Sony and UMG are suing Suno again

AI music generators that incorporate user-generated content may face copyright challenges from rights holders.

Sony and Universal Music Group (UMG) have filed a new lawsuit against AI music generator Suno. The music giants allege that Suno's latest model, v6, still infringes on their copyrights. Their core argument is that v6 was trained on outputs generated by previous Suno models, which themselves were trained on unlicensed music. Sony and UMG claim this process is a form of "model laundering," where infringing content is passed through new models. They assert that even if v6 was trained on new data, it benefits from the unauthorized copies and infringing data used in earlier versions. Suno has previously stated that v6 was trained from scratch with a new dataset, including user data, but has not provided extensive details. Sony and UMG are among the few major labels that have not reached a licensing agreement with Suno.

The music industry's legal challenge against Suno indicates a broader battle over AI training data and output ownership. By pursuing Suno again with the "model laundering" argument, Sony and UMG are establishing a precedent that subsequent AI models are liable if trained on outputs derived from infringing prior models, even if directly trained on new datasets. This approach could significantly complicate the development of generative AI that relies on user-generated content or indirectly benefits from previously acquired data.

This lawsuit suggests that AI companies, particularly in creative fields, must rigorously audit their entire training pipeline, not just the final dataset. The music labels' stance implies that any reliance on outputs from potentially infringing earlier models, or any benefit derived from them, constitutes continued infringement. This could force AI developers to implement stricter data provenance tracking and develop entirely new methods for training models that can demonstrably prove a lack of lineage to any unlicensed material.