A federal judge has allowed music publishers to continue pursuing copyright infringement claims against Anthropic, rejecting the artificial intelligence company’s motion to dismiss parts of the lawsuit.
US District Judge Eumi Lee on Monday ruled that Universal Music Group, Concord Music Group and ABKCO can press forward with claims that Anthropic bears legal responsibility when users of its Claude chatbot generate copyrighted lyrics.
The decision, which you can read in full here, keeps alive key portions of a lawsuit filed in 2023 alleging infringement of lyrics from over 500 songs by artists including Beyoncé, the Rolling Stones, and the Beach Boys.
The ruling marks the second attempt by Amazon and Google–backed Anthropic to dismiss the infringement claims. Lee previously granted a dismissal motion but allowed the publishers to refile amended complaints.
This time, the judge found the publishers adequately argued that Anthropic could have been aware of user infringement and profited from allowing it to continue by using “guardrails” designed to prevent copyright violations. The judge’s ruling addressed secondary liability and DMCA claims.
“Based on these allegations, Anthropic had actual knowledge of specific acts of infringement by Claude users with respect to Publishers’ lyrics.”
Eumi Lee, United States District Judge
“Publishers claim that, in developing and improving these guardrails, Anthropic ‘collected Claude user prompts and output data, including specific infringing output copying lyrics.’ And Anthropic supposedly became aware of the user prompts seeking such lyrics because, each time this occurred, the guardrails detected the users’ actions and Anthropic received a corresponding notice.
“Based on these allegations, Anthropic had actual knowledge of specific acts of infringement by Claude users with respect to Publishers’ lyrics,” Judge Lee wrote on Monday (October 6).
The judge also decided to keep the claims that Anthropic is responsible for its users’ copyright violations, rejecting Anthropic’s argument that the publishers failed to show direct financial benefit from the alleged violations. Lee also found it was plausible that Anthropic profits each time users request lyrics.
“Publishers allege that Anthropic ‘is paid every time . . . end users submit[] a request for Publishers’ song lyrics, and it is paid again every time its Claude API generates output copying and relying on those lyrics.’”
The judge wrote: “It is plausible that the availability of Publishers’ lyrics draws customers to use Claude because, as Publishers contend, Claude would not be as popular and valuable as it is but for ‘the substantial underlying text corpus that includes Publishers’ copyrighted lyrics.’”
Additionally, the judge let stand a separate claim under the Digital Millennium Copyright Act. The music publishers accuse Anthropic of intentionally removing information about copyright management information (CMI) from its AI training datasets. They claim Anthropic’s co-founders used a tool called Newspaper to get rid of these ownership details better than other options to hide their actions.
“It is plausible that the availability of Publishers’ lyrics draws customers to use Claude because, as Publishers contend, Claude would not be as popular and valuable as it is but for ‘the substantial underlying text corpus that includes Publishers’ copyrighted lyrics.’”
Eumi Lee, United States District Judge
Judge Lee wrote: “In that case, the court found that plaintiffs sufficiently alleged that defendant ‘intentionally removed CMI to conceal copyright infringement.’ … It reached this conclusion based on plaintiffs’ allegations that defendant (1) was aware its AI system was particularly prone to “memorizing and generating outputs of CMI unless CMI was removed from its training data’; and (2) took steps to reduce the chances that its program would produce CMI.”
The dispute is one of several high-profile cases testing whether AI companies are legally allowed to use copyrighted work when training their AI models. Other big companies like Microsoft and Meta, and AI firms like OpenAI, Suno and Udio are also facing similar lawsuits. Anthropic is among the first AI companies to settle one of these cases, agreeing in August to pay $1.5 billion to a group of authors.
In the case against music publishers, US Magistrate Judge Susan van Keulen made it clear how frustrated the court is with both parties over the handling of discovery. The court has ordered five joint discovery submissions on Monday, with another eight expected to come. In her ruling, Judge van Keulen threatened to sanction both sides for failing to comply with federal rules of civil procedure.
The judge wrote in a separate court document: “Counsel with primary responsibility for discovery will appear in-person on October 10, 2025 at 9:30 a.m. to show cause as to why both sides should not be sanctioned for the failure to comply with the Federal Rules of Civil Procedure and this Court’s standing order regarding civil discovery.”
UMG, Concord and ABKCO on Monday said they are “heedful of the Court’s directives, and will cooperate with Defendant Anthropic PBC to resolve as many disputes as possible before the scheduled hearing this Friday, October 10.”
Publishers are seeking to expand their lawsuit to add a new claim against Anthropic for allegedly distributing copyrighted lyrics without a license, based on evidence that Anthropic used BitTorrent to acquire unauthorized material.Music Business Worldwide
