Anthropic AI gets first legal 'win' in copyright suit vs authors

Anthropic AI has received a positive ruling in a United States court, in a copyright suit brought by authors.

A federal judge has ruled that Anthropic's use of lawfully acquired books to train its AI models constitutes fair use. The court concluded that the company’s approach was sufficiently transformative and did not violate the authors’ copyright.

Three authors alleged their works were used without consent to train the Claude family of large language models developed by Anthropic.

Judge William Alsup sided with Anthropic in regards to the training, on purchased books, but said a separate trial must be held to assess the legality of the firm’s broader data practices.

The ruling noted Anthropic had created a digital library by scanning print books and scraping online repositories, including some known piracy sites.

The court drew a legal distinction between transformative use in training and the method of obtaining content. It marks the first legal endorsement of fair use for AI training. Other claims against the company remain unresolved.