ElevenLabs requires authorization before it clones a voice

The writer's problem

An author considering a synthetic narrator built from a specific person's voice, their own, a narrator's, or hypothetically someone else's, needs to know what a voice-cloning platform actually requires before it will build that model. ElevenLabs' own Terms of Service and Prohibited Use Policy, last updated 31 March 2026 and 17 August 2026 respectively, set out that requirement directly.

What the documents show

The Terms of Service state that creating a User Voice Model requires uploading recordings of your voice or a voice you are authorized to share with us, making authorization from the voice's owner a condition of the feature rather than an afterthought; the same clause grants ElevenLabs a license to use the resulting model to operate and improve its services, while stating the company will not commercialize your voice on a standalone basis without your permission to do so. The Prohibited Use Policy is more specific about misuse: it bars unauthorized, deceptive or harmful impersonation, defined to include creating audio that replicates the voice of another person... without consent or legal right, including to take unauthorized action on behalf of such individual, in a way that harasses or causes harm, or in a manner intended to deceive others about whether the voice was generated by artificial intelligence. The same policy also lists evading voice verification mechanisms such as Voice CAPTCHA as a separate prohibited act, which indicates the platform runs an automated check intended to confirm a speaker's presence or consent at the point a clone is created, though neither document details exactly how that check operates.

The editorial choice

This is an editorial safeguard an author should treat as a floor, not a ceiling: the platform's own rule against cloning a voice without consent or legal right describes a minimum condition for using the product lawfully under its terms, not a guarantee that a hired narrator's separate contract with the author addresses ownership, ongoing use, or compensation for a synthetic version of their voice.

What stays with the author

ElevenLabs' documents describe what its own platform requires and prohibits; they do not describe what a narrator's recording contract should say about a synthetic voice built from their performance. Negotiating that separately with the narrator, rather than relying on the platform's use policy to cover it, remains the author's or publisher's task.

• Has the person whose voice would be cloned given consent specifically for that use, beyond simply being audible in an uploaded recording?
• Does the narrator's contract address a synthetic version of their voice, or does it predate the possibility?
• Would using a cloned voice for narration require disclosing synthetic narration to the retailer or the listener?

A platform that names its own consent requirement in writing gives an author something concrete to point to; it does not by itself settle what the author owes the person whose voice is being used.