Labels Sue Suno for "Model Laundering" in AI Music v6

The music giants filed a fresh federal lawsuit asserting that Suno's newly rolled-out v6 music engine was trained directly on the outputs of its previous copyright-infringing models.

According to the joint complaint, attempting to sanitize algorithmic infringement by feeding generated audio into a fresh training run is nothing more than legal money laundering for weights and biases. The record companies labeled the system the direct fruit of a poisonous tree, arguing that copyrighted catalog tracks were originally scraped into earlier iterations, baked into those systems' song outputs, and then recycled straight back into v6 under the guise of an all-new dataset.

During the promotional push for v6, Suno representative Jack Brody announced that the network was trained entirely from scratch on a brand-new dataset featuring user creations. Company spokesperson Rachel Racusen subsequently confirmed that user tracks were included in the training pipeline, though she pointedly refused to clarify whether those audio files were original files uploaded by humans or previous synthesized tracks spat out by Suno itself.

Neither Sony nor UMG signed off on licensing arrangements with the generative platform, continuing their offensive originally initiated in 2024. Sony simultaneously pursued legal action against competitor Udio, seeking massive statutory damages over more than 30,000 catalog recordings by artists like Elvis Presley, Beyoncé, Jefferson Airplane, Kings of Leon, ZZ Top, and Harry Styles.

The tech industry might have stumbled into a digital laundering operation no traditional copyright statute was ever drafted to handle, or legacy record executives simply cannot stomach the reality that algorithmic mimicry refuses to show up on their quarterly royalty statements.

Source: The Verge