IP Law Daily, COPYRIGHT NEWS: Music industry sues Anthropic over alleged large-scale copyright piracy, (Jan 29, 2026)

By George Basharis, J.D.

Lawsuit accuses the AI developer building and expanding its Claude models by illegally torrenting songbooks and copying lyrics at scale, then monetizing the resulting outputs.

A coalition of leading music publishers has filed a copyright infringement complaint accusing Anthropic and two of its founders of building and expanding their artificial intelligence business through systematic piracy of copyrighted musical works, including the alleged use of BitTorrent to download songbooks from pirate libraries and the continued copying of lyrics in AI training and output. The lawsuit frames the conduct as distinct from earlier litigation over AI-generated lyrics and seeks to hold the company and its executives liable for direct and secondary infringement tied to both mass torrenting and ongoing model deployment (Concord Music Group, Inc. v. Anthropic PBC, No. 5:24-cv-03811-EKL (N.D. Cal. filed Jan. 28, 2026)).

The plaintiffs include Concord Music Group, multiple Universal-affiliated publishing entities, and ABKCO Music, which together say the defendants copied hundreds of thousands of musical compositions without authorization, undermining established licensing markets and enabling the distribution of infringing content at internet scale. The complaint was filed in the U.S. District Court for the Northern District of California and seeks damages and injunctive relief.

Alleged infringement. The lawsuit advances two primary claims. First, it alleges that Anthropic and its founders downloaded enormous quantities of copyrighted books, including songbooks and lyric collections containing publishers’ works, from well-known pirate libraries using BitTorrent. Second, it alleges that Anthropic trained and released successive versions of its Claude large language models using unauthorized copies of lyrics and musical compositions, resulting in outputs that reproduce or derive from protected works and compete with licensed offerings.

On the torrenting allegations, the publishers contend that Anthropic deliberately used BitTorrent to obtain pirated copies from Library Genesis and Pirate Library Mirror, sites described as notorious for hosting infringing content. Because BitTorrent operates as a peer-to-peer system in which users simultaneously upload and download file fragments, the complaint asserts that Anthropic not only acquired unauthorized copies but also redistributed them to others during the process, violating reproduction and distribution rights with each transaction.

The publishers say they discovered the alleged torrenting only after public rulings and filings in separate litigation against Anthropic revealed internal practices and testimony. They allege that at least one founder personally participated in the activity and that senior leadership discussed and authorized the downloads despite knowing the sources were illegal.

Stockpiled texts. The complaint describes Anthropic as maintaining a vast central library of text assembled through scraping, copying physical books, and torrenting pirate libraries. According to the publishers, that library was retained and reused for multiple purposes, including training new generations of Claude models. The publishers argue that even if some materials were later filtered or used selectively, the initial acts of copying constituted standalone infringement.

The suit further alleges that Anthropic continued to release and monetize new Claude models after earlier complaints were filed, including paid consumer subscriptions and enterprise offerings, all while relying on training data that incorporated unauthorized lyrics. The publishers say Anthropic earns revenue each time users prompt the models and receive text outputs, including outputs that reproduce or rely on copyrighted lyrics, and that commercial customers integrating the Claude API similarly generate infringing outputs.

Outputs and use. Beyond training, the complaint focuses on model outputs and user behavior. It alleges that Anthropic expected and encouraged lyric-related use cases, citing internal testing and examples in which employees and contractors prompted models for song lyrics or derivative text. According to the publishers, Anthropic monitored user prompts and outputs and continued to deploy models despite awareness that users were requesting copyrighted lyrics and that the models were capable of generating them.

The publishers allege direct infringement based on Anthropic’s own copying and public display of lyrics in outputs, as well as contributory and vicarious infringement for enabling and profiting from users’ infringing requests. They also allege violations tied to the removal or absence of copyright management information when lyrics are reproduced without attribution.

Market harm. Central to the complaint is the assertion that Anthropic’s practices undercut existing and emerging licensing markets. The publishers say they routinely license lyrics and musical compositions for print, digital books, aggregators, websites, and other platforms, and that authorized options are widely available to consumers. By providing free or subscription-based access to AI-generated lyrics and lyric-like outputs trained on unauthorized copies, the publishers argue that Anthropic substitutes for licensed services and depresses the value of their catalogs.

The complaint also points to alleged competitive harm from AI-generated lyrics and songs that draw on publishers’ works and compete with original compositions. According to the publishers, the scale of copying makes damages difficult to quantify and threatens incentives for songwriters and publishers to invest in creative output.

Prior litigation. The publishers characterize the case as a follow-on to earlier litigation over AI training and outputs but emphasize that the torrenting claims are separate and were not litigated previously because the conduct was not disclosed during discovery. They allege that when they sought to amend earlier pleadings to add torrenting claims, Anthropic opposed the amendment on the ground that it would fundamentally change the case, prompting the publishers to bring a new action focused on those allegations.

Claims and relief. The complaint asserts multiple counts of direct copyright infringement based on torrenting and AI training and output, along with contributory and vicarious infringement. It seeks statutory and actual damages, permanent injunctive relief barring further unauthorized copying and distribution, and other remedies available under the Copyright Act.

Anthropic and the individual defendants are alleged to be subject to personal jurisdiction in the Northern District of California based on residence, headquarters, and alleged acts of infringement occurring in the district. The case places renewed focus on how courts will assess large-scale data acquisition practices for AI development, particularly where plaintiffs allege deliberate use of pirate sources and ongoing commercial exploitation.

The Case is No. 5:24-cv-03811-EKL.

Attorneys: Alexander Kaplan (Oppenheim & Zebrak, LLP) for Concord Music Group, Inc. Sonal N. Mehta (Wilmer Cutler Pickering Hale and Dorr LLP) for Anthropic PBC.

Companies: Concord Music Group, Inc.; Anthropic PBC

News: Copyright TechnologyInternet CaliforniaNews