed a copyright lawsuit on December 22 against OpenAI, Google, Meta, Anthropic, Perplexity, and xAI for allegedly training their models on pirated books. The catch, though, is that he and five co-plaintiffs deliberately avoided class-action status, positioning themselves to claim up to $150,000 per infringed work instead of accepting the pennies-on-the-dollar settlements that have become standard in AI copyright disputes.
The math is brutal for defendants. When Anthropic settled a similar class action in August for $1.5 billion, authors received roughly $3,000 per work, just 2% of the statutory maximum. If Carreyrou’s strategy succeeds and other authors follow suit, AI companies could face billions in individual judgments across thousands of copyrighted works.
Carreyrou made his position clear during November court hearings, calling pirated training data Anthropic’s “original sin” and rejecting the August settlement entirely.
Emmanuel Umahi
The complaint explicitly argues that “LLM companies should not be able to so easily extinguish thousands upon thousands of high-value claims at bargain-basement rates.” It’s a direct challenge to the class-action playbook that has allowed AI companies to resolve mass infringement claims cheaply.
The five other plaintiffs—Lisa Barretta, Philip Shishkin, Jane Adams, Mathew Sacks, and Michael Kochin—join Carreyrou in betting that individual lawsuits will force better compensation than collective settlements. Their legal team includes Kyle Roche of Freedman Normand Friedland, whom Carreyrou profiled in a 2023 New York Times investigation.
This marks the first copyright lawsuit against xAI, Elon Musk’s AI startup. xAI’s response was succinct: “Legacy Media Lies.” Other defendants didn’t immediately comment.
Why It Matters for AI Economics
The timing couldn’t be more pointed. OpenAI is reportedly negotiating funding that could value it at $830 billion, while Anthropic explores valuations above $300 billion. Courts will now assess whether settlements negotiated when these companies were worth a fraction of their current value adequately compensate the creators whose work built multi-hundred-billion-dollar businesses.
For AI developers and users, this case could fundamentally reshape training economics. If courts side with authors demanding maximum statutory damages, companies face two expensive paths: negotiate comprehensive licensing deals upfront or risk nine-figure judgments per lawsuit.
Either path increases costs significantly. That could trickle down to users through higher subscription prices, reduced free-tier access, or fundamental changes to how LLMs are trained. The “free access” model that currently drives 650 million weekly users across major AI platforms becomes harder to sustain when training costs multiply fifty-fold.
During a November hearing in the Anthropic class action, US District Judge William Alsup criticized a law firm co-founded by Roche for gathering authors to opt out of settlements in search of “a sweeter deal.” Whether that skepticism extends to Carreyrou’s individual lawsuit strategy remains to be seen.
Emmanuel Umahi
