Particle.news

Download on the App Store

Court Declares AI Training From Purchased Books Fair Use, Sets December Piracy Damages Trial

Developers now face a clear legal mandate for rigorous data sourcing following the ruling that pirate texts incur infringement liability.

Image
Image
Image

Overview

  • U.S. District Judge William Alsup found that destructively scanning legally acquired books for AI model training is “spectacularly transformative” and qualifies as fair use under copyright law.
  • The ruling applies the first-sale doctrine to permit AI firms to use text from purchased books without obtaining separate licenses or paying authors for training purposes.
  • Alsup rejected the use of pirated content from shadow libraries as unfair and infringing, stating that unauthorized copies offer no legal fair-use defense.
  • A December trial has been scheduled to determine the financial damages Anthropic and similar developers must pay for infringement related to unlicensed pirated texts.
  • The verdict elevates “data diligence” into a critical compliance requirement for AI companies seeking to avoid future litigation and reputational harm.