Judge Criticizes $1.5 Billion Settlement Between Anthropic and Authors Over AI Training Data

A federal judge in San Francisco has raised substantial concerns regarding a proposed $1.5 billion settlement between artificial intelligence company Anthropic and authors who allege that nearly 465,000 of their books were used without authorization to train the company's chatbot, Claude.

U.S. District Judge William Alsup criticized the settlement's terms, questioning the validity and fairness of the agreement. He expressed particular apprehension about the claims process and the accuracy of the number of books involved, emphasizing the necessity to ensure that all affected authors are properly informed and treated equitably.

Alsup has scheduled a follow-up hearing for September 25 to review whether his concerns have been addressed. He also set deadlines for the submission of a verified list of pirated works by September 15 and the claims form by September 22. The judge warned that the case might still proceed to trial, citing unease about the involvement of groups such as the Authors Guild and the Association of American Publishers.

This development follows a June ruling in which Alsup found that while training AI chatbots on copyrighted books wasn't illegal, Anthropic had wrongfully acquired millions of books through pirate websites to enhance its AI models.