Particle.news

Download on the App Store

Judge Rules Anthropic’s AI Training on Books Is Fair Use, Pirating Violates Copyright

By distinguishing transformative AI training from the unauthorized retention of millions of pirated books, the ruling sets a December trial on damages

Anthropic wins AI copyright ruling, judge says training on purchased books is fair use
Image
A view shows a Microsoft logo at Microsoft offices in Issy-les-Moulineaux near Paris, France, March 21, 2025. REUTERS/Gonzalo Fuentes/File Photo

Overview

  • Judge William Alsup found that using copyrighted books to train Anthropic’s Claude model is “exceedingly transformative” and qualifies as fair use under U.S. law.
  • The court determined that storing over seven million pirated books in a central library exceeds fair use and constitutes willful copyright infringement.
  • Alsup denied summary judgment on piracy claims and ordered a jury trial in December 2025 to assess damages, which could reach $150,000 per work.
  • This decision marks the first detailed federal ruling on fair use in generative AI training and is expected to shape dozens of similar lawsuits against major AI developers.
  • Anthropic welcomed the fair use finding but said it disagrees with the piracy ruling and is evaluating its legal options ahead of the upcoming trial.