Suno’s Quest for Secrecy: A Look into the Legal Battle over AI Training Data
Suno, an innovative player in the AI music generation space, is currently embroiled in a legal showdown with powerhouse music labels Universal Music Group (UMG) and Sony Music Entertainment. At the heart of the matter is a request from Suno to keep the exact number of audio files used to train its generative AI model under wraps. The stakes are high, as they argue that disclosing this information would give competitors a potentially unfair advantage.
The Court Filing and Its Context
On May 29, Suno submitted a statement to the U.S. District Court for the District of Massachusetts. This filing was a direct response to a request from Inner City Press, which advocated for the unsealing of specific training data evidence pertinent to the copyright infringement claims brought forth by UMG and Sony. Suno clarified that it only seeks to keep secret one key metric—the total volume of audio files allegedly used in training its model.
Suno’s legal team contended that their request for confidentiality is about a narrow piece of information, as they do not aim to conceal details like the identity of specific recordings from the labels’ claims. In fact, UMG and Sony have already indicated in their court filings that there are “millions” of sound recordings involved in the case.
Reasons Behind the Request
Suno’s Chief Technology Officer, Georg Kucsko, provided a compelling argument for why this specific figure should remain undisclosed. Kucsko asserted that in the fast-paced world of generative AI, the size of a training corpus is not just a number; it reflects critical technical decisions and strategic considerations related to model performance. If competitors were to access this figure, they could potentially benchmark their systems against Suno’s, glean critical insights into the company’s model design, and exploit Suno’s trade secrets to gain an unfair market edge.
Counterarguments and Public Interest
Inner City Press’s reporter, Matthew Russell Lee, argued that understanding what music was used to train Suno’s AI is essential to the copyright infringement issue. He suggested that the sealed materials directly relate to the core of the case against Suno. Suno, however, vehemently disagreed with this characterization. They maintained that the specificity of the figure they want to seal is minimal and not central to the legal arguments being made. Suno’s lawyers emphasized that the plaintiffs’ claims revolve around identified works rather than the aggregate size of the training dataset.
Previous Precedents in Sealing Information
The legal landscape for AI companies is characterized by a lack of clarity when it comes to proprietary data and intellectual property. Suno pointed to several earlier cases—such as The New York Times v. Microsoft, Kadrey v. Meta Platforms, and Concord Music Group v. Anthropic—where courts have approved motions to seal similar types of information, citing competitive risks as a substantial concern.
In this instance, it is noteworthy that the plaintiffs have not opposed Suno’s request to keep this information sealed, although they reserve the right to contest its propriety later on.
Ongoing Negotiations and Broader Implications
As the case unfolds, it is essential to note that the landscape is dynamic. While Warner Music Group, once a co-plaintiff, settled with Suno and established a licensing deal, UMG and Sony remain steadfast in their pursuit of damages and injunctive relief. Their attempts to access terms from the Warner Music settlement have been blocked, leaving them at a negotiating impasse with Suno.
The original lawsuit stems from UMG and Sony’s assertion that Suno used “millions” of their copyrighted sound recordings without consent. Illustrating a complex backdrop, the case continues to evolve, with a deadline for dispositive motions set for January 8, 2027, indicating that the legal battle is far from over.
In this intricate dance of copyright law, technology, and competitive strategy, Suno’s request encapsulates larger themes of what constitutes fair play in the increasingly competitive landscape of AI-driven music generation.