A federal judge delivered a major victory for Anthropic in a copyright lawsuit about artificial intelligence training methods. Judge William Alsup ruled that companies can legally use purchased books to teach AI systems without violating copyright laws. The decision came on June 24, 2025, from the Northern District of California court. Alsup said converting text into AI knowledge differs from copying or sharing copyrighted material. The ruling establishes that learning from legally obtained content falls under fair use protections.
The judge drew a sharp line between legal and illegal data collection practices. Anthropic faces continued legal trouble for using stolen books from piracy websites Book3 and LibGen. Alsup rejected arguments that downloading pirated content could be justified for AI development purposes. The court will hold a separate trial to determine damages for the illegal downloading. Authors retain the right to pursue additional legal action against the company.
The split decision creates important guidelines for artificial intelligence companies moving forward. Tech firms can train their systems on content they purchase or access through legal channels. Companies cannot defend using pirated materials even when the final product transforms the original work. The ruling affects how AI developers collect training data for future projects. Legal experts expect other courts to follow similar reasoning in upcoming technology cases.
The judge drew a sharp line between legal and illegal data collection practices. Anthropic faces continued legal trouble for using stolen books from piracy websites Book3 and LibGen. Alsup rejected arguments that downloading pirated content could be justified for AI development purposes. The court will hold a separate trial to determine damages for the illegal downloading. Authors retain the right to pursue additional legal action against the company.
The split decision creates important guidelines for artificial intelligence companies moving forward. Tech firms can train their systems on content they purchase or access through legal channels. Companies cannot defend using pirated materials even when the final product transforms the original work. The ruling affects how AI developers collect training data for future projects. Legal experts expect other courts to follow similar reasoning in upcoming technology cases.
