Music Publishers Ask Court to Reject Anthropic's Copyright Defenses Over AI Lyrics Use
Publishers say Anthropic admitted at least 100 of their songs were in Claude's training data and never once sought a license, calling the infringement "massive scale."

Universal Music Publishing Group, Concord Music Group, and ABKCO asked a federal judge to rule that Anthropic infringed their copyrights and to reject the AI company's fair use defense. The motion for partial summary judgment, filed on Monday, March 23, in the Northern District of California, is the latest step in a case first brought in October 2023 over the alleged infringement of 499 copyrighted musical works.
The filing states that Anthropic admits "at least one Claude model was trained on a dataset containing the lyrics to at least one hundred (100) of Publishers' Works" and that Anthropic "does not deny that the lyrics to Publishers' Works are included in Claude's training data" and "has never sought nor obtained a license from Publishers to use the Works." The suit broadly covers at least 500 songs, with works by the Rolling Stones, Beyoncé, and the Beach Boys among those cited. The original complaint also named specific titles including Katy Perry's "California Gurls," Maroon 5's "Moves Like Jagger," and the Rolling Stones' "Gimme Shelter."
The publishers characterized the conduct in stark terms in their motion. "Having established that Anthropic copied and ingested songwriters' lyrics without permission or compensation, trained its Chatbot (Claude) to serve up those lyrics on demand, and spit out AI-generated derivatives that compete directly with human songwriters, the plaintiffs move for summary judgment. The evidence in this case is overwhelming," the filing states, as quoted by Music Business Worldwide.
The publishers are asking the court to rule in their favor on core elements of the case without a full trial, a step that is only granted when the underlying facts are not meaningfully contested. The motion is accompanied by a 47-page statement setting out 218 "undisputed facts" supported by deposition testimony and internal Anthropic documents.
On fair use, the publishers argue that Anthropic's use was purely commercial, noting the company is a "for-profit technology company valued at $380 billion or more" with a revenue run rate approaching $14 billion, that it copied lyrics in full, and that its output directly competes with licensed lyrics services such as LyricFind and Musixmatch. The publishers represent the company's counsel through Matt Oppenheim of Oppenheim + Zebrak; Anthropic is represented by Sonal Mehta of Wilmer Cutler Pickering Hale & Dorr.
Internal Anthropic communications feature prominently in the filing. An August 2023 internal memo is cited as stating that AI models like Claude "memorize A LOT, like a LOT." Co-founder Benjamin Mann reportedly testified that certain content is "worth memorizing." CEO Dario Amodei is quoted as stating in an April 2024 interview that AI models should not be "verbatim outputting copyrighted content," while also testifying in his deposition that doing so is "against the law."
The internal communications underscore an allegation the publishers have pressed since they filed an amended complaint: that Anthropic intentionally removed copyright management information during dataset curation. U.S. District Judge Eumi Lee declined to dismiss that claim in April 2025. At that time, she wrote that "taking these allegations together (as true), the court can plausibly infer that Anthropic had a reasonable basis to know that the dataset curation and training processes it engaged in, whereby it removed CMI (i.e., its intentional removal of CMI), concealed its own infringement."
Anthropic has broadly denied the allegations but had not yet formally argued fair use in the music publishers' case at the time the motion was filed. The company's legal position exists in a fractured judicial landscape: U.S. District Judge William Alsup in San Francisco ruled in a separate case involving book authors that Anthropic's use of books for AI training was "quintessentially transformative," siding with the company. The publishers contend their case is distinct because of Claude's specific documented ability to reproduce lyrics verbatim on demand.
The filing argues that when Claude outputs lyrics in response to user requests, "Anthropic provides the same service to its users as Publishers' licensees, without Publishers' authorization or any restrictions on that use."
A January 2025 partial settlement between Anthropic and the publishers had served as a temporary truce, defusing a preliminary injunction motion that sought to require Anthropic to implement "effective guardrails" and to prohibit it from creating or using unauthorized copies of publishers' lyrics in future model training. That settlement did not resolve the underlying copyright infringement dispute, which the March 23 summary judgment motion now forces directly before Judge Lee.
This case is separate from the publishers' second, larger lawsuit against Anthropic filed in January 2026, which covers more than 20,000 songs and seeks over $3 billion in statutory damages. That suit also names Anthropic CEO Dario Amodei and co-founder Benjamin Mann as defendants. Anthropic previously settled a class-action lawsuit with a group of authors for $1.5 billion, a resolution that did not carry over to the music publishers' fight and has so far set no precedent binding Judge Lee's court.
Sources:
Know something we missed? Have a correction or additional information?
Submit a Tip

