Anthropic Pays $1.5B to Settle Copyright Case Over Pirated Books Used to Train Claude

Quick Facts

• U.S. District Judge Araceli Martinez-Olguin approved the $1.5 billion settlement on July 20, 2026, calling it the largest known copyright recovery in U.S. history.
• Authors and publishers will receive $3,000 per work for roughly 500,000 titles Anthropic copied from pirate libraries, with 92% of eligible copyright holders filing claims.
• The settlement closes a case where Anthropic faced theoretical statutory damages exceeding $70 billion had the matter gone to trial.

A federal judge in San Francisco approved a $1.5 billion settlement between Anthropic PBC and a class of authors and publishers on Monday, ending a two-year copyright battle over the training data used to build Claude. The ruling marks the largest known copyright settlement in American legal history.

U.S. District Judge Araceli Martinez-Olguin granted final approval, rejecting objections that the payout was too small. She said those complaints were “not grounded in a realistic assessment of the overall risks and rewards of a trial.”

How the Case Unfolded

Authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson sued Anthropic in 2024, alleging the company downloaded hundreds of thousands of books from shadow libraries including Library Genesis and Pirate Library Mirror to train its Claude models. Anthropic had also purchased physical books, scanned them, and stored the files in a central digital library it aimed to build from “all the books in the world.”

In June 2025, Judge William Alsup issued the first federal ruling to apply the fair use doctrine to AI training on copyrighted books. He found that using legally acquired books to train an AI was “spectacularly” transformative and therefore fair use. He drew the line at piracy. Copying books from shadow libraries was not protected, he ruled.

With that ruling in place, the case narrowed to 482,460 pirated titles registered with the U.S. Copyright Office. Statutory damages of up to $150,000 per work put Anthropic’s theoretical liability above $70 billion. The parties settled in August 2025 before a December trial date. A preliminary approval came in September 2025, and the final approval arrived July 20, 2026.

Who Gets Paid and When

Copyright holders will receive $3,000 per work. Anthropic is funding the settlement in four installments, with $300 million due by October 2, 2025, and the remaining amounts spread through September 2027. Attorneys were awarded just over $101 million of the $187.5 million they requested in fees.

More than 92% of eligible copyright holders filed claims, a claim-in rate that Anthropic Deputy General Counsel Aparna Sridhar called notable. “We are pleased that more than 91% of authors and publishers covered by the settlement have claimed their share of the payment,” Sridhar said in a statement. She added that the underlying fair use ruling on legally acquired books “remains the law today.”

Authors’ lead attorney Justin Nelson called the result “the largest known copyright recovery in history” and said distributions to the class would be made “as promptly as possible.”

What This Means for AI Companies

The Anthropic case is one of dozens of copyright suits filed against AI developers by authors, news outlets, and other content owners. It is the first major U.S. case to reach a final settlement.

Keith Kupferschmid, president and CEO of the Copyright Alliance, said the outcome confirms AI companies can compensate rights holders without halting development. “It also proves what we have been saying all along,” he said, “that AI companies can afford to compensate copyright owners for their works without it undermining their ability to continue to innovate and compete.”

Kashyap Kompella, CEO of RPA2AI Research, said courts in 2026 are shifting focus from whether AI training constitutes fair use to how training data was gathered, specifically whether data was pirated or collected in violation of contractual agreements. That distinction is now central to every remaining case in the pipeline.

For software and AI company executives, the Anthropic settlement sets a concrete price for data sourcing decisions made years ago. The $3,000-per-work figure and the 92% claim rate give opposing counsel in future cases a benchmark for settlement negotiations. The fair use ruling on legally acquired material stands, but the piracy line is now firmly drawn.

Read more: Anthropic settles with authors and publishers for $1.5B in landmark copyright case