Voice and Likeness Law for Creator Marketing: ELVIS Act, AB 2602, NO FAKES
A plain-language map of the voice and likeness legal floor: Tennessee's ELVIS Act, California's AB 2602, and the pending federal NO FAKES Act. Not legal advice.
Voices and faces are licensable assets now, in the plain contractual sense — a creator's cloned voice can narrate a brand's ads, an avatar can front a campaign nobody filmed. That shift happened fast enough that the law is still catching up in pieces, state by state, rather than as one settled framework. If you're a creator licensing your voice or likeness out, or a brand commissioning someone else's, three developments currently form the actual floor you're standing on. None of this is legal advice — treat it as a map of where the lines are being drawn, then get counsel for the specific deal.
Tennessee's ELVIS Act: the first state law naming AI voice cloning directly
Tennessee's Ensuring Likeness, Voice, and Image Security Act — the ELVIS Act — was signed into law on March 21, 2024, and took effect July 1, 2024. It's the first state law to name AI voice cloning specifically, and it did that by updating Tennessee's existing Personal Rights Protection Act to add voice as a protected personal attribute, sitting alongside the name, photograph and likeness protections the statute already covered.
The detail that matters most for anyone building with AI voice tools: the Act applies to a clone that merely sounds like the individual. You don't need to have used the person's actual name, or a literal recording of them, for it to count as an unauthorized imitation of their voice. That closes the loophole an obvious workaround would otherwise exploit — "we never said it was them, we just made something that sounds exactly like them."
It's also unusual in reaching beyond civil liability into criminal enforcement, which most right-of-publicity statutes don't do. That combination — a specific, named carve-out for AI voice imitation, applied even without using the person's name, backed by both civil and criminal exposure — is why the ELVIS Act gets treated as the template other states are watching, not just a Tennessee curiosity.
California's AB 2602: the contract-side fix
Where the ELVIS Act targets unauthorized cloning outright, California's AB 2602 targets a quieter failure mode: consent that technically exists but shouldn't count. The law renders digital-replica provisions in a contract unenforceable unless the use is specifically described and the performer had professional representation when they agreed to it.
That's aimed squarely at the boilerplate clause — a line buried in a much larger agreement granting rights to create "digital replicas" for "any and all future uses," signed by someone without an agent or lawyer in the room to flag what they were actually agreeing to. Under AB 2602, that kind of clause doesn't hold up on its own. Two things have to both be true for a digital-replica grant to survive: the intended use has to be described specifically enough that the performer knew what they were licensing, and they need to have had professional representation at the time they signed.
The practical translation is straightforward even outside California: a broad, vague, forever grant of "digital replica rights" is a contract red flag regardless of which state's law ultimately governs it, because it's exactly the shape of clause this kind of statute exists to void.
The federal picture: NO FAKES, proposed but not law
There is no single federal voice-and-likeness right yet. The federal NO FAKES Act, which would create a national right covering voice and likeness with a notice-and-takedown mechanism, has been introduced in Congress but has not been enacted. Until that changes, if it does, the actual floor is a patchwork: state right-of-publicity laws that vary significantly in what they cover, plus whatever the contract itself specifies. That's precisely why Tennessee's and California's approaches matter so much right now rather than being background trivia — for the moment, they're the sharpest edges of the law, not one example among many equivalent ones.
What this means in practice
If you're a creator or talent licensing your voice or likeness out, the emerging standard across these laws points to the same handful of non-negotiables:
- Written consent, not a verbal agreement or an assumption based on past work together.
- A specific description of the use — which content, which platforms, roughly how long — rather than an open-ended grant covering anything the licensee might do with it later.
- A term limit. A license with no end date and no renegotiation point is the shape of agreement that ages worst, both practically and, per AB 2602's logic, legally.
- Representation or counsel reviewing it before you sign, if there's any way to arrange that — under a law like AB 2602, whether you had representation isn't just good practice, it can be the difference in whether the clause is enforceable at all.
If you're a brand or agency commissioning a voice or avatar, the safer default is to assume the strictest applicable regime rather than the most permissive one. Get consent in writing regardless of which state the talent is in, since AI-voice-specific protection is actively spreading beyond Tennessee. Match the actual use to the described use — a license scoped to "social ads" doesn't cover reusing the same cloned voice in a different product's campaign. And don't treat a voice that merely sounds like someone, without using their name, as a safe workaround; that's the specific gap the ELVIS Act was written to close.
This is also where working with a creator's existing team pays off. Versely's guide for influencer marketing agencies puts the same principle plainly: nobody should generate a creator's likeness without an explicit written license covering synthetic use, and that belongs in the contract before anything gets produced — not patched in after a client asks where the clip came from.
Building with consent as the default, not an afterthought
None of this is a reason to avoid voice cloning — it's a reason to build the consent step into the workflow instead of treating it as paperwork that happens somewhere else. Versely's AI voice cloning tools are built for cloning a voice with the speaker's own participation — recording the sample directly — which is the version of this that was never in legal question in the first place. The voice cloning glossary entry puts the underlying rule as plainly as the statutes above do: a clone of someone else's voice needs their permission, and both platform terms and the law in most jurisdictions treat it that way.
If you're building narration for your own videos, cloning your own voice sidesteps the entire question — there's no third-party consent to secure when the voice being cloned is yours. And if the goal is simply a wide range of natural-sounding narration without cloning anyone, Versely's voiceover library covers dozens of languages and named voices without touching likeness rights at all.
FAQ
Does the ELVIS Act only protect musicians?
No. It updated Tennessee's Personal Rights Protection Act — an existing right-of-publicity statute that already covered a person's name, photograph and likeness — by adding voice as a protected attribute alongside those, including AI-generated imitations. The protection isn't limited to recording artists.
Is the federal NO FAKES Act law right now?
No. It has been introduced in Congress but has not been enacted. Until that changes, there is no single federal voice-and-likeness right — only state laws like Tennessee's and California's, plus ordinary contract terms.
What makes a digital-replica clause enforceable under AB 2602?
Two things, together: the intended use has to be specifically described rather than open-ended, and the performer needs to have had professional representation when they agreed to it. A clause missing either one can be rendered unenforceable.
Does a voice clone need to use the person's actual name to be covered by the ELVIS Act?
No. The Act applies to an imitation that merely sounds like the individual, so leaving their name out of it isn't a workaround.
This is general information, not legal advice — voice and likeness law is moving fast and varies by state and country, so for an actual license or release, get counsel who covers the specific jurisdictions involved.