Music Publishers Sue Anthropic for Alleged Illegal Song Downloads

Music publishers Concord Music Group and Universal Music Group have initiated legal proceedings against artificial intelligence firm Anthropic in the United States, alleging large-scale copyright infringement. According to court documents and reporting by Reuters, the publishers accuse Anthropic of illegally downloading and using more than 20,000 copyrighted songs—including sheet music, lyrics, and full compositions—to train its conversational AI chatbot, Claude.

The lawsuit cites works by renowned artists such as The Rolling Stones, Neil Diamond, and Elton John, alongside material from artists represented by Concord, including Common, Killer Mike, and Korn. The case emerged from evidence uncovered during the discovery phase of the earlier Bartz v. Anthropic lawsuit, in which thousands of unauthorized downloads were found.

Publishers contend that the illegal acquisition and use of these songs directly contributed to the advancement of Anthropic’s product offerings. The lawsuit’s estimated damages exceed $3 billion, potentially making it one of the most significant non-class action copyright disputes in U.S. legal history.

A statement from the complaint alleges, “While Anthropic misleadingly claims to be an AI ‘safety and research’ company, its record of illegal torrenting of copyrighted works makes clear that its multibillion-dollar business empire has in fact been built on piracy.”

This legal team previously led the Bartz v. Anthropic case, which concluded with a $1.5 billion award to 500,000 affected writers and authors. In that matter, each author received $3,000 per work after it was demonstrated that Anthropic had used copyrighted written content to train its AI models.

Anthropic, which carries a reported valuation of around $350 billion, is now facing similar allegations in the musical domain. The current lawsuit relies heavily on findings from the Bartz litigation, which revealed the scale and method of Anthropic’s content acquisition for AI training.

In Bartz, U.S. District Judge William Alsup distinguished between the legality of training AI models on copyrighted works and acquiring such works through piracy. He maintained that while model training might be lawful under certain circumstances, illegally downloading copyrighted material remains a clear violation. This precedent forms the publishers’ core argument against Anthropic’s use of music files.

The publishers allege that Anthropic used illegal means, such as torrenting, to obtain the works in question. Evidence provided during the Bartz discovery phase documented thousands of such downloads, and the current case asserts that this practice extended well beyond text-based works to include a broad range of musical compositions.

The disputed materials are believed to have directly powered the development of Claude, Anthropic’s large language model-based chatbot. Large language models, or LLMs, are a type of AI program trained on extensive datasets to generate human-like text responses and, in some cases, perform a variety of language-based tasks.

The close parallels between Anthropic’s alleged unauthorized use of song material and the practices challenged in Bartz mean this lawsuit could redefine the boundaries of copyright protection in relation to AI training. With a multibillion-dollar claim at stake, the case is set to test established legal guidance and highlight ongoing tensions between creators and AI developers over the use of protected content.

Reference: dataconomy.com