A federal judge has delivered a pivotal decision favoring AI company Anthropic in a copyright lawsuit, ruling that training artificial intelligence systems on published books without author consent constitutes fair use.
This judgment, handed down by Judge William Alsup in the U.S. District Court for the Northern District of California, is the first major ruling of its kind and sets a significant legal precedent. Although not binding for other courts, this decision has substantial implications for the ongoing legal battles over how copyrighted material is used to train generative AI systems.
Creative Industry Pushes Back as Authors Sue Over AI’s Use of Copyrighted Works
The legal question at the center of this case has been a contentious one for years, as creative industries voice concerns over AI tools trained on copyrighted content. Since 2023, authors, artists, music labels, and media companies have filed numerous lawsuits against AI firms, protesting the use of their work without consent.
Many artists have signed open letters and urged for regulation, while some content owners have begun negotiating licensing deals with AI developers to protect their work and establish usage terms.
Judge Rules AI Book Training Is Fair Use in Landmark Win for AnthropicThe lawsuit was filed by authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson, who alleged that Anthropic had infringed copyright by digitizing and using millions of books — some purchased, others pirated — to train its language models, including versions of its Claude AI system.
While Judge Alsup acknowledged that the authors’ works had expressive elements deserving of strong copyright protection, he ruled that the use of these works for AI training was fair due to the highly transformative nature of the technology involved.
Fair Use Upheld, But Ethical and Legal Content Sourcing Still Required for AI Training
Alsup’s decision emphasized the transformative purpose of using copyrighted books in AI training, stating that the models created something new rather than replicating or replacing the original content. He cited the four-factor test from the Copyright Act, noting that all but one factor favored fair use.
While downloading pirated books is not protected under fair use, the judge found that copying entire works for training purposes was reasonable, as the AI models did not offer public access to those texts or threaten the market for the original books.
Although the ruling supports the legality of training AI on copyrighted material under fair use, Alsup clarified that AI developers must still obtain content through lawful means. Anthropic remains liable for the use of pirated copies and will face further trial on that issue.
The judge stated that purchasing a book after previously downloading it illegally does not erase the original act of infringement, though it may influence the damages awarded. This nuanced decision both advances the legal standing of AI development and underscores the importance of ethical content sourcing.
