Anthropic to pay $1.5 billion in historic AI copyright settlement

A U.S. federal judge in San Francisco approved a $1.5 billion settlement of a class‑action lawsuit against AI firm Anthropic. The case alleged that Anthropic used pirated copies of millions of books to train its Claude chatbot without permission. The court found that while training on copyrighted text may be fair use, the acquisition of pirated copies was unlawful. Roughly 482,000 works were covered, with about 91 % already claimed, and claimants will receive roughly $3,000 per work. Payments will be split between authors and publishers; Bloomsbury, the London publisher of the Harry Potter series, has 14,087 titles in the settlement and expects a multimillion‑pound payout. Lawyers for the class will receive about $101 million in fees. The settlement is described as the largest copyright payout in U.S. history and sets a precedent for how AI developers must obtain data for model training.