Major music publishers have launched what could become one of the costliest copyright battles in AI history, filing a sweeping lawsuit against Anthropic and two of its co-founders over the alleged mass theft of tens of thousands of protected works.
The Lawsuit and What's at Stake
Sony Music Publishing, Warner Chappell, and a coalition of music publishers filed suit on August 29 in the U.S. District Court for the Northern District of California, targeting Anthropic as a company and personally naming co-founders Dario Amodei and Benjamin Mann as individual defendants [1][2]. The complaint seeks up to $150,000 per infringed work, plus up to $25,000 for each instance in which copyright management information was stripped from content [1]. With "tens of thousands" of works allegedly involved, the total exposure could run into several billion dollars if a court awards maximum damages [1].
The publishers describe the alleged conduct in stark terms. The complaint calls it "one of the largest and most blatant ongoing thefts of intellectual property in history," accusing Anthropic and its founders of conducting "a brazen campaign of illegally torrenting, scraping, and downloading copyrighted works on a massive scale in order to develop, operate, and reap enormous profits" from the Claude series of AI models [1].
Anthropic pushed back. "We disagree with the publishers' claims and we intend to defend ourselves robustly in court," a company spokesperson told TechCrunch [2].
Alleged Methods of Acquisition
The lawsuit lays out specific accusations about how Anthropic allegedly obtained the copyrighted material. According to the complaint, co-founder Benjamin Mann personally used BitTorrent to download more than five million pirated books, while other employees downloaded at least two million additional titles from Pirate Library Mirror [1]. The suit also claims Anthropic scraped song lyrics from services like MusixMatch and LyricFind — platforms that had themselves paid licensing fees to the labels whose content was allegedly taken [1].
Among the specific songs the complaint identifies as appearing in Anthropic's training data are Marvin Gaye and Tammi Terrell's "Ain't No Mountain High Enough," Bon Jovi's "Livin' On a Prayer," Earth, Wind & Fire's "September," Leonard Cohen's "Hallelujah," and Taylor Swift's "Paper Rings" [1].
A Pattern of Legal Pressure
This lawsuit does not arrive in a vacuum. Anthropic has faced a mounting wave of intellectual property litigation, and the new filing builds directly on prior cases [2]. Some of the same attorneys behind this suit also represented Concord Music Group and Universal Music Group in a separate case filed in January, and were involved in the landmark Bartz v. Anthropic matter, in which a group of authors accused Anthropic of using copyrighted works to train Claude [2].
That earlier case resulted in a $1.5 billion settlement — a significant legal milestone [1][2]. Notably, the judge in Bartz ruled that while using copyrighted works for AI training may be permissible, acquiring that content through piracy is not [2]. The current music publishers' suit leans heavily on that distinction, centering its argument on the alleged illegality of how the training data was obtained rather than solely on the act of training itself [2].
Beyond Bartz, Anthropic has also faced separate suits from Universal Music Group, Concord, ABKCO, BMG, and Round Hill Music [1]. The new Sony and Warner Chappell filing is described as particularly broad, extending the legal theory developed in prior cases and adding new allegations of "flagrant piracy" through torrenting to obtain millions of book copies, some of which contained lyrics and sheet music [2].
Why Music Publishers Are Moving Now
The timing reflects a broader industry calculation. Having watched the Bartz case establish that piracy-based data acquisition is legally vulnerable — and result in a billion-dollar-plus payout — music publishers appear to be applying the same framework to their own catalogs. The involvement of the same legal team across multiple cases suggests a coordinated litigation strategy designed to press AI companies on the specific question of data provenance rather than the more contested fair-use arguments around training generally.
The personal naming of Amodei and Mann as defendants is also notable. Holding individual founders liable, rather than limiting claims to the corporate entity, raises the personal stakes and signals that plaintiffs intend to pursue the most aggressive legal posture available to them [1].
What to Watch Next
Several threads will determine how this case unfolds. Courts will need to assess whether the evidence of torrenting and scraping meets the legal threshold for willful infringement — a finding that would justify the maximum per-work damages the publishers are seeking. The Bartz precedent is favorable to plaintiffs on the piracy question, but the scale of claimed damages here is substantially larger.
Anthropic's defense strategy will also be worth tracking. The company has signaled it will contest the claims vigorously, but it must now manage simultaneous litigation across multiple fronts while continuing to develop and commercialize Claude. Whether any of these cases move toward settlement — and at what price — may ultimately shape how the entire AI industry accounts for the origins of its training data.
