Grimes offered her voice as a licensed model with a royalty split

Elf.tech's own 2023 terms set an equal royalty share and an approval step for releasing music with Grimes's AI voiceprint.

Historical event

April 1, 2023

Site publication

September 18, 2026

What happened

In April 2023, the musician Grimes announced that producers could use an AI model of her voice, distributed through a web platform called elf.tech, to record new vocals in her voiceprint. The offer was unusual for the period: rather than pursuing takedowns against unauthorised clones, as other artists did that year, Grimes published terms letting anyone use the model in exchange for an agreed royalty share. The platform has since changed; as retrieved on 16 September 2026, it displays only the words 'I am evolving', with none of the 2023 terms visible on the live site.

What the documents say

The 2023 terms survive in an archived capture of elf.tech's own application code, taken by the Wayback Machine on 22 May 2023. That text states registered creators could upload vocals to be 'transformed into a GrimesAI-1 voiceprint', that 'creators will own these new recordings' and that GrimesAI-1 'does not claim any ownership of the sound recording or the underlying composition' unless it covers an existing Grimes song. It sets the royalty term precisely: creators 'agree to share an equal % of the master recording royalties' across streaming and on-chain sales with GrimesAI-1, and must email a named address to request approval before releasing on any platform other than elf.tech itself, identifying the company CreateSafe, Inc. as the counterparty. An archived capture of the connect page from 20 April 2023 shows the service was wallet-gated and built around the crypto platform Triniti, matching the on-chain framing of the royalty terms.

Why it matters for makers

The equal-share structure names a mechanism absent from most voice-licensing offers: an automatic, disclosed revenue share tied to distribution, rather than a flat fee or a silent grant. A maker who used GrimesAI-1 kept ownership of their new recording outright, but took on an ongoing royalty obligation and an approval step for anything released off-platform — two separate conditions a casual reading of 'she let anyone use her voice' would miss entirely.

What to check before you use it

Because the terms above exist only in an archived capture of application code rather than a current, live page, a maker cannot confirm whether similar terms still apply to any successor service. Before assuming a comparable deal is available elsewhere, check whether the offering platform is still operating, request current written terms rather than relying on 2023 reporting, and confirm who the counterparty and rights-holder actually are, since the entity named here was a company, not the artist personally.

• Is the voice model's operator still the same legal entity as when the terms were published?
• Does the royalty obligation apply to streaming only, or also to sync and on-chain resale?
• Is approval required before release on any platform other than the one hosting the model?

Elf.tech's current one-line placeholder is itself a data point: a 2023 licensing structure, however clearly documented at the time, is not a permanent guarantee, and a maker should treat any archived terms as a historical record to verify afresh rather than a standing offer.

Sources & reading trail

Archived application code carrying elf.tech's own creator FAQ text: ownership, royalty split and approval terms.

Source published: Not established · Retrieved: 16 September 2026

Archived application code showing the wallet-gated, Triniti-powered access flow around the voice model in April 2023.

Source published: Not established · Retrieved: 16 September 2026

The current live site, confirming the 2023 terms are no longer published and the platform now shows only a placeholder.

Source published: Not established · Retrieved: 16 September 2026

Papers, terms and official documents establish the record; the maker reading and the checks are Signal to Song editorial analysis. This retrospective draft does not imply the site published on the event date.

Continue reading

• Holly Herndon released her own voice as a governed instrument
• An open-source voice cloner ships with no consent gate at all
• A voice clone’s consent has to satisfy a statute and a platform
• Browse the complete the archive

Published September 18, 2026, not on the date of the event described.