Anthropic's $1.5B copyright settlement approved, 350 authors opt out

On Monday, US District Judge Araceli Martínez-Olguín approved a $1.5 billion settlement between Anthropic and a class of authors, concluding the largest copyright class-action ever certified. The settlement resolves claims over Anthropic's use of copyrighted books for AI training. Notably, the court had previously ruled that training AI on books was fair use, but that piracy of works was likely not. Despite objections from some authors who argued payouts were too low ($3,000 per work) and lawyers' fees too high, the judge found the settlement fair. Approximately 95% of the class received notifications, and over 91% of impacted authors and publishers have already filed claims. Only 350 class members opted out, while 54 others filed late objections or opt-out requests. The judge noted that $3,000 per work was four times the minimum statutory damages, and the high participation rate indicated widespread acceptance.

The judge also reduced lawyers' fees from the originally requested 20% ($300M) to less than 7% of the settlement fund (~$101M) citing concerns over inflated projections. Lead plaintiffs' service awards were cut from $50,000 to $15,000 each, deemed more reasonable. Anthropic's deputy general counsel expressed satisfaction that the case established fair use for AI training and that most rightsholders have claimed their share. Authors' lawyers celebrated the approval as a step toward accountability. While unlikely, the judge noted that any remaining funds could be redistributed among class members. Anthropic also agreed to comply with a discovery order regarding its training data.