Sony Music Publishing and Warner Chappell Music are demanding substantial compensation from Anthropic.

The two companies launched legal proceedings against Anthropic and its cofounders Dario Amodei and Benjamin Mann in a Northern California federal court on Friday.

The filing states that Anthropic and its founders carried out an extensive operation involving illegal downloading and scraping of protected material on a vast scale to create and profit from the Claude family of artificial intelligence models.

The publishers claim Anthropic gathered thousands of copyrighted tracks, among them the 1980s hit Eye of the Tiger, Marvin Gayes Ain't No Mountain High Enough, Mariah Careys All I Want for Christmas is You, and Taylor Swifts Paper Rings.

According to the complaint, Anthropic obtained the material through unauthorized channels including Library Genesis and Pirate Library Mirror. A court decision in June 2025 already found that the company had downloaded millions of pirated books for Claude training.

The suit notes that many of the pirated books contained lyrics and sheet music for hundreds of the publishers compositions, such as Livin On a Prayer, September, Great Balls of Fire, Ramblin Man, and Hallelujah.

The publishers allege that the resulting Claude models can reproduce identical or nearly identical versions of these protected works when responding to user prompts. They further argue that training on such material enables the AI to generate lyrics that compete directly with original human compositions.

The companies have asked for a jury trial and are seeking statutory damages reaching 150000 dollars for each work used in training Claude.

The complaint emphasizes that even groundbreaking technologies must operate within legal limits, and Anthropics models are no exception.

The artificial intelligence sector has encountered numerous copyright challenges since models like Claude and ChatGPT began reshaping industries.

Developers require enormous datasets to build these systems and often resort to questionable acquisition methods. Recent reports have also accused firms of harvesting and destroying antique volumes to feed language models.

In September of last year Anthropic settled a class action over pirated books for more than 1.5 billion dollars. OpenAI continues to face similar claims, including actions brought by The New York Times and Encyclopedia Britannica.