Licensing your voice to AI platforms: consent, payouts and how to take it back

Licensing your voice to an AI platform means letting the company host a model of your voice that other people can use, usually in exchange for a share of what they pay. Whether that is a good deal depends almost entirely on three things in the platform's own terms: who says yes to each use, how the money is worked out, and what happens when you want out.

This guide reads those terms for two platforms that publish them, ElevenLabs' Voice Library and Kits.ai's artist voice models, as we read them on 24 Oct 2026. It is not legal advice. If you want the defensive side, how to stop people cloning you without permission, read our guide to protecting your voice from AI cloning first. This one is about doing it on purpose.

Two models of consent

The two platforms make very different choices about when you, the voice owner, get a say.

Consent once, up front (ElevenLabs). You build a Professional Voice Clone, share it to the Voice Library, and from then on other users can generate audio with it. The Voice Library Addendum says that by sharing you "authorize us to make available the User Voice Models for others to access and use". You do not approve individual uses. Your protection is ElevenLabs' general Prohibited Use Policy and its moderation tools, which the Addendum itself says are not guaranteed to stop every misuse.

Consent per commercial use (Kits.ai). Kits' Artist AI Model License Terms give any subscriber a personal, non-commercial licence to what they make with an artist's model. To use an output commercially, the user has to register it on the artist's model page, and the artist "may approve or deny" the request in their "sole discretion". If the artist says no, the user keeps only the personal licence.

Getting in: verification and eligibility

ElevenLabs' Addendum sets the entry rules: the model must be based on your own voice and made with ElevenLabs' tools, you must have the rights to create and share it, and you must consent to the processing of your voice data under its privacy policy. Payouts require a Professional Voice Clone that has been reviewed and shared in the library.

The Voice Actor Payouts page says the feature is available on the Creator plan and above, since that is where Professional Voice Clones live, while the payouts documentation says a paid subscription is not required to receive the rewards themselves. Read both before you budget for it.

The money: how splits actually work

Neither platform publishes a single percentage in the documents we read, so we are not quoting one.

ElevenLabs pays on usage. Its payouts documentation says earnings "are calculated based on usage by paid users and exclude use by free users", and the Addendum repeats that "in no case will Rewards be provided" for use by a free user. Your default rate is set by the notice period you choose, but the page does not list the rates. Payouts run through Stripe Connect, roughly every 6 to 8 days once you pass a minimum that is $10 in most countries. For non-US earners, the US applies a default 30% withholding tax on earnings from US customers' usage, which a tax treaty may reduce if you file the right form.

Kits.ai works the other way round. Under section 1.5 of its licence, the user who commercialises an approved output agrees to pay the artist a percentage of "Gross Revenues", with the percentage "set forth on the Artist AI Model Page". Gross revenues are defined widely: streaming royalties, publishing and performance income, sales, and sync and master-use fees for the output or any larger work that includes it. The user must add the artist's publishing information when they distribute, and the royalty is due within 30 days after the end of each calendar quarter. Crucially, the licence says Kits' parent Arpeggi "merely facilitates" the model, that it is the user's "sole responsibility to remit the Royalties", and that the artist enforces the terms against the user directly.

That is a real split of the song's income, which is good, but collecting it is your problem. A per-use payout from the platform is smaller per use but arrives without you chasing anyone.

Taking it back: revocation and what survives

This is where most of the risk sits.

On ElevenLabs you can remove a shared voice "at any time", but anyone who saved it keeps access for the notice period you set, which runs from immediate removal to two years. The Addendum says that once you share, "you will not be able to reduce your Notice Period". It also says outputs generated before the notice period ends "will continue to exist", and termination does not undo rights licensed up to that point. ElevenLabs can also remove a voice at its own discretion. In plain terms: you can stop new use, but you cannot recall audio already made.

On Kits, the output licences are described as irrevocable except under section 4.1, which lets the artist terminate all of a user's licences if they materially breach the terms. Once ended, the user must stop using the rights granted. The royalty clause survives termination. There is also a clause users should notice: section 1.6 has every user promise not to sue the artist or other users over outputs that sound similar or identical to theirs.

A checklist before you share

1. Pick the notice period knowing you cannot shorten it. A longer one may pay a better default rate on ElevenLabs, but it is also how long your voice stays out there after you change your mind.
2. Assume every output is permanent. Neither set of terms lets you recall audio already generated.
3. Know who pays you. Platform-paid usage (ElevenLabs) or user-paid royalties (Kits) are different businesses with different admin.
4. Check your other contracts. A label, publisher or management deal may already cover your name, likeness or voice. The ElevenLabs Addendum makes you promise you hold those rights.
5. Register the publishing details correctly if you take the Kits route, since the licence relies on users adding them when they distribute. Our guide to metadata and split sheets covers the basics.
6. Re-read the terms on the day you sign. Both platforms can change their terms, and the rates are not in the documents we read.

What we think

Licensing your voice can be a fair deal, but only one of these models treats you as an artist rather than a supplier. Per-use approval, as on Kits, is slower and needs you to answer requests, but it is the only arrangement here where your voice cannot end up on a record you would never have agreed to. Blanket consent with a moderation backstop is fine for narration and voiceover; for a singer whose voice is their brand, we think it is too much to hand over.

The detail that worries us most is shared by both: revocation stops the future, not the past. Platforms should publish their rates next to the notice periods, let artists shorten a notice period at least once, and give artists a list of what was generated with their voice. Until they do, treat your notice period as the real price of the deal, and set it short.

FAQ

Can I stop an AI platform using my voice after I share it? On ElevenLabs you can remove it at any time, but people who saved it keep access for your notice period, and earlier outputs remain.

Do I get paid when free users use my voice on ElevenLabs? No. Its terms say rewards come only from paid users.

Who pays the royalty on Kits.ai? The user who commercialises the approved output pays the artist directly; the platform says it only facilitates the model.

Is there a standard revenue split? Not in the terms we read. ElevenLabs ties rates to the notice period without listing them, and Kits sets a percentage per artist on each model page.