Labels Accuse AI Music Firm of Laundering Copyright Infringement in New Lawsuit
Sony Music Entertainment and Universal Music Group have filed a new 45-page lawsuit against AI music company Suno, alleging its v6 model was trained on outputs from earlier infringing models. The labels argue this does not eliminate infringement but merely launders it, calling the new model "fruit of the same poisoned tree." Suno has not yet commented on the suit.
The lawsuit, filed in Massachusetts federal court, names Sony Music, Universal Music Group, and several of their imprints as plaintiffs. It targets Suno's v6 model, which the labels say was trained on outputs from earlier models that were already accused of infringement. Suno had previously claimed its newer models avoided using copyrighted material directly, but the labels argue this distinction is meaningless if the training data itself was derived from infringing sources. The case builds on an earlier legal action between the same parties, and Suno has not yet issued a public response.
The labels' argument uses the metaphor of a "poisoned tree" to suggest that any model built on prior infringing outputs inherits the original legal problem. This approach could set a precedent for how courts treat iterative AI training, where successive versions of a model are refined using data from earlier versions. The outcome may hinge on whether judges accept that indirect training on copyrighted material constitutes infringement, even when the final model's outputs do not directly copy existing songs.
This lawsuit could shape how AI music companies develop future models, as they may face legal risk even when using their own prior outputs. If the labels' argument gains traction, it may force AI firms to prove the provenance of every layer of training data, potentially slowing innovation. Artists and rights holders could feel empowered to challenge AI systems that mimic their work indirectly, while smaller AI startups might struggle with added compliance costs. The case may also influence broader debates on copyright in generative AI across creative industries.