San Francisco: A U.S. federal judge has granted final approval to Anthropic's $1.5 billion settlement with a class of authors who accused the artificial intelligence company of using pirated books to train its Claude chatbot, marking one of the largest copyright settlements in U.S. legal history.
The settlement stems from a ruling last year by U.S. District Judge William Alsup, who held that Anthropic's use of copyrighted books to train its AI model constituted fair use under U.S. copyright law. However, the court also found that the company infringed copyright by storing more than seven million pirated books in a central digital library, regardless of whether all of them were used for training. The case was scheduled to proceed to trial on damages before the parties reached a settlement.
According to court filings, claims were submitted for more than 92% of the approximately 480,000 works covered under the settlement. Eligible authors are expected to receive an average payout of around $3,000 per work, although the final amount will vary depending on approved claims.
Under U.S. copyright law, statutory damages in such cases can reach up to $150,000 per infringed work, had the matter proceeded to trial and liability been established.
Several authors objected to the settlement, arguing that the compensation was inadequate, that legal fees were excessive, or that some copyright holders had been excluded.
U.S. District Judge Araceli Martinez-Olguin rejected those objections, stating that criticisms of the settlement amount did not adequately account for the uncertainties and risks associated with litigation. However, the court reduced the attorneys' fee award, approving more than $101 million instead of the $187.5 million requested.
Anthropic's Response
Anthropic said the settlement followed the court's earlier ruling that training AI models on books constitutes fair use under copyright law. Deputy General Counsel Aparna Sridhar said the company reached the agreement after that decision and noted that the ruling remains in effect.
The case is among several lawsuits filed by authors, publishers and news organisations against AI companies over the use of copyrighted material for training large language models. While this settlement resolves the claims of participating authors, some copyright holders opted out and continue to pursue separate legal action against Anthropic.
