Sony and Universal Music Group have filed a new copyright suit against Suno, the developer of AI-generated vocalists. The complaint targets Suno’s most recent release, the v6 model, which the plaintiffs say continues to rely on music that was ripped from YouTube and other platforms without permission. Both labels argue that the model’s outputs still infringe on their copyrighted recordings, despite Suno’s claim of a fresh training regime.
Suno’s co-founder Jack Brody told The Verge that v6 was “trained from the ground up, with a new set of data,” adding that the process incorporated “user data.” A Suno spokesperson, Rachel Racusen, later clarified to the same outlet that the training set included “creations from users, preference signals from our community, and licensed content from our partners,” while stopping short of confirming whether uploaded audio files were part of the corpus.
Sony’s filing further alleges that Suno employed a technique known as distillation to build v6, effectively copying the behavior of earlier “teacher” models that were themselves trained on unlicensed recordings. The plaintiffs contend that this approach allows the new system to reproduce the same infringing results without directly ingesting the original tracks, thereby extending the reach of the alleged copyright violations.
The complaint states that even a model not directly trained on the plaintiffs’ recordings is “informed by, and benefits from, Suno’s retained unauthorized copies.” This language underscores the labels’ position that any derivative learning, regardless of direct exposure, still draws on the underlying infringing dataset, creating a legal gray area around the use of aggregated user-generated content in AI training pipelines.
Both Sony and Universal have refused to sign a licensing deal with Suno, a stance that fuels the lawsuit’s demand for the company to rebuild its technology from scratch. The plaintiffs argue that without a clean-room data set, any future AI vocalist will inevitably incorporate protected works, perpetuating the alleged infringement. The case adds to a growing wave of music-industry actions targeting generative-AI platforms that rely on large, often unverified, data collections.