Music Publishers Accuse Anthropic of Copyright Piracy in Claude Training
Sony Music Publishing, Warner Chappell and numerous other music publishers have sued Anthropic, alleging that the AI company used thousands of copyrighted works to train its Claude model through illegal torrenting, scraping and downloading. The complaint, filed late Friday in the U.S. District Court for the Northern District of California, also names Anthropic co-founders Dario Amodei and Benjamin Mann.
The publishers describe the alleged conduct as a “brazen campaign” of intellectual-property theft. They contend that Anthropic obtained copyrighted material without permission and used it in the development of Claude, its AI model family.
Publishers focus on how material was obtained
The case is not framed simply as a dispute over whether copyrighted works may be used for AI training. The complaint alleges “blatant theft” and says Anthropic engaged in illegal torrenting to obtain millions of copies of books, including works containing lyrics and sheet music.
That emphasis on acquisition is significant because it mirrors an issue already tested in litigation involving Anthropic. In the Bartz case, a judge ruled that use of copyrighted works was legal, while obtaining that material through piracy was not, resulting in a $1.5 billion payment order.
The legal backdrop is set by Anthropic copyright ruling on pirated acquisition and its finding on pirated acquisition, while the new complaint extends the allegations to music-related works and a wider claimed collection of materials.
A broader copyright challenge for Claude
Anthropic has faced other intellectual-property claims. Some of the lawyers involved in the new lawsuit also represent Concord Music Group and Universal Music Group in a case filed in January, and they led Bartz v. Anthropic, in which authors alleged that copyrighted works were used to train products including Claude.
The new action is described as particularly broad and builds on those earlier arguments. Its claims combine alleged use of protected works with allegations about the methods used to acquire them, including torrenting and downloading on a large scale.
Anthropic had not responded before publication. The court will determine the claims, but the filing makes training-data provenance a concrete business issue for organizations developing, procuring or deploying generative-AI systems.
Business implication
Companies using generative AI should ensure that vendor assessments address not only model performance and contractual rights, but also the documented origin of training data and the exposure created by unresolved copyright disputes.