A US judge ruled that Anthropic's use of copyrighted books to train its Claude AI model is a transformative use, not violating copyright law. This decision, part of a lawsuit filed by three authors including Andrea Bartz, allows Anthropic's fair use defense. However, the judge rejected Anthropic's request to dismiss the case entirely due to the firm's use of pirated copies. Anthropic, backed by Amazon and Alphabet, could still face significant damages. This ruling is one of the first to address the widespread legal questions surrounding AI training data.

Prepared by Jonathan Pierce and reviewed by editorial team.