Universal Music Group (UMG) has filed another sweeping copyright lawsuit against artificial intelligence firm Anthropic, this time potentially seeking more than $3 billion in damages by focusing on accusations that the Claude creator used “pirate libraries” of music to build its models.
Six months after a judge ruled in a separate case that AI companies could face huge damages for training models on illegally-acquired datasets, UMG says in a Wednesday (Jan. 28) lawsuit that that’s exactly what Anthropic did to teach Claude to spit out new song lyrics.
Related
The new case, which comes as UMG and other majors have begun striking huge settlements with other AI firms, claims that Anthropic’s “multibillion-dollar business empire” was “built on piracy.”
“Publishers recognize the great potential of ethical AI as a powerful tool for the future,” write lawyers for UMG and numerous other publishers in the lawsuit, obtained by Billboard. “However, it remains crucial that AI technology be developed and employed ethically and responsibly, in a manner that protects the rights of publishers and songwriters, their livelihoods, and the creative ecosystem as a whole. Doing so will ensure that AI enhances — rather than imperils — human creativity.”
The lawsuit claims Anthropic illegally acquired sheet music to “hundreds or thousands” of songs owned UMG and the other publishers, including iconic tunes like The Rolling Stones’ “Wild Horses,” Neil Diamond’s “Sweet Caroline,” Elton John’s “Bennie and the Jets,” Creedance Clearwater Revival’s “Have You Ever Seen The Rain,” Maroon Five’s “She Will Be Loved,” and Katy Perry’s “California Gurls.”
The new case adds to an earlier lawsuit filed against Anthropic in 2023 by UMG, Concord Music Group, ABKCO and a slew of other music publishers, claiming the AI firm was violating copyrights en masse by using songs without authorization to teach its Claude models how to create new lyrics. Like other AI companies facing a flood of such litigation over the past three years, Anthropic strongly denies those allegations, arguing that the training is instead a so-called “fair use” of copyrighted materials.
In a ruling last year in a separate case filed against Anthropic by book authors, a federal judge ruled that AI training itself is covered by fair use — a major legal win for the AI industry. But the ruling came with a huge catch: the judge ruled that tech companies must legally acquire the training materials in the first place and would likely be on the hook for enormous infringement damages if they didn’t.
In the wake of that ruling, UMG and the music publishers moved to update their lawsuit with new allegations that Anthropic had relied on millions of books containing sheet music it illegally downloaded from file-sharing sites. But that bid to amend the case was denied by the judge as procedurally improper.
The result of that decision is Wednesday’s new lawsuit, in which UMG and the other publishers expressly cite that procedural backstory and say they were forced to file a new case that is “distinct and separate” to attack Anthropic’s alleged piracy.
“Publishers recently discovered that defendants downloaded by torrenting an enormous number of unauthorized copies of publishers’ works from illegal shadow libraries to avoid paying for those works,” the publishers write. “To the extent defendants now try to absolve themselves of liability for this blatant theft by claiming that Anthropic later used some subset of these stolen works for AI training, any such claimed use is irrelevant.”
A spokesperson for Anthropic did not immediately return a request for comment on Wednesday.
In a statement to Billboard, the music companies said they were forced to file the second case because of Anthropic’s “persistent and brazen infringement” in using illegally acquired works: “In total, we are suing for infringement of more than 20,000 songs, with potential statutory damages of more than $3 billion. We believe this will be one of the largest (if not the single-largest) non-class action copyright cases filed in the U.S.”
