Universal and Sony Sue AI Music Generator Suno Over New V6 Model, Calling It 'Fruit of the Same Poisoned Tree'

Universal Music Group and Sony Music Entertainment have filed a second, 45-page copyright infringement lawsuit against AI music generator Suno, targeting its newly launched v6 model. The labels argue that despite Suno securing new licensing deals, v6 was trained on user interactions and outputs generated by older models that illegally ingested 60,202 copyrighted recordings, making the new system "the fruit of the same poisoned tree." The filing, submitted in Boston federal court on September 18, 2026, marks a direct challenge to the AI industry's attempt to pivot toward legitimacy by simply signing new checks.[1][2]

The lawsuit arrives as a new chapter in an ongoing legal saga, landing just nine days after Suno launched v6. At the time of the launch, Suno touted partnerships with Warner Music Group, BMG, and Believe, presenting the update as a clean break from its controversial origins. The company claimed the model was built from scratch using licensed music, user data, and accumulated learnings.[1][3]

The core of the labels' new argument centers entirely on that "user data." UMG and Sony allege that the user interactions Suno relied upon are actually preference signals and synthetic audio outputs derived from prior models. Because those earlier models were trained on unlicensed catalogs, the labels argue the resulting user data carries the exact same copyright taint.[1][2]

"Training a 'new' model on the outputs of an infringing model does not eliminate the infringement; it launders it," the joint complaint states. The filing argues that the value of the original copied recordings was passed into the tainted models, from those models into their outputs, and ultimately into v6.[1]

The financial stakes attached to this legal theory are existential for the generative AI sector. During discovery in the initial lawsuit, the labels utilized audio fingerprinting technology from Audible Magic to pinpoint 60,202 specific sound recordings inside Suno's training data. Under US copyright law, willful infringement carries statutory damages of up to $150,000 per work, setting a theoretical maximum liability of over $9 billion.[1][2]

Additionally, the new complaint introduces a stream-ripping claim, accusing Suno of bypassing YouTube's anti-downloading protections to acquire its initial training data. The labels are seeking up to $2,500 for each act of circumvention, which could add another $150 million to the total damages if applied to all 60,202 tracks.[1][2]

Suno maintains that its practices fall squarely under the fair use doctrine, arguing that learning from existing music to build a new back-end product is not infringement. In a statement responding to the suit, a Suno spokesperson called the labels' claims "fundamentally flawed on both the facts and the law," reiterating that v6 was trained on licensed content and community preference signals to help users make new music.[2]

The lawsuit highlights a growing and permanent fracture among major rights holders regarding AI strategy. Warner Music Group, initially a plaintiff alongside UMG and Sony in the earlier litigation, settled its claims and became a licensing partner for v6. UMG and Sony, however, are now using those very deals against Suno, arguing that the agreements prove a functioning, commercial market exists for AI training data—directly undermining the company's fair use defense.[1][3]

The case now proceeds in the Massachusetts federal docket, running parallel to the original lawsuit. The outcome will likely hinge on the "knowledge distillation" theory—whether courts accept that iterative software improvements carry forward infringing knowledge. If the judges agree with the labels, AI companies across all mediums may be forced to prove their new models have fully unlearned past data, regardless of how many new licenses they sign.[1][2]