Grimes lets anyone make songs with her AI voice

While the music industry is busy sending cease-and-desist letters to AI clones, Grimes is taking the exact opposite route: she is giving her voice away to anyone who wants it.

This week, the experimental pop artist launched Elf.Tech, a new AI voice software that lets anyone create songs using an automated replica of her vocal chords. The concept is straightforward. You upload a recording of your own singing, the system Grimes-ifies the track, and you can mix the vocals into beats to produce a song that sounds remarkably authentic.

Half the royalties, all the clout

Of course, there is a catch, but it is surprisingly generous.

Grimes announced on Twitter that she is totally fine with creators releasing these songs commercially, provided she gets a 50% cut of master recording royalties. In exchange, artists get an official Grimes feature credit and access to distribution channels.

She noted that there is a small chance they can organize publishing royalties as well, though that part is not guaranteed yet.

Her manager, Daouda Leonard, told Rolling Stone that Grimes views creativity as an ongoing conversation rather than something to gatekeep. Instead of trying to control what she sees as a gift from the universe, she chose to open-source her voice to see what new experiences emerge.

Now that the tool is live, the internet is already seeing a wave of newly minted, AI-powered Grimes tracks.

Why it matters

Grimes just handed the music business a blueprint for how to monetize AI instead of fighting a losing legal war.

For bedroom producers and indie musicians, Elf.Tech turns an unreachable A-list collaboration into a self-serve plugin. You get an iconic, ethereal vocal identity on your track without paying upfront studio fees, split down the middle if the track blows up.

We think this is one of the smartest cultural moves an artist has made around generative tech. Deepfakes are happening whether labels like it or not, so offering the official, highest-quality model and capturing 50% of the upside is infinitely smarter than playing whack-a-mole with copyright takedowns.

The real test will be brand dilution. When thousands of algorithmic tracks hit streaming platforms under your name, the quality control vanishes overnight. But as a business model, it flips copyright panic into a scalable franchise.