US music publisher sues Anthropic, argues against ‘fair use’ of AI

Music publishers Universal Music Group, Concord and ABKCO have asked a California judge to rule that U.S. copyright law does not exempt artificial intelligence startup Anthropic from liability for copying song lyrics to train its AI-powered chatbot Claude.

The publisher’s bill, filed Monday in federal court in San Jose, raises serious questions in the legal battle between creators and tech companies. The question is: Does the doctrine of “fair use” apply to the millions of copies of copyrighted materials used to train AI models?

The publisher argued in Monday’s filing that Claude’s AI-generated lyrics were derivative of the publisher’s lyrics and were not fair use because they “competed with and diluted the market.” The publisher also said Claude was illegally reproducing lyrics on demand without permission.

An Anthropic spokesperson did not immediately respond to a request for comment on Tuesday’s motion.

The plaintiffs said in a statement that Anthropic is committing “massive copyright infringement” and that the evidence is “overwhelming.” The case is one of dozens of disputes between copyright holders, such as authors and news organizations, and technology giants such as OpenAI, Microsoft, and Meta Platforms over the training of AI systems.

Anthropic, backed by Amazon and Google, was the first major AI company to settle one of the lawsuits, agreeing last year to pay $1.5 billion to a group of authors to settle a class-action suit. The music publisher filed a lawsuit in 2023 alleging that Anthropic infringed the copyrights to the lyrics of at least 500 songs by musicians including Beyoncé, the Rolling Stones and the Beach Boys.

All pending litigation is likely to revolve around whether AI systems are making fair use of copyrighted material by using it to create new and innovative content. The publishers on Monday asked U.S. District Judge Yumi Lee to rule before trial that Anthropic infringed their copyright and reject Anthropic’s fair use defense.

Anthropic denies the allegations, but has not yet claimed fair use in the music publisher’s lawsuit. In a separate lawsuit last year, U.S. District Judge William Alsup in San Francisco said Anthropic’s use of books for AI training was “truly transformative” and sided with the company on the issue.

Another judge in the Northern California court, Alsup and Lee, issued different rulings regarding fair use in AI training. The publisher said Monday that unlike the author in these cases, Claude’s record of reprinting works on demand is “overwhelming.”