Published: 10/14/2026
Source: AI Musicpreneur — “Jamendo refiles Nvidia AI-training lawsuit”
Jamendo has filed a new federal complaint alleging that Nvidia used the MTG-Jamendo music dataset to train commercial AI systems despite a noncommercial research license. The new case was filed October 12 in the Northern District of California after Jamendo voluntarily dismissed a closely related June action.
The complaint concerns Nvidia’s Fugatto and Audio Flamingo projects. Jamendo is seeking at least €17.8 million and says the dataset contains more than 55,000 tracks from about 3,500 artists under a Creative Commons Attribution-NonCommercial license.
These are allegations, not court findings. Nvidia’s reported position is that it used open-source data, while the dispute turns in part on whether the dataset’s noncommercial restrictions permitted the challenged training activity.
Why it matters: The case could help clarify a central AI music copyright question: does public access to a dataset authorize commercial model training when its license expressly limits use to noncommercial research? Musicians and model developers should watch the license analysis, not just the copyright headlines.