Sony Music Group and Universal Music Group have filed a new copyright infringement lawsuit against AI music generator Suno, targeting its latest v6 model. The labels allege that Suno engaged in model laundering by training v6 on user outputs and synthetic data generated by earlier model versions, which had been originally trained on copyrighted sound recordings without authorization.
Suno confirmed that v6 utilized community interaction data and creations alongside licensed partner content, though it did not detail specific dataset compositions. The record labels argue that using synthetic data or model distillation from an infringing parent model perpetuates underlying copyright violations, referring to the resulting architecture as fruit of a poisoned tree.
The legal battle comes as major record labels split strategies between litigation and licensing deals with generative music platforms. The suit raises critical legal questions regarding model distillation, synthetic data training, and whether subsequent model generations can clean initial copyright liabilities.
Why it matters
Establishes a major legal test around model laundering and whether synthetic data derived from copyrighted sources violates copyright law.
Increases intellectual property litigation risks for generative media startups using model distillation or synthetic output for retraining.
Source: theverge.com



