What just happened? Major music publishers have filed a $3 billion federal copyright lawsuit against AI giant Anthropic, alleging that the company engaged in "flagrant piracy" of more than 20,000 copyrighted musical works as part of the data used to train its Claude generative AI models.
The complaint, filed January 28 in the US District Court for the Northern District of California, could become one of the largest non-class action copyright cases in US history if damages are awarded as sought.
The plaintiffs – led by Universal Music Group (UMG), Concord Music Group, and ABKCO Music – allege Anthropic systematically downloaded copyrighted songs, sheet music, lyrics, and musical compositions from pirate sources and incorporated the content into its training data without authorization.
According to the court filings, this represents a dramatic escalation from an earlier complaint that initially cited just 500 infringing works.
The same legal team represented authors in the Bartz v. Anthropic case, which focused on Anthropic's use of copyrighted books. That dispute culminated in a $1.5 billion settlement, with roughly $3,000 paid per infringing work after Judge William Alsup ruled that acquiring pirated content was illegal even if training on copyrighted material itself could fall under fair use.
In their complaint, the music publishers assert that Anthropic not only benefited commercially from the alleged infringement but also failed to secure licenses for the use of these works.
The suit names Anthropic's co-founders, including CEO Dario Amodei and co-founder Benjamin Mann, as individual defendants.
Statutory copyright law allows for significant per-work damages when infringement is established. With tens of thousands of works at issue, the publishers argue that more than $3 billion is justified. They claim that the scale and scope of alleged misconduct far exceed what could be framed as incidental data usage, characterizing it instead as systematic and unauthorized distribution of copyrighted material.
The case could set a precedent for how AI companies can source and use copyrighted material. A ruling in favor of the publishers could accelerate the adoption of licensed training datasets or even push toward compulsory licensing.
Anthropic has not responded to the new lawsuit at the time of publication.