United States of America: District Court for the Northern District of California issued order finding that Artificial Intelligence training with copyrighted works qualifies as fair use in certain circumstances (Bartz et al. v. Anthropic) (Case 3:24-cv-05417-WHA)

On 23 June 2025, the United States District Court for the Northern District of California issued an order in Bartz et al. v. Anthropic, finding that the use of copyrighted works to train generative Artificial Intelligence (AI) models qualifies as transformative and therefore constitutes “fair use” under U.S. copyright law in certain circumstances. The ruling applies to developers of large language models (LLMs), particularly those using copyrighted content, including books to enhance model performance. The Court distinguished between three categories of books used to train Claude, Anthropic's AI. First, it identified digitised books used to train Claude, which it deemed transformative and non-infringing. Second, it identified legally purchased books that were scanned and used solely within the company, which also qualified as fair use. Third, it identified pirated books used to build a general-purpose research library, which the court ruled constituted copyright infringement. While the Court granted summary judgement on the fair use question for the first two categories, it ordered further discovery regarding the pirated copies for potential trial proceedings.