Sony Music Publishing, Warner Chappell and numerous other music publishers have sued Anthropic in the US District Court for the Northern District of California. The complaint accuses the company of illegally downloading and using tens of thousands of copyrighted musical compositions — mostly song lyrics — to train its Claude models.

The suit was filed late Friday and first reported by Music Business Worldwide. An Anthropic spokesperson told TechCrunch: "We disagree with the publishers' claims and we intend to defend ourselves robustly in court."

The founders are named personally

This is the unusual part of the complaint: CEO Dario Amodei and co-founder Benjamin Mann are named alongside the company as individual defendants, for their alleged role in directing and overseeing the torrenting of copyrighted files.

The forty-eight-page complaint states: "Dr. Amodei expressly directed, approved, controlled, and intentionally induced these infringements." The plaintiffs call it "one of the largest and most blatant ongoing thefts of intellectual property in history."

They are seeking up to $150,000 per infringed work, and up to $25,000 per violation for the unlawful removal of copyright management information — copyright notices and other identifying details.

The same weak spot, a second time

Anthropic has already lost this fight once. In September 2025 it agreed to the largest copyright settlement in US history, paying $1.5 billion to authors and publishers for using pirated books in AI training.

What sank the company in that case was not using copyrighted data for training — it was acquiring it through illegal torrents. The new lawsuit aims at exactly that point.

Anthropic allegedly torrented at least seven million books from the pirate libraries LibGen and PiLiMi. The suit treats that downloading as a standalone infringement, regardless of whether those works ever reached a commercial Claude model.

The other allegations

  • Scraping licensed platforms. Lyrics were allegedly pulled from services such as MusixMatch and LyricFind without publisher consent, breaching those platforms' terms of service.
  • Datasets. The complaint challenges the use of Books3, The Pile and Common Crawl, which it says contain unauthorised content.
  • Physical copies. The company is accused of scanning and destroying used songbooks and sheet music collections.

The synthetic data loophole

Technically, this is the most interesting part. Anthropic has publicly denied using LibGen and PiLiMi books directly to train commercial Claude models. The plaintiffs argue that the denial rests on how the company defines "training."

They allege it trained at least one commercial Claude model on synthetic data generated by a non-commercial model that had itself learned from those pirated texts, and that it used such a model to give reinforcement feedback to a commercial Claude model.

If that holds up, it collapses a defence common across the industry: "we did not use pirated data in the commercial model" loses its meaning when an intermediate model launders the data through.

Some of the lawyers behind this suit also represent Concord Music Group and Universal Music Group in a case filed in January, and led the case that produced the $1.5 billion payment. The full scope of the claims will emerge during discovery.