Sony and Universal Music Group have filed a second lawsuit against Suno, alleging that its latest v6 model continues to violate copyright laws. The record labels argue the system remains trained on user outputs generated by earlier versions, which themselves relied on unlicensed audio scraped from YouTube and other sources. This legal action targets two major industry partners who refused to sign a licensing deal with the startup. In the complaint shared with The Verge, Sony and UMG describe the practice as model laundering. They claim training a new model on infringing outputs does not erase the violation but rather passes the value of the original expression to the updated system.
This dispute highlights a specific legal vulnerability for generative audio tools that rely on iterative training. While Suno previously settled with some partners, these two giants maintain that the core training data remains problematic regardless of model version updates. The case sets a precedent for how future developers must handle data provenance when building successive iterations of their systems.
* The suit specifically cites the v6 model’s reliance on prior user outputs.
* Sony and UMG remain the primary holdouts on licensing agreements with Suno.
* The labels argue the new model simply washes the original infringement rather than curing it.
