UMG and Sony Move to Prevent Suno From Concealing Audio Files Used for AI Training
The two major labels filed a memorandum of law to reveal the total number of files Suno used to train its models
Universal Music Group (UMG) and Sony Music Entertainment have filed a new memorandum of law, urging the court to make public the complete list of audio files Suno used to train its generative models. This is the latest move in the copyright lawsuit that the labels brought against the AI music company back in 2024.
The details:
The total number of audio recordings Suno used to train its models is referred to as the “Model Training Figure.”
In the memorandum, the major labels’ legal team cited precedent stating that since the Model Training Figure is part of their proposed amended complaint, such information is where “the presumption of public access is at its strongest.”
Suno has argued “serious commercial harm” as a reason to keep the figure under wraps. This claim relates to the idea of competitors copying its business model.
Suno has already admitted to “showing the program tens of millions of instances of different kinds of recordings.”
Background:
UMG and Sony redacted the number from their court documents, but they maintained their right to contest concealing the number entirely.
The labels recently moved to add 61,026 recordings to the suit after using the content recognition service Audible Magic to scan Suno’s training data for their copyrighted works.
👋 Disclosures & Transparency Block
This article was written with information sourced from Music Business Worldwide.
We covered it because of the ongoing conversation around AI in the music industry.
