top of page
trademark breadcrumb.png

Music Publishers v. Anthropic

Music industry giants challenge AI firm Anthropic for alleged unauthorized use of song lyrics in training its AI chatbot Claude.

 

Short Description


Universal Music Publishing Group, Concord, and ABKCO filed a copyright infringement lawsuit against Anthropic (maker of the AI model Claude), claiming the company used at least 500 copyrighted song lyrics without permission to train its AI. The publishers seek damages, injunctions, and transparency about Anthropic’s use of their materials. The case is ongoing in U.S. federal courts.

 

Facts


1.     In October 2023, a lawsuit was filed by three music publishers in a Tennessee federal court, accusing Anthropic of training Claude on song lyrics without a license and enabling reproduction of lyrics in responses.


2.     The complaint cited specific songs (e.g., Sam Cooke’s "A Change Is Gonna Come," Beyoncé’s "Halo") and alleged widespread, systematic infringement.


3.     The lawsuit sought a permanent injunction, monetary damages (statutory up to $150,000 per work), attorney’s fees, and disclosure of how the training data was obtained


4.     In December 2024, the court approved partial ‘guardrails’ requiring Anthropic to implement safeguards in Claude to prevent infringing output of song lyrics.


5.     In March 2025, Judge Eumi Lee denied the publishers' request for a preliminary injunction, finding they could not demonstrate irreparable harm.


6.     At the same time, the court dismissed publishers’ claims on contributory infringement, vicarious infringement, and DMCA violations—with leave to amend—though direct copyright claims remain.


7.     The court also denied requests for access to personal user data, emphasizing privacy protection.


8.     In May 2025, an issue arose where Anthropic's expert report included a fabricated citation generated by Claude; Anthropic’s law firm accepted responsibility and implemented stricter review measures.

 

Findings


  • The court recognized direct copyright infringement claims still stand, though other infringement theories were dismissed

  • Guardrails were established to reduce the chance of Claude producing infringing lyrics

  • Motion for preliminary injunction was denied—publishers failed to prove irreparable harm 

  • The court protected Anthropic users’ privacy by refusing to turn over personal identifying information

  • The AI hallucination in a legal citation raised concerns about AI accuracy in legal proceedings 

 

Suggestions (Legal Significance)


  • AI companies must build robust safeguards to prevent generation of copyrighted content.

  • Courts require concrete evidence of irreparable harm before granting injunctions—it’s not enough to claim potential harm.

  • Privacy considerations can limit discovery—even in high-stakes litigation, user data may be protected.

  • AI reliability is critical in legal contexts—errors like hallucinated citations must be mitigated through manual review and oversight.

 

Judgment / Status


  • No final judgment yet.

  • The case is ongoing in the U.S. District Court, Northern District of California (Case: Concord Music Group Inc. v. Anthropic PBC).

  • Guardrails are in place; only direct infringement remains litigated.

  • Preliminary injunction denied; privacy and AI-hallucination issues highlighted.

  • Ongoing developments include potential amendment filings and continued proceedings.

 
 
 

Comments

bottom of page