Anthropic’s record-breaking $1.5 billion settlement for an AI copyright lawsuit filed by writers has been approved by a federal judge in San Francisco. The company was sued by a group of authors in 2024, who accused Anthropic of using pirated copies of their work to train its Claude large language models. William Alsup, the case’s original judge, ruled that Anthropic’s use of the writers’ work to train Claude counted as fair use. At the same, he found that the company infringed their copyright when it stored 7 million pirated books to a central library 

Alsup rejected the deal in September last year over concerns that it was shoving the terms “down the throat of authors.” He also said it was “nowhere close to complete” and left a lot of questions that needed to be addressed. Since then, Anthropic has launched an official website for the settlement, which includes a list of affected works so that authors could know for sure if they can file for a claim. To note, the settlement covers more than 480,000 works. 

US District Judge Araceli Martinez-Olguin, who took over the case after Alsup retired, has now approved the settlement terms. Authors are expected to get approximately $3,000 per affected work, while Anthropic is required to destroy the pirated copies of their books. Some of the plaintiffs, however, found the payout too low and opted out of the settlement to file separate lawsuits against Anthropic.