Sony Music Publishing and Warner Chappell have filed a 48-page lawsuit against Anthropic, naming both the company and its founders Dario Amodei and Benjamin Mann personally. The complaint, filed in Northern California, seeks up to $150,000 per song across tens of thousands of compositions — potentially exposing Anthropic to $1.5 billion in damages.
This isn’t the first time a record label has sued an AI company over training data, but the scale and the personal naming of executives make it different. Sony and Warner are not just challenging Anthropic’s right to use song lyrics in training — they’re arguing that the company knowingly built its models on copyrighted material without permission.
Anthropic has called the lawsuit “recycled” — suggesting the claims are baseless and merely repurpose arguments already made in earlier cases. However, the stakes are notably higher this time. A judgment in favor of the record labels could establish precedent affecting every AI company that trains on music, text, or other creative works.
The timing matters. The AI industry has been operating under relatively untested assumptions about fair use of training data. Courts have yet to deliver a definitive ruling on whether training large language models constitutes infringement. A loss for Anthropic would force the entire sector to reconsider how it sources training material — or face similar lawsuits.
For now, Anthropic faces the prospect of a lengthy legal battle alongside its existing operations. The company recently posted its first profit of $10.9 billion in Q2 2026, but this lawsuit could become a significant distraction and potential liability. Other AI labs — including OpenAI and Google — are watching closely. Whatever ruling emerges from this case will likely shape the industry’s approach to copyrighted content for years to come.