Anthropic to Pay $1.5 Billion in Largest Copyright Settlement in US History

A US federal judge has given final approval to Anthropic's $1.5 billion (€1.3bn) settlement with a group of authors, closing the largest copyright class action in US history and marking the first major resolution of a wave of lawsuits against AI companies over the use of copyrighted books in model training.

The Ruling

US District Judge Araceli Martínez-Olguín approved the deal on July 20 in San Francisco, rejecting objections from some authors who argued the payout was too small. The judge called those complaints "not grounded in a realistic assessment of the overall risks and rewards of a trial."

The settlement, first proposed last September, resolves a class action filed by authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson in August 2024. They accused Anthropic of using pirated books to train its Claude chatbot.

What Authors Get

Under the deal, authors and publishers receive $3,000 (€2,630) per work for an estimated 500,000 titles covered by the settlement. More than 91% of eligible claimants have already filed, according to Anthropic.

The Fair Use Twist

The case followed a June 2025 ruling by then-presiding judge William Alsup, who delivered a split decision. Alsup found that training Claude on lawfully acquired books qualified as fair use, since the process is transformative rather than a straight reproduction of the original work.

Where Anthropic lost was on the means of acquisition. Alsup ruled the company had downloaded and stored more than 7 million pirated books in what internal documents called a "central library" — a repository that existed independent of any specific training run. That exposed Anthropic to statutory damages of up to $150,000 per work, a potential bill running into the hundreds of billions of dollars had the case gone to trial in December.

Awkward Timing

The ruling lands at an uncomfortable moment for Anthropic, which spent much of this year positioning itself as the industry's leading voice against unauthorized model distillation. In February, the company accused three Chinese labs — DeepSeek, Moonshot AI, and MiniMax — of running coordinated campaigns using roughly 24,000 fake accounts and 16 million queries to extract Claude's outputs to train competing models.

Critics including Elon Musk pointed out the obvious irony: a company built in part on unlicensed books objecting to a rival extracting value from its own model without payment.

Broader Implications

Anthropic's deputy general counsel Aparna Sridhar framed the outcome as closure, noting the fair use ruling remains intact and the settlement was about resolving the "central library" issue. The plaintiffs' lead attorney Justin Nelson described it as the largest publicly known copyright recovery in history.

The case is one of dozens still working through US courts against OpenAI, Google, Meta, and others over AI training practices. The Alsup fair use ruling — that training on copyrighted material is transformative — gives other AI companies a strong legal precedent. But the $1.5 billion price tag for illegally sourcing that material sends an unmistakable signal: how you acquire the data matters.

Key takeaway: Training AI on copyrighted material may be fair use, but pirating millions of books to build your training dataset is not. The $1.5 billion settlement establishes that AI companies can be held accountable for how they source their training data — even when the training itself is legal.