Anthropic PBC said in a filing to the US Court of Appeals for the Ninth Circuit on 26 Aug 2025 that it has reached a settlement with a class of US authors who accused the artificial-intelligence company of infringing their copyrights while training its Claude language models. The parties signed a binding term sheet on 25 Aug and asked the appellate court to pause all proceedings while they finalise the agreement. Financial terms were not disclosed, and the settlement is expected to be completed by 3 September. “This historic settlement will benefit all class members,” said Justin Nelson of Susman Godfrey LLP, counsel for the authors. The case, Bartz v. Anthropic PBC, centres on allegations that the company downloaded two pirate databases containing as many as seven million books. Judge William Alsup had ruled in June that training on lawfully obtained books constituted fair use but allowed claims tied to the allegedly pirated copies to proceed and certified a nationwide class of authors. Anthropic faced a December 2025 trial in San Francisco and potential liability running into the billions of dollars before the agreement was reached. The deal removes one of the highest-profile US copyright challenges confronting generative-AI developers, though similar lawsuits against other companies remain active.