Universal Music Group’s Legal Battle
Universal Music Group and its colleagues—the muses of rightsholders—filed lawsuits against the artificial intelligence platform Anthropic this week, seeking multi-million dollar damages.
However, the judge in the case suggested that Anthropic may still face significant financial repercussions for the music industry, which publishers state they will continue to pursue “energetically”.
As reported earlier today (March 26), a federal judge in California requested UMG and other music publishers (including Concord and Abkco) to prevent the AI company from using their songs to train AI models.
In an order issued on Tuesday (March 25), Judge Eum K. Lee of the US District Court in the Northern District of California denied the publishers’ petition for a preliminary injunction that would ban Anthropic from using copyrighted materials controlled by publishers for future AI model training.
It is noteworthy that while the court denied the preliminary injunction, it did issue two separate, yet related orders on March 25, granting UMG and other plaintiff publishers important investigative tools to potentially strengthen their legal arguments.
Following the decision, a group of music publishers involved in the case stated to MBW that they “remain very confident” in their legal battle against Anthropic.
Legal Statements and Responses
You can read the application of music publishers in its entirety below:
“Despite the narrow court decision, we remain very confident in our case against Anthropic. In response to our preliminary petition for a court ban, Anthropic has already lost the advantage of our claims against the compromised results of our songs, having entered into a reservation that demands ‘guardrails’ to prevent such violations of the results, thereby addressing a critical aspect of our motion.”
Joint statement of music publishers
Publishers added: “The court clearly refused to resolve the fair use issue or the validity of our claims for violations at this early stage of the proceedings. While the court denied the remaining part of our petition that would require Anthropic to refrain from training on our texts, it did so on narrow grounds, indicating that our damages could ultimately be compensated through monetary damages that we intend to pursue.”
“This case remains vital for protecting creators from the wholesale theft of their copyright, specifically regarding Anthropic and other artificial intelligence companies. We expect that, as we proceed with this case and develop a complete record of discovery, our claims will be substantiated.”
Within the court’s decision this week, there was concern regarding the scope of the requested injunction, which would encompass not only the 500 works cited in the lawsuit but potentially “hundreds of thousands” of songs.
“(Music) publishers are essentially asking the courts to define the licensing landscape for AI training, where the crucial issue of fair use remains unresolved,” Judge Lee noted in a 13-page order, which you can read in full here.
The court observed that, according to the publishers’ own evidence, “the market for AI licensing has actually grown during this litigation, rather than contracted”—citing companies like OpenAI that have engaged in monetized content licensing deals over the past year with IP Owners such as News Corp, Financial Times, and Shutterstock.
Judge Lee asserted that this fact indicates that a commercial licensing structure for AI training already exists (and is expanding), suggesting that if UMG and others eventually win their lawsuit against Anthropic, the court could assess monetary damages rather than an injunction at this stage.
Judge Lee emphasized that the latest ruling did not resolve the question of whether training generative AI models with protected copyright materials constitutes a violation or falls under “fair use,” which remains an “open issue” to be addressed in subsequent proceedings.
UMG, Concord, and ABKCO sued Anthropic in 2023, alleging that the company trained its AI chatbot, Claude, on the lyrics of at least 500 songs from artists including Beyoncé, the Rolling Stones, and the Beach Boys without permission.
As noted in our earlier coverage of this development, this week marked only one aspect of the legal claims made by publishers against Anthropic, and importantly, it did not affect previous favorable rulings for the publishers.
The case effectively has two main components:
- First, can the copyrights in Claude’s outputs (what users see when interacting with AI) be protected;
- Secondly, can Anthropic use copyrighted texts as training data (i.e., to create their own AI models).
Publishers previously achieved a significant victory on the first issue in January 2025 when Anthropic agreed to implement “guardrails” to prevent Claude from reproducing copyrighted content in its outputs to users.
This agreement, which the court approved, remains intact and was not challenged in this new ruling.
At that time, UMG stated: “We are pleased that the court approved these crucial ‘guardrails’—Anthropic must implement them to protect publishers’ copyrights, effectively acknowledging the validity of our claims regarding violations of Claude’s outputs.”
Today’s developments represent a significant milestone in the ongoing copyright battle between copyright holders and artificial intelligence companies. The Anthropic case is one of several legal disputes concerning the use of copyrighted material for AI training.
The outcome could have far-reaching implications for both creative industries and the evolution of artificial intelligence technologies.
In light of the trial, Anthropic acquired an additional USD 4 billion investment from Amazon at the end of 2023, bringing the total amount of Amazon’s investment in artificial intelligence to USD 8 billion.
