Music publishers sue Anthropic over pirated training data

Sony Music Publishing, Warner Chappell and a group of other music publishers have sued Anthropic, along with co-founders Dario Amodei and Benjamin Mann, in the U.S. District Court for the Northern District of California. As TechCrunch reports, the publishers describe what they call a "brazen campaign" of illegally torrenting, scraping and downloading copyrighted material to train Claude. The complaint focuses on how the works were obtained, alleging that Anthropic pulled in millions of pirated copies of books, some of which contained song lyrics and sheet music. An Anthropic spokesperson said the company disagrees with the claims and intends to defend itself in court.

The timing is what makes this more than routine legal noise. In July 2026, Anthropic lost the Bartz case and was ordered to pay $1.5 billion, after a judge drew a sharp line: training on copyrighted text can be lawful, but acquiring that text through piracy is not. The publishers are now pointing at the same weak spot. Earlier suits from Concord and Universal Music Group in January 2026 took a similar angle, so a pattern is forming around how the training data was sourced rather than whether training itself is fair use.

Why it matters

If you build on frontier models, the legal risk is shifting from the abstract question of fair use to a concrete one: where did the training data come from and can the vendor prove it. A run of piracy-based rulings could push labs to pay for licensed corpora, which would change model costs and possibly which data future models are trained on.