Anthropic Faces Legal Scrutiny Over $1.5 Billion AI Copyright Settlement

A federal judge questioned details about Anthropic's proposed $1.5 billion settlement with authors accusing it of unauthorized book use for AI training. The settlement involves over 480,000 works, but its fairness is contested. Authors allege Anthropic's infringement, with alternative lawsuits emerging.

A federal judge has demanded further details about the proposed $1.5 billion settlement between AI company Anthropic and authors. The authors accused the tech firm of misusing their books for training its chatbot, Claude. During a hearing in San Francisco, U.S. District Judge Araceli Martinez-Olguin paused final approval, seeking clarity on issues like legal fees and compensation to lead plaintiffs.

The case represents a significant legal battle in the AI realm, one of many brought by copyright owners against tech giants. Originally approved by Judge William Alsup in September, it marks the largest documented copyright settlement in the U.S. to date. Authors claim the settlement inadequately compensates them, excessively rewards attorneys, or excludes certain copyright holders.

The litigation against Anthropic began in 2024 after authors alleged the company used pirated book copies to refine Claude's algorithm. While Alsup ruled the training was fair use, he penalized Anthropic for retaining over seven million pirated books. The trial concerning Anthropic's financial penalties is scheduled for December, with more authors pursuing separate legal action.