Judge criticizes $1.5B Anthropic deal with authors

Photo - Judge criticizes $1.5B Anthropic deal with authors
Anthropic offered authors $1.5B in compensation for using 465,000 books to train its AI. The judge questioned the deal’s legality and threatened to send the case to trial.
US federal judge William Alsup sharply criticized a $1.5B settlement between AI company Anthropic and a group of writers who accused it of illegally using nearly 465,000 books to train its Claude chatbot.

The deal, struck last week, offered authors and publishers $3,000 per book to avoid a December trial. But Alsup called the agreement “full of pitfalls” and doubted he could approve it. “We’ll see if I can close my eyes and sign off on this,” the judge said, scheduling the next hearing for September, 25.
Screenshot of court document. Source: bloomberglaw.com

Screenshot of court document. Source: bloomberglaw.com

In June, Alsup had already issued a mixed ruling: he did not deem training AI on books inherently illegal but noted that Anthropic had obtained millions of works from pirate sites.

The judge ordered Anthropic to submit a final list of all books used by September, 15, to prevent new lawsuits “out of thin air.” He also voiced concern that the claims process might lack transparency and deprive authors of fair compensation.

The hearing was attended by leaders of the Authors Guild and the Association of American Publishers, as well as three plaintiff authors, though none spoke. Plaintiffs’ attorney Justin Nelson assured the court that the money would be distributed fairly, stressing the case’s high public importance.

Still, Alsup hinted he was prepared to allow a full trial if his concerns were not addressed. “I have a troubling sense of all the shadow players,” the judge remarked.

avatar
Author
Writes about DeFi and cryptocurrencies from a technological perspective.