Anthropic's $1.5 Billion Author Settlement Wins Final Approval, the Largest in US Copyright History

The largest known copyright settlement in US history is now final. On Monday, July 20, US District Judge Araceli Martinez-Olguin of the Northern District of California granted final approval to Anthropic's $1.5 billion settlement with book authors in Bartz v. Anthropic, the case over millions of pirated books used to train the Claude models.

From fair use to a record payout

The case produced one of the defining rulings of the AI copyright era. In June 2025, Judge William Alsup found that training on the books themselves qualified as fair use — but that downloading and storing more than 7 million pirated titles from shadow libraries such as LibGen was infringement. That split ruling pushed Anthropic to settle rather than face a damages trial. Alsup granted preliminary approval in September 2025 and has since retired; the case passed to Martinez-Olguin for the final sign-off.

What authors get

The fund covers more than 480,000 works, translating to roughly $3,000-3,100 per book. Uptake has been unusually strong for a class action: 440,490 claims had been filed by April, and more than 91% of covered authors and publishers claimed their share. Class counsel at Susman Godfrey and Lieff Cabraser were awarded $101 million in fees — 6.8% of the fund under the lodestar method, well below the $187.5 million they had sought.

Objections overruled

Some rightsholders argued the settlement was too small, that lawyers were overpaid, or that certain works were wrongly excluded. The judge rejected all of it, writing that the objections were "not grounded in a realistic assessment of the overall risks and rewards of a trial." No hard payment date was set, but class counsel said distributions will begin "as promptly as possible." Anthropic must also destroy its copies of the pirated books.

The benchmark is set

The final approval lands the same day Sony Music opened a new $4.5 billion front against AI music generator Udio — a sign that the Anthropic number is already functioning as the industry's reference price for training-data liability.