Anthropic's $1.5 Billion AI Copyright Settlement Gets Green Light from Judge

A federal judge in San Francisco has signed off on Anthropic's $1.5 billion settlement with writers who accused the AI company of using stolen copies of their work to train its Claude chatbot. The deal, the largest of its kind in an AI copyright case, resolves claims that Anthropic built a central library of 7 million pirated books without permission. Authors whose works were included will receive roughly $3,000 per title, and the company must destroy all unauthorized copies.

The case originally landed before Judge William Alsup, who last September rejected an earlier version of the settlement, calling it incomplete and saying it had been shoved "down the throat of authors." Alsup retired before finalizing the terms, leaving the matter to Judge Araceli Martinez-Olguin. She approved the revised deal after Anthropic launched a public website listing more than 480,000 affected works, letting authors check eligibility and file claims.

Not everyone is satisfied. Some plaintiffs found the per-work payout too low and have opted out to pursue separate lawsuits against Anthropic. The settlement covers only the copyright infringement claims tied to the pirated book library. Judge Alsup had previously ruled that using the works to train Claude qualified as fair use, but storing the copies in a central repository crossed the line into infringement.

Source: Engadget