Judge seeks details on Anthropic’s $1.5 billion author settlement tied to Claude training

A U.S. federal judge on May 14, 2026, declined to grant final approval to Anthropic’s proposed $1.5 billion settlement with authors who accused the AI company of misusing books to train its Claude chatbot. U.S. District Judge Araceli Martinez-Olguin asked lawyers in San Francisco for more information on issues including attorneys’ fees and payments to lead plaintiffs. The agreement, first preliminarily approved in September 2025, is the largest known U.S. copyright settlement and the first major AI training case in the country to settle. Authors and copyright holders filed claims covering more than 92% of the 480,000-plus works in the settlement. The lawsuit, filed in 2024, alleged Anthropic, backed by Amazon and Alphabet, used pirated books without permission. A judge ruled in June 2025 that training Claude was fair use, but found Anthropic violated rights by storing more than 7 million pirated books in a “central library.” More than 25 opt-out authors filed a new complaint on May 13, 2026.