Music Publishers Sue Anthropic and Founders Over Torrented Datasets

Major music publishers including Sony and Warner have filed a copyright infringement lawsuit against artificial intelligence company Anthropic, naming chief executive Dario Amodei and co-founder Benjamin Mann as individual defendants.

The complaint, lodged late last Friday in the U.S. District Court for the Northern District of California, alleges that Anthropic downloaded and utilized pirated datasets that contained songbooks, sheet music collections, and song lyrics. The filing accuses the defendants of direct and contributory copyright infringement through torrenting, unauthorized web scraping, and AI model training, as well as DMCA violations for removing copyright management information.

According to TorrentFreak, the lawsuit heavily leverages evidence uncovered during previous litigation against the AI company. Last September, Anthropic agreed to a $1.5 billion settlement with book authors in the Bartz v. Anthropic class action involving seven million pirated titles. The music publishers argue that the exact same dataset contained significant amounts of their proprietary sheet music and lyric compilations.

## Unpacking Internal Communications

The lawsuit draws directly from internal Anthropic Slack exchanges revealed in the Bartz proceedings. In those communications, Mann allegedly discussed downloading files from pirate repository Library Genesis, known as LibGen, and shared screenshots of the process. Mann referred to a script he designed to automate the downloads as "a cute little libgen babysitter," according to the complaint reported by TorrentFreak.

The publishers claim the peer-to-peer nature of the torrent downloads exacerbated the legal injury. As the complaint notes, "Indeed, each pirated work that Defendants torrented was likely shared thousands if not tens of thousands of times, depriving Music Publishers of substantial revenue."

## Why It Matters

By targeting the founders personally alongside the corporate entity, the music publishing industry is attempting to establish executive liability for AI training data acquisition. As AI companies seek blanket training corpora across the open web, rightsholders are using discovery material from earlier class actions to launch fresh, high-stakes legal challenges.