Universal Music Group and Sony Music Entertainment filed a second federal complaint against Suno in Boston, alleging unauthorized copying of 60,202 sound recordings to build its generative models. The labels argue that Suno’s v6 models, described by Suno as trained from scratch with Warner, BMG and Believe partners, are still “the fruit of the same poisoned tree,” because they were built using outputs and preference data from earlier models allegedly trained on unlicensed recordings.
The Suno fight extends well beyond streaming. Generative music is increasingly relevant to games, where AI could enable infinite soundtracks.
That makes the provenance of the models powering those experiences a gaming issue too. If courts determine that models trained on copyrighted recordings require licenses or that outputs derived from those models carry infringement risk: game developers and platforms integrating generative music could face an entirely new layer of rights complexity.
The opportunity is enormous: but before that becomes infrastructure, the industry needs clarity on what these models can legally learn from and what developers can safely build with them.
— Miles Coda
Editor-in-Chief, Music & Gaming