About 120,000 authors and other copyright holders are seeking compensation through a proposed US$1.5 billion class-action settlement with Anthropic, the AI company behind the Claude language model. The deal addresses allegations that Anthropic used copyrighted books without permission to train its artificial intelligence systems, according to court documents filed in the United States.
The settlement covers more than 480,000 works, with claims filed by rights holders for 91% of them, as revealed in judicial filings released Thursday [1]. If approved, the agreement would be the largest copyright settlement in U.S. history, and a final approval hearing is scheduled for next month.
Anthropic is the first and, so far, only major AI firm to reach a class-action settlement with authors in the U.S. over the use of copyrighted material for AI training. “This claims rate is another reason why this settlement is so historic and demonstrates overwhelming support from the class,” Justin Nelson of Susman Godfrey, one of the lead authors’ lawyers, told Reuters on Friday. The average claims rate in U.S. consumer class actions is just 9%, according to a 2019 report by the U.S. Federal Trade Commission.
However, the agreement has drawn objections from some authors, who argue the amount is insufficient, that lawyers are overcompensated, or that foreign rights holders have been improperly excluded. Anthropic has declined to comment on the matter.
The lawsuit, filed in 2024, alleged that Anthropic—backed by Amazon and Alphabet—used pirated copies of books to train Claude, its AI model, without permission or payment. The case is among dozens of similar suits brought by authors and news agencies against tech companies over the use of copyrighted content to train large language models. It is the first major U.S. case to reach a settlement.
In June 2025, U.S. District Judge William Alsup ruled that Anthropic’s use of the books for AI training was “fair use,” but found the company violated authors’ rights by storing more than 7 million pirated books in a central repository, which was not strictly necessary for training. A trial was set for December to determine damages, which could have reached hundreds of billions of dollars.
Instead, Anthropic agreed to the $1.5 billion settlement last year. U.S. District Judge Araceli Martinez-Olguin is expected to hold a hearing on May 14 to consider final approval of the deal.
As part of the settlement, the law firms Susman Godfrey and Lieff Cabraser have requested $187.5 million in legal fees—12.5% of the total fund—down from an initial request of $300 million after Judge Alsup questioned the amount.
The outcome of the hearing could set a precedent for how AI companies and copyright holders resolve disputes over the use of creative works in training artificial intelligence.
Source: UOL
