In our recent report on Suno’s new v6 models, we explained that they were trained on a fresh dataset including music licensed from WMG and BMG, plus “user data”.
Now Universal Music Group and Sony Music have trained their attention on the latter, filing a new, second copyright-infringement complaint against Suno. You can read the full filing here.
The new complaint cites 60,202 sound recordings that the labels believe Suno has copied in its past training process, adding that “in reality they remain only a small portion of Plaintiff’s works that Suno infringed”. But the two majors are also squarely focused on its latest models.
“Suno has admitted that it trained v6 using ‘user interactions’ with previous iterations of its models. But these ‘interactions’ are the outputs of, and preference data derived from, Suno’s prior models,” claims the filing.
It also suggests that v6 was trained on “synthetic outputs” of the “prior tainted models”, building this into an argument that “Suno trained v6 with material that exists only because Suno copied Plaintiffs’ recordings without authorization”.
“v6 is not a fresh start; it is the fruit of the same poisoned tree,” is the key line. “Suno’s development of v6 continued, rather than ended, its unauthorized exploitation of Plaintiffs’ Copyrighted Recordings.”
In our recent interview with Suno’s chief product officer Jack Brody, he said that “I think of this moment as really marking a new chapter for Suno”. But in the legal context, UMG and Sony Music are determined to prove that it’s just another chapter in the same copyright-infringement playbook that spurred them to sue in the first place.
Suno will doubtless respond to the new complaint with its own counter-filing, and we’ll bring you details of that – as well as any response statement in the meantime – as we get them.