Federal Judge OKs $1.5 Billion Settlement Over AI Training with Unlicensed Books
Anthropic will pay roughly $3,000 per book to thousands of authors after a court approved a $1.5 billion copyright settlement for using pirated copies to train its Claude chatbot.
A U.S. district judge in San Francisco has given final approval to a $1.5 billion settlement that requires AI firm Anthropic to compensate authors whose books were used without permission to train its Claude chatbot. The ruling states that about 91% of the more than 482,000 titles covered have been claimed by authors or publishers, who will receive roughly $3,000 per work.
Judge Araceli Martínez-Olguín described the agreement as providing “meaningful relief” to the affected creators. The case originated after bestselling novelist Andrea Bartz and two other writers sued Anthropic in 2024, alleging the company obtained millions of books from piracy sites.
Earlier, Judge William Alsup had issued a preliminary approval and ruled that while training AI on copyrighted books is not per se illegal, Anthropic’s acquisition of the texts was wrongful.
"The settlement is the largest known copyright recovery in history," plaintiff attorney Justin Nelson said. "We are pleased that more than 91% of authors and publishers covered by the settlement have claimed their share of the payment, and we’re looking forward to bringing this matter to a close," Anthropic deputy general counsel Aparna Sridhar said.
The settlement marks the first major resolution among numerous AI-related copyright lawsuits pending in U.S. courts.