Context: In October 2023, eight music publishers, including Universal Music and Concord, sued Claude LLM maker Anthropic in the Middle District of Tennessee for copyright infringement. The complainants alleged that Anthropic used at least 500 songs by musicians such as Beyoncé, the Rolling Stones, and the Beach Boys to train its artificial intelligence model Claude. Last August, after the court transferred the case to the Northern District of California, the companies filed for a preliminary injunction. A hearing for that motion was held in November 2024.
What’s new: Judge Eumi Lee of the United States District Court for the Northern District of California yesterday denied that motion (March 25, 2025 decision (PDF)), ruling that the publishers’ request was too broad and that they had failed to show Anthropic’s conduct caused them “irreparable harm”.
Direct impact: This will give Anthropic a boost in its case against the music publishers, despite the complainants stating that they remain “very confident” in their broader case against the AI provider. Preliminary injunctions are typically quite difficult to get, and this loss will be a major challenge for the publishers, but it doesn’t mean it is completely hopeless. The outcome of this case will be particularly interesting, given Anthropic has not formally denied the publishers’ allegations that Claude’s training set contains the 500 songs-in-suit.
Wider ramifications: This case is also one of a growing number of cases being filed in the U.S. (and recently Germany) over the infringement of copyright in music works, books, and newspapers by AI providers. With the industry moving at such a fast pace, these initial suits will be crucial in forming the case law surrounding AI and its place in the world of music, literature and journalism.
The eight plaintiffs include Concord Music Group, Inc., Capitol CMG, Inc., Universal Music Corp., Songs of Universal, Inc., Universal Music – MGB NA LLC, Polygram Publishing, Inc., Universal Music – Z Tunes LLC, and ABKCO Music, Inc.
This is the original complaint:
In their suit, the complainants alleged that, when prompted, Claude responded with output containing “verbatim or near-verbatim copies” of at least 500 of their songs and this has caused “incalculable harm”.
Anthropic has not yet disputed that Claude’s training material includes the songs-in-suit. “If the works were included in the training set, it is likely because they are commonly found all across the Internet,” the defendant has argued. However, the company has alleged that Claude’s intended purpose is not to reproduce existing materials in response to user queries but “to generate original outputs that meet the full range of customer needs.”
In a statement today, the music publishers said they “remain very confident in [their] case against Anthropic more broadly.”
Meanwhile, a spokesperson for Anthropic said the company was pleased that the court did not grant the publishers’ “disruptive and amorphous request.”
Counsel
The music publishers are being represented by a team at Oppenheim + Zebrak: Matthew Oppenheim, Nicholas Hailey, Audrey L. Adu-Appiah, Jennifer Pariser, Andrew Guerra, and Timothy Chung. As well as Jeffrey G. Knowles at Coblentz Patch Duffy & Bass.
Anthropic is being represented by Latham Watkins’s Joseph Wetzel, Andrew Gass, Brittany Lovejoy, Ivana Dukanovic, Sarang Damle, Allison Stillman, and Rachel Horn.
Anthropic is also facing a class action brought by three book authors over copyright infringement (August 20, 2024 ai fray article). In that case, the plaintiffs are being represented by Susman Godfrey’s Justin Nelson (who also acts for the Authors Guild and at least 28 different authors in a class-action against OpenAI in the Southern District of New York (February 5, 2024 class action complaint)).
This latest suit is one of several that have recently emerged involving AI and copyright infringement in the music industry. Last November German music rights collecting society GEMA filed a lawsuit concerning the use of song lyrics in the training of, and their output by, OpenAI (November 14, 2024 ai fray article). This came after a group of publishers (including some of the plaintiffs in this case), such as UMG, Capitol Records, Sony Music and Warner, sued AI provider Suno — an AI service provider that generates music from its internal repertoire — in the United States District Court for the District of Massachusetts (June 24, 2024 complaint (PDF)). In January, GEMA also sued Suno, alleging that the company generated “audio content that is confusingly similar” to various famous original songs (January 25, 2025 ai fray article).