By Blake Brittain
March 24 (Reuters) - Music publishers Universal Music Group UMG.AS, Concord and ABKCO have asked a judge in California to rule that U.S. copyright law does not insulate artificial intelligence startup Anthropic from liability for copying their song lyrics to train its AI-powered chatbot Claude.
The publishers' request, filed on Monday in federal court in San Jose, tees up a critical question in the legal battle between creators and tech companies: Does the doctrine of "fair use" apply to the copying of millions of copyrighted works to train AI models?
The publishers argued in Monday's filing that Claude's AI-generated lyrics are not fair use because they are derivatives of the publishers' lyrics that "compete with and dilute the market" for them. The publishers also said that Claude unlawfully reproduces their lyrics on demand without permission.
Spokespeople for Anthropic did not immediately respond to a request for comment on the motion on Tuesday.
The plaintiffs said in a statement that Anthropic has "committed copyright infringement on a massive scale" and that the evidence was "overwhelming."
The lawsuit is one of dozens of disputes between copyright owners such as authors and news outlets and tech giants including OpenAI, Microsoft and Meta Platforms over the training of their AI systems. Amazon and Google-backed Anthropic was the first major AI company to settle one of the cases, agreeing last year to pay a group of authors $1.5 billion to resolve a class-action lawsuit.
The music publishers sued Anthropic in 2023, alleging that it infringed their copyrights in lyrics from at least 500 songs by musicians including Beyonce, the Rolling Stones and the Beach Boys.
All of the pending cases will likely revolve around whether AI systems make fair use of copyrighted material by using it to create new, transformative content. The publishers asked U.S. District Judge Eumi Lee on Monday to rule before trial that Anthropic infringed their copyrights and reject Anthropic's fair use defense.
Anthropic has denied the allegations, but has not yet argued for fair use in the music publishers' case. U.S. District Judge William Alsup in San Francisco said in a separate case last year that Anthropic’s use of books for AI training was “quintessentially transformative," siding with the company on the issue.
Alsup and another judge in Lee's Northern California court issued diverging rulings on fair use in AI training. The publishers said on Monday that unlike the authors in those cases, their record of Claude reproducing their work on demand is "overwhelming."
The case is Concord Music Group Inc v. Anthropic PBC, U.S. District Court for the Northern District of California, No. 5:24-cv-03811.
For the music publishers: Matt Oppenheim of Oppenheim + Zebrak
For Anthropic: Sonal Mehta of Wilmer Cutler Pickering Hale & Dorr
Read more:
Music publishers sue AI company Anthropic over song lyrics
Music publishers fend off Anthropic's bid to dismiss some AI copyright claims
AI copyright battles enter pivotal year as US courts weigh fair use