Anthropic settles AI book-training lawsuit with authors

Anthropic has reached a settlement in a notable class action lawsuit involving various authors of both fiction and nonfiction, as revealed in a recent filing submitted to the Ninth Circuit Court of Appeals. The company had previously secured a partial win in a lower court decision and was in the midst of appealing that ruling when the settlement was announced. While the specifics of the agreement remain undisclosed, Anthropic did not provide immediate comments when approached for further information. The lawsuit, known as Bartz v. Anthropic, centers around the company's utilization of literary works as training data for its advanced language models. In a prior court ruling, it was determined that Anthropic’s use of these texts could be classified as fair use. However, the fact that many of the books in question were allegedly pirated led to substantial financial repercussions for the company. Despite these challenges, Anthropic celebrated the earlier ruling, viewing it as a positive development for the generative AI landscape. The company expressed confidence in its intentions, stating, "We believe it’s clear that we acquired books for one purpose only — building large language models — and the court clearly held that use was fair," in a statement to NPR after the June ruling.