A federal judge in San Francisco has signed off on Anthropic’s $1.5 billion settlement with a group of authors who sued the AI company over the use of their works to train its Claude models. The deal, the largest ever in an AI copyright case, was approved by U.S. District Judge Araceli Martinez-Olguin, who took over the case after the original judge, William Alsup, retired. The settlement covers more than 480,000 works and requires Anthropic to pay roughly $3,000 per affected piece, while also destroying pirated copies of the books it had stored.

The lawsuit, filed in 2024, accused Anthropic of using illegally obtained copies of the authors’ work to train Claude. Judge Alsup had earlier ruled that using the works for training fell under fair use, but he found the company violated copyright when it kept 7 million pirated books in a central library. He rejected an earlier version of the settlement last September, calling it incomplete and saying it was being forced on authors. Since then, Anthropic launched a website listing all affected works, allowing authors to check if they qualify for a claim.

Not everyone is satisfied with the payout. Some plaintiffs opted out of the settlement to file separate lawsuits against Anthropic, arguing the per-work compensation is too low. Despite the approval, the case highlights ongoing tensions between the AI industry and content creators over how training data is sourced and compensated. The deal’s final approval marks a major milestone, but the legal fight over AI’s use of copyrighted material is far from over.

Source: Engadget