Federal Judge Rules AI Training on Copyrighted Books Fair Use with Trial Pending

A U.S. federal judge ruled that using copyrighted books to train large language models (LLMs) like Anthropic's Claude constitutes fair use, a decision that could set a significant precedent for the AI industry. Judge William Alsup found that while Anthropic initially pirated millions of books from shadow libraries, the transformative nature of training AI on these works—similar to how humans learn by reading—justifies their use under fair use protections. The ruling emphasized that the AI's outputs do not reproduce specific creative elements or authors' styles, but rather synthesize a new, original expression. However, the judge rejected blanket fair use protection for the practice of downloading pirated books, leaving some legal questions unresolved. This decision marks a pivotal moment in ongoing legal battles over AI training data, with artists, authors, and media companies continuing to challenge unauthorized use of their work. AI companies argue that training on copyrighted content is necessary for innovation and creativity, while critics fear the ruling enables exploitation of creative labor without compensation.