Judge approves Anthropic's $1.5B copyright settlement
A federal judge granted final approval to the largest US copyright settlement on record, covering about 500,000 books Anthropic downloaded from pirate libraries.
A federal judge in San Francisco has granted final approval to Anthropic's $1.5 billion settlement with book authors, closing the largest copyright settlement in US history and setting a reference price for training data taken from pirate libraries.
The terms
US District Judge Araceli Martínez-Olguín signed off on the deal this week, overruling objections from authors who argued the amount was too low, that attorneys were overcompensated, or that some rights holders were wrongly excluded. The settlement covers roughly 500,000 works at an expected payout of about $3,000 per title. Participation is strikingly high: about 91% of eligible authors filed claims, and reporting puts opt-outs at only around 350 — leaving very few plaintiffs positioned to pursue separate suits.
How the case got here
The class action, brought by three authors in 2024 (Bartz v. Anthropic), produced one of the most consequential rulings in AI law: then-Judge William Alsup found that training on lawfully purchased books qualifies as fair use, while downloading and retaining millions of pirated copies from shadow libraries does not. That split ruling exposed Anthropic to statutory damages that could theoretically have reached hundreds of billions of dollars, pushing it to settle in 2025. Alsup has since retired, leaving Martínez-Olguín to complete the approval.
Why it matters
The approval converts a theoretical legal risk into a concrete market signal: acquisition method, not training itself, is where US copyright liability currently bites, and the going rate for a pirated book is about $3,000. Every AI lab facing similar claims — and every plaintiff's lawyer drafting them, including in the pending Sony-Udio music dispute — now has a benchmark. For authors, the near-universal claims rate suggests creators would rather take structured compensation than gamble on litigation, a dynamic that may accelerate licensing deals as the cheaper path for labs going forward.