A high-profile copyright infringement case involving the training of artificial intelligence on pirated texts has concluded in the United States. The court approved a settlement between Anthropic and a group of writers, the largest in the history of such cases, amounting to $1.5 billion. The settlement concerns the use of approximately 482,000 works obtained from sites with pirated content.

The case began in 2024 when writers accused Anthropic of training its AI model Claude on books obtained from pirate sites. In 2025, Judge William Alsup ruled that training AI on works could fall under the fair use doctrine but recognized as a separate violation Anthropic's creation of a collection of 7 million pirated books.

The settlement avoided a jury trial where potential damages could have reached hundreds of billions of dollars. However, the ruling on the applicability of the fair use doctrine under certain circumstances remains in effect. Payments under the settlement will amount to nearly $3,000 per book, underscoring the seriousness of the authors' claims.

This case raises important questions about the boundaries of using copyrighted materials for AI training and the protection of intellectual property in the digital age. Companies working with AI must consider the risks associated with using content and actively monitor the development of legal practice in this area. It is important to understand that copyright infringement can not only lead to significant financial losses but also affect a company's reputation.

Legiscan