A federal district court in California is hearing arguments in a lawsuit where music publishers accuse AI firm Anthropic of copyright infringement. Concord Music Group, representing the publishers, alleges that Anthropic trained its AI model, Claude, using copyrighted song lyrics without permission. They claim Claude can generate lyrics similar to existing works or even verbatim copies, such as an instance where it produced lyrics for "American Pie" when prompted about Buddy Holly. This unauthorized use, publishers argue, devalues their work and cuts into licensing revenue. Anthropic, however, is asserting the Fair Use doctrine, which permits limited use of copyrighted material for purposes like transformation or criticism. Recently, Anthropic settled a similar $1.5 billion case with book publishers, where the use of pirated material was a significant factor. Anthropic also contends that Claude's outputs are user-driven and that they are not directly responsible for infringement. The publishers are seeking substantial financial penalties for each copyright violation, while also requesting compensation for lost profits. They hope a recent appeals court ruling against fair use in a similar Thomson Reuters case will bolster their arguments. However, the precedent from that ruling may not be binding in California.