Two Consequential Rulings Find Fair Use Where Copyrighted Works Were Used to Train LLMs

The week of June 23, 2025 saw a flurry of high-impact decisions concerning copyright and artificial intelligence (AI). Two judges in the U.S. District Court for the Northern District of California issued substantive decisions regarding fair use and generative AI model training. Both courts found fair use and that using copyrighted works to train large language models (LLMs) is highly transformative.

On June 23, 2025, in Bartz v. Anthropic PBC, the court granted summary judgment for defendant Anthropic PBC, finding that using copyrighted works for training LLMs was a fair use, and that Anthropic’s creation of a digital library by purchasing and digitizing millions of print books was likewise a fair use. The court, however, found that Anthropic’s use of downloaded pirated copies of books to build its central library was not justified as fair use, and thus denied summary judgment for Anthropic on its argument that the pirated library copies must be treated as training copies.

Just two days later, on June 25, 2025, the court in Kadrey v. Meta Platforms, Inc. granted defendant Meta’s motion for summary judgment on its fair use defense against claims that using the plaintiffs’ books for LLM training is copyright infringement.

Given the number of pending cases involving issues of fair use and AI model training, these decisions may influence how other courts approach these issues.

Read more on Bartz v. Anthropic PBC here and Kadrey v. Meta Platforms, Inc. here.