The music industry’s fight over artificial intelligence has entered another major courtroom battle.

Sony Music Publishing and Warner Chappell Music have sued AI company Anthropic, alleging that copyrighted musical compositions were improperly obtained and used to train the company’s Claude artificial-intelligence models.

The federal lawsuit was filed in California and adds another major music-industry challenge to the way generative-AI companies acquire the enormous quantities of material used to develop their systems.

These are allegations, not judicial findings. Anthropic disputes the claims and is expected to defend itself in court.

What Sony and Warner allege

The publishers allege that Anthropic obtained copyrighted lyrics and sheet music without authorization and used the material in Claude’s training data.

Reuters reported that the complaint identifies music associated with The Beatles, Taylor Swift, Michael Jackson and hundreds of other artists. It also alleges that Claude can reproduce copyrighted lyrics when prompted and generate new lyrics that compete with publishers’ works.

The publishers are seeking statutory damages of up to $150,000 for each work they say was infringed, as well as a court order preventing Anthropic from using the material.

Anthropic had not provided Reuters with an immediate response to the new complaint at publication time. The company has disputed copyright claims involving Claude in other litigation.

The distinction between an allegation and a proven violation is essential. A complaint presents the plaintiffs’ version of events; the court has not determined that Anthropic infringed these copyrights.

Using AI is not the same question as training AI

The lawsuit arrives as musicians and producers increasingly debate whether artificial intelligence should be viewed as a threat or as another creative technology. But two separate issues are often blended together.

The first is whether an artist should be able to use AI creatively. The second is how the AI system obtained the material from which it learned.

A musician can believe that AI is a legitimate production tool while also arguing that copyrighted recordings, lyrics or compositions should not be copied without authorization to create that tool.

That distinction will become increasingly important as AI moves deeper into recording studios.

Supporters of AI-assisted music frequently compare the technology to innovations that once frightened parts of the music business. Sampling, synthesizers, drum machines, Auto-Tune and digital recording all generated arguments about authenticity before becoming part of the industry’s creative vocabulary.

Generative AI presents a different complication. A drum machine does not need to study a library of copyrighted songs before producing a kick drum. A generative model may learn patterns from enormous datasets containing language, composition and other creative material.

The controversy is therefore not only about what an artist creates with AI. It is also about what was allegedly used to build the AI itself.

A copyright question with industry-wide consequences

The music business has spent decades developing licensing systems for songs that are sampled, reproduced, publicly performed or synchronized into films and television. Generative AI is forcing courts to confront a newer question: does training a model on copyrighted creative work require permission or compensation?

Technology companies and copyright owners have sharply different interpretations of existing law. Lawsuits brought by authors, publishers, record companies and news organizations are now testing those boundaries.

The Sony and Warner case against Anthropic could therefore carry consequences far beyond Claude. The fight is ultimately about who controls the economic value created when decades of human creativity become training material for machines.

As AI becomes more deeply integrated into entertainment, that question is unlikely to disappear.