Anthropic's $1.5B author payout gets green light: $3,000 per pirated book

A federal judge gave the final stamp of approval to Anthropic’s $1.5 billion class action settlement on Monday, closing a major chapter in the legal brawl over AI training data. Judge Araceli Martínez-Olguín said the deal provides “meaningful relief,” setting the payout at roughly $3,000 for each copyrighted book the company allegedly scraped without permission. For context, the plaintiff’s law firm is calling it the “largest known copyright recovery in history” — a number that puts a tangible, and expensive, price tag on the once-wild-west practice of hoovering up text to train large language models.

This isn’t just a big check. It’s a signal. The original suit, filed by authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson in 2024, accused Anthropic of a “Napster-style downloading of millions of works.” That framing stuck, and even after Anthropic scored a partial victory from the now-retired Judge William Alsup, the company still chose to settle rather than let a jury decide what fair use looks like when you’re building a Claude competitor. The September 2025 agreement now turns into a real payout.

Anthropic general counsel Aparna Sridhar told Reuters that over 91% of covered authors and publishers have already claimed their share, adding the company is “looking forward to bringing this matter to a close.” Getting nine out of ten eligible claimants to sign on suggests the per-book rate, while a rounding error for a company valued in the tens of billions, felt sufficient to most members of the class. It’s a clean break for a firm that wants to look like the responsible player in a room full of companies still duking it out in court.

Don’t mistake the resolution for universal peace, though. Anthropic is still staring down separate lawsuits from entities like Chicken Soup for the Soul and a contingent of authors who flatly argue that $3,000 per book is a lowball offer for the wholesale ingestion of their life’s work. The settlement closes the biggest liability, but it doesn’t settle the core ethical debate. It just proves that when the legal pressure gets high enough, even the most ambitious AI labs will write a massive check to make the problem go away.