Grimes says she will split 50% of royalties with fans who deepfake her music

The musician says she has “no label or legal bindings.”

While major record labels have been trying to fend off songs using AI voice tracks of famous artists, Grimes has other ideas.

“I’ll split 50% royalties on any successful AI generated song that uses my voice,” Grimes posted on Twitter Sunday night. “Feel free to use my voice without penalty,” she said, claiming she has no label and “no legal bindings.”

At this point, Grimes’ Sunday night tweet seems to be just that — a late-night tweet that maybe, possibly, potentially could become something in the future. Grimes didn’t add much detail as to how arrangements like this would work but said the profit sharing could apply to “viral” or “super popular” tracks made using her voice that are already floating around.

Grimes isn’t the first artist to embrace voice cloning and artificial intelligence tools. Holly Herndon, an experimental musician, introduced her own artificial voice called Holly Plus in 2021. Herndon allows users to upload audio files and receive a new version sung in her voice. Only members of Herndon’s decentralised autonomous organisation (DAO) are able to profit from the voice model.

Voice models trained on a corpus of recordings of an artist are now more accessible than ever, yielding bizarre, hilarious, and slightly creepy results. A song generated using voice models of Drake and The Weeknd went suspiciously viral last week, only to be uniformly yanked from streaming platforms shortly after. In the midst of the song ‘Heart on My Sleeve’ going viral, Universal Music Group issued a sternly worded statement claiming that training AI models on their artists’ work was a copyright violation.

WriterChris Saunders