SZA - one of the biggest voices in contemporary R&B - searched a database used to train AI music models and found 238 of her own songs inside it. Her reaction wasn't legal, it was personal: she came out publicly against the entire AI music industry and called on artists not to hand over their own work to the machines that will later imitate them.
Fact
The facts: 238 of her songs were in the training set. SZA called the practice exploitation of artists. Her label Sony is already suing the AI music generator Suno - one of the cases that will determine whether training on copyrighted music is legal.
What matters here is who's raising it, not the lawsuit itself. When an engineer says 'the model was trained on public data,' the world shrugs. When a voice of this caliber says 'these are my songs and nobody asked me,' the subject stops being technical and becomes personal. I've stood behind stages for twenty years while a crowd danced to a voice it knew - and I know that voice, and how personal the feeling it carries really is. The machine takes the timbre; the right to give it away stays outside its reach. And as long as that right is defended only in court, instead of being built into the tool itself, everyone who sings, plays, or paints - including you - watches the case as their own.
Sources
CourtListener - federal court record: UMG Recordings, Inc. et al. (incl. Sony Music Entertainment) v. Suno, Inc., No. 1:24-cv-11611 (D. Mass., filed 24.06.2024)Rolling Stone - SZA on AI musicThe visual is generated code-art, no third-party images.