Suno asks a federal court to seal the number of audio files it trained on
Suno wants one number kept out of public view: how many audio files it used to train its AI music model. On Friday, May 29, the company asked the US District Court in Boston to impound that figure inside the copyright case brought by Universal Music and Sony Music.
The filing is narrow on purpose. Suno told the court it wants to seal a single piece of information across the labels’ recent papers: the total volume of audio files the plaintiffs say it copied to train its generative model.
The one figure Suno wants kept quiet
Suno’s lawyers were clear about the limits of the request. The company is not trying to hide the identity of any specific recording the labels claim sat in its training data. It is also not sealing the 61,026 copyrighted works that UMG and Sony are moving to add to the complaint.
The argument rests on a declaration from Georg Kucsko, Suno’s Chief Technology Officer. He stated the number is not publicly available and has not been disclosed outside the litigation. His worry is competitive: if rivals learn the size of the training set, they could benchmark their own systems against Suno’s.
A journalist is pushing the other way
The request did not go unopposed. New York-based Inner City Press, run by journalist Matthew Lee, filed to stop the impoundment and force the number into the open.
Lee told the court these are not minor discovery scraps. They are core pleadings tied to the labels’ motion to amend the complaint, and the figure speaks to a public question about how artists’ recordings were used.
These are not peripheral discovery documents. They are the core pleadings and supporting submissions for Plaintiffs' motion to amend the complaint.
Suno’s reply is that the point is now moot. The court told Suno to make a good-cause showing for sealing, and Suno says it has done that.
Why the count carries weight
The size of a training set is not a footnote in an AI copyright case. It is the scale of the alleged copying, stated as one clean number.
UMG and Sony already say discovery showed Suno trained on millions of their recordings. Warner took a different road and settled with Suno in November 2025. Sony and Universal are still in Boston, and a loss on fair use could push Suno onto a fully licensed model.
This is the second sealing fight in the case. Suno has also resisted handing over the terms of its Warner license, and the labels have pushed back there too. The pattern is a company trying to keep the mechanics of its model, and its deals, behind the curtain. It echoes the wider fight over undisclosed training data across the industry.
Frequently asked questions
What is Suno trying to keep secret in the UMG and Sony lawsuit?
Suno asked the court to impound one figure, the total number of audio files used to train its AI model. It is not sealing the names of specific recordings or the 61,026 works the labels want to add to the case.
Why does Suno want the training data count sealed?
In a court declaration, Suno CTO Georg Kucsko argued the number is not publicly available and that competitors could use it to benchmark their own AI systems against Suno's, causing competitive harm.
Who is fighting to make the number public?
Inner City Press, run by journalist Matthew Lee, filed an opposition urging the court to deny the impoundment. Lee argues the figure is a matter of direct public concern, especially for the artists whose recordings are at issue.
What is the status of the Suno lawsuit?
Warner Music settled with Suno in November 2025. Sony Music and Universal Music are still litigating in the US District Court in Boston and are moving to expand the case to 61,026 recordings.
About the author
Covering AI music tools, industry shifts, and news for music creators and professionals. Twice-weekly newsletter at aimusicpreneur.com.
