Voice Cloning Consent and the Law on Voice and Likeness

Consent is the whole ethics of voice cloning in one word, and most people get it half right. They ask. The person says yes. Nobody writes down what the yes covered. Then the clone ends up in a fundraising ad the person never imagined, saying a sentence she would never say, and a friendship or a reputation goes with it.

Consent that holds up

The module on AI images and video showed a short consent note. A note is fine for a one-off. For anything that will be reused, a written agreement should answer every question below. Plain language is fine. Vague language is not.

Question	Weak answer	Strong answer
Who is cloned, by whom, on what service?	"Your voice"	Marisol Vega's voice, cloned by the Harlan Valley Library Friends on her own ElevenLabs account and shared to theirs
What will it say?	"Library stuff"	Scripts for three explainers on the levy; she sees each script before it is voiced
Where will it appear?	"Online"	The group's website, podcast feed and social accounts. No paid ads, no phone calls
For how long?	Not stated	November 1, 2026 to May 31, 2027
Payment?	Not stated	$150 per finished explainer, or "volunteered, no payment," in writing
Withdrawal?	Not stated	In writing at any time; new audio stops at once; published audio comes down within 14 days
What happens to the voice model afterward?	Not stated	Deleted from the group's account at the end date or on withdrawal; she keeps her own

Notice the third row. ElevenLabs' docs say a Professional Voice Clone can only be made of your own voice, verified by reading lines into the microphone. That design is itself a consent mechanism. The clean way to use someone else's professional clone is for them to make it on their own account and share it, so they can revoke it.

Some lines this course does not cross, whatever a form says. No clones of public officials, candidates, opponents or sources. No clones of people who cannot consent, including the dead, unless an estate grants it for a specific, labeled purpose. No "sounds like" prompts aimed at a real person. And no synthetic voice on a phone dialer without reading the robocall section below.

The law as of October 2026

There is no single U.S. law on voice. There is a patchwork of state publicity rights, a few new AI-specific laws, a federal agency ruling on phone calls and a federal bill still waiting for a vote. The U.S. Copyright Office's July 2024 report on digital replicas found gaps in existing protection and recommended that Congress pass a federal law.

Law	Where	Status	What it does
ELVIS Act	Tennessee	Signed March 21, 2024; effective July 1, 2024	Adds voice, actual or simulated, to the state's publicity right; covers tools whose primary purpose is unauthorized replicas
AB 2602 and AB 1836	California	Signed 2024, in effect	Contract terms for digital replicas must be specific, with counsel or a union; replicas of deceased performers need estate consent
Synthetic performer ad law	New York	Effective June 9, 2026	Conspicuous disclosure when an advertiser knows a synthetic performer is used; audio-only ads excluded; $1,000 then $5,000 penalties
Synthetic performer ad law	California	Enacted September 2026; effective January 1, 2027	Disclosure when an ad prominently features a synthetic performer, audio-only ads included
NO FAKES Act of 2026	Federal	S. 4591 on the Senate calendar since June 24, 2026; H.R. 8915 in House committee	Would create a federal right over digital replicas of voice and likeness
FCC Declaratory Ruling 24-17	Federal	Adopted February 2, 2024; in effect	AI voices are "artificial" under the TCPA, so robocalls using them need prior consent
Idaho Code 67-6628A	Idaho	Added 2024; in effect	A candidate deceptively shown through synthetic media in an electioneering communication can sue; disclosure is a defense

Tennessee's ELVIS Act

Governor Bill Lee signed the Ensuring Likeness Voice and Image Security Act on March 21, 2024, and state leaders called it the first law of its kind to protect musicians from AI-generated synthetic media. It defines voice as a sound readily identifiable and attributable to a particular person, whether the actual voice or a simulation. According to a Davis Wright Tremaine summary, it reaches three kinds of conduct: unauthorized commercial use of a voice or likeness; publishing or distributing one knowing it was unauthorized; and distributing a tool whose "primary purpose or function" is producing someone's voice or likeness without permission. Exceptions cover news, public affairs, commentary, criticism, satire, parody and incidental uses.

That third clause is the unusual one. It reaches the people who build tools as well as the people who use them.

California and New York

California's AB 2602 makes contract clauses allowing a digital replica of a performer unenforceable unless they describe the uses with reasonable specificity and the performer had counsel or union representation. AB 1836 requires consent from the estate for digital replicas of a deceased personality. Both were signed in 2024.

The newer laws target ads. New York's took effect June 9, 2026: an advertiser with actual knowledge that a synthetic performer is used must disclose it conspicuously. It excludes clearly non-human figures, replicas of identifiable people (which fall under other law) and, per a Mondaq comparison, audio-only ads. California enacted a similar law in September 2026, effective January 1, 2027, with no "actual knowledge" qualifier and explicit coverage of audio-only ads; suggested wording is close to "this performance features a synthetic performer." For a radio spot with a synthetic narrator aimed at California listeners, that matters.

The NO FAKES Act, and the argument over it

The NO FAKES Act of 2026, S. 4591, sponsored by Sen. Chris Coons with Sens. Marsha Blackburn, Thom Tillis and others, was introduced May 20, 2026. The Senate Judiciary Committee approved it unanimously on June 18, and it was placed on the Senate calendar on June 24. The House companion, H.R. 8915, was referred to the House Judiciary Committee in May. As of October 2026, neither chamber had voted on it.

Supporters, described in coverage of the committee vote as a cross-industry coalition, say it gives everyone, not only celebrities, a federal right against unauthorized replicas, with liability for platforms that knowingly host them. Blackburn's office says the revised bill adds a counter-notice procedure, exemptions for libraries and researchers, and First Amendment exclusions. It would also preempt future state laws on digital replicas.

The Electronic Frontier Foundation, writing about the 2025 version, called it a disaster for internet speech. Its objections: a takedown system that would push platforms toward filters that over-remove lawful content, subpoenas to unmask users without a judge's review, and compliance costs that favor big companies. Whether the 2026 text answers those objections is exactly what to read for if it reaches the floor.

Robocalls: the FCC's ruling

On February 2, 2024, the FCC unanimously confirmed that the Telephone Consumer Protection Act's limits on "artificial or prerecorded voice" calls cover AI voices, including clones. So a call using one needs the called party's prior express consent unless an emergency purpose or exemption applies. The ruling followed a January 2024 incident in which New Hampshire voters received calls with an AI imitation of President Biden's voice telling them not to vote in the primary. The FCC fined the consultant behind the calls $6 million in September 2024; the company that carried them agreed to a $1 million settlement. A New Hampshire jury acquitted the consultant of state voter suppression and impersonation charges in June 2025. The criminal case failed. The federal phone rules did not change.

Idaho's election deepfake law

Idaho Code 67-6628A, the Freedom From AI-Rigged (FAIR) Elections Act, lets a candidate whose speech or conduct is deceptively represented through synthetic media in an electioneering communication seek an injunction and damages from whoever created it. The module on AI images and video covers it in detail. For audio, the disclosure defense is the part to know: "This (video/audio) has been manipulated," spoken clearly at the beginning and end, and at intervals of no more than two minutes in audio over two minutes long.