Anthropic has reached a settlement with a coalition of American writers in a collective lawsuit stemming from the use of books, including pirated ones, to train its large language models. Initially, the court had determined that utilizing legally purchased books for model training constituted fair use. However, the presence of pirated materials still exposed the company to the risk of significant financial penalties. Anthropic had previously viewed the court's ruling as a favorable indication for the progression of generative AI models. The exact details of the settlement remain confidential, and Anthropic has chosen not to comment further.