Anthropic Wins Right to Use Copyrighted Books to Train AI Models

Anthropic PBC convinced a California federal judge that using copyrighted books to train its generative AI models qualifies as fair use. Authors’ claims that the maker of the ‘Claude’ large language model infringed their copyrights by copying pirated books, though, will proceed toward trial.

Senior District Judge William Alsup partially granted summary judgment to the Amazon-backed AI company in an order issued Monday in the US District Court for the Northern District of California.

The ruling represents a win for the AI industry, in which companies including OpenAI Inc. and Meta Platforms Inc. have argued that training with unlicensed copyrighted content is both legal necessary for AI to work. Dozens of similar lawsuits remain pending against AI’s biggest players.

The decision comes about a month after a hearing where Alsup said he was leaning toward finding Anthropic violated copyright law when it made initial copies of pirated books, but that its subsequent use of the material to train AI models is fair use.

The named-plaintiff authors, including Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson, sued Anthropic in 2024, alleging in a proposed class action the company illegally downloaded versions of their works and tried to hide the extent of the infringement.

Susman Godfrey LLP, Lieff Cabraser Heimann & Bernstein LLP, and Cowan Debaets Abrahams & Sheppard LLP represent the authors. Anthropic is represented by Arnold & Porter Kaye Scholer LLP, Latham & Watkins LLP, and Lex Lumina LLP.

The case is Bartz v. Anthropic PBC, N.D. Cal., No. 3:24-cv-05417, order on fair use issued 6/23/25.