Synthetic Voice Rights in the USA: What the Law Protects
There is no single federal law in the United States that says who owns a voice, which is why synthetic voice rights feel like a patchwork. Today a person’s voice is protected mainly through state right-of-publicity and privacy law, contract terms, consumer protection rules and, for robocalls, federal telecom enforcement. A federal bill, the NO FAKES Act, would add a national right, but at the last reporting we found it had cleared a Senate committee and had not become law. If you hire voice talent, build with AI audio, or are a performer worried about your own sound, this guide walks through what applies now and what to put in writing.
TL;DR
No federal statute currently gives every person a general right over their synthetic voice. The NO FAKES Act would, but it was still a bill when we last checked.
Protection today comes from a mix of state publicity and privacy laws, contracts, consumer protection and FCC rules on AI-cloned robocalls.
The safe rule for any commercial use: get written, specific consent for the replica, the uses, the term and the payment.
The penalties can be steep: the FCC finalized a $6 million fine against an individual behind an AI-cloned robocall campaign.
Your agreement, not the technology, is what keeps a project clean. Spell out replicas explicitly.
Table of Contents
Is AI voice cloning illegal in the United States?
Cloning a voice is not, by itself, a crime anywhere in the country. What creates liability is what you do with the clone and whether the person agreed. A cloned voice used for a private experiment sits in a very different place from one used to sell a product, mislead voters or impersonate a relative on the phone.
Three questions usually decide whether a use is a problem:
Is the voice recognizable as a specific person? A generic synthetic narrator is a different matter from a voice built to sound like a named actor or a private individual.
Is the use commercial or deceptive? Ads, product endorsements, fake calls and political messages draw the most legal attention.
Did the person consent, in writing, to this specific use? General consent to record a session rarely covers building a reusable model.
In short, the risk lives in the combination of recognizability, commercial or deceptive purpose and missing consent. Remove consent and the other two make a claim easy to bring.
What protects a voice today
Because there is no one national rule, a voice is protected by several overlapping layers. We find it helpful to think of them as a stack.
State right of publicity and privacy
Many states recognize a right to control the commercial use of your name, image and, in a number of them, your voice. Those laws typically protect against unauthorized commercial use, which is the theory voice actors have relied on in court. Coverage differs a great deal: some states are explicit that voice is included, some protect only living people, some extend rights after death, and some leave voice to court interpretation. The number of states adding AI-specific language is growing, so check the current statute where your talent lives and where your campaign runs, not just where your studio sits.
New York readers should note that the state has moved on digital replicas in contracts, a topic we return to below when we discuss agreements.
Contract law
For professionals, the contract is often the strongest protection. A recording session agreement can allow or forbid training a model, creating a replica and reusing a performance in later projects. If the contract is silent, you are left arguing over implied rights, which is expensive for both sides. Some states now limit how far a contract can go in granting replica rights when the language is vague, so vague drafting is a risk for the party who wrote it.
Copyright, and why it is the wrong tool
Copyright protects fixed creative works such as a recording, a script or a song. It does not protect the abstract qualities of a voice: its timbre, accent or delivery. That gap is exactly why the U.S. Copyright Office examined digital replicas in its report, Copyright and Artificial Intelligence, Part 1: Digital Replicas. The Office suggested that federal protection should reach all individuals, not only celebrities, because everyone is exposed to the harms of a convincing replica regardless of fame or commercial value.
If you want a refresher on how copyright and permissions work for the finished piece, our guide to content usage rights covers the basics creators and brands need.
Federal telecom rules
One area where federal law already bites is phone calls. The FCC has confirmed that AI voice synthesis falls within the Telephone Consumer Protection Act’s restrictions on artificial or prerecorded voices, which means a cloned voice on an outbound call generally needs the same consent a robocall does. The consequences are not theoretical, as the next section shows.
What happens when a synthetic voice is used without authorization
Remedies depend on the route taken. A performer whose voice was cloned for an ad might bring a state publicity or privacy claim and seek damages and an injunction. A client who relied on a vendor that lacked rights may face a contract dispute. And regulators can act independently of any private lawsuit.
The clearest federal example is the New Hampshire primary robocall case. In 2024 the FCC finalized a $6 million fine against an individual and a $1 million fine against a telecommunications carrier tied to an AI-cloned voice robocall campaign aimed at voters. That is a deception case rather than a licensing dispute, but it shows how seriously enforcers treat convincing synthetic voices when they are used to mislead.
For brands, the practical lesson is about exposure you cannot see. If an agency, a freelancer or a text-to-speech vendor supplies a voice, you inherit the risk when the rights chain is unclear. Ask where the voice came from before the spot runs, not after a complaint arrives.
Where the NO FAKES Act stands
The Nurture Originals, Foster Art, and Keep Entertainment Safe Act of 2026, known as the NO FAKES Act (S. 4591), would create a federal right against unauthorized digital replicas of a person’s voice or visual likeness. According to a Holland & Knight summary, the Senate Judiciary Committee advanced it unanimously by voice vote on June 18, 2026. The summary describes a new federal intellectual property right for every individual and liability for people and companies that produce or distribute unauthorized replicas.
Advancing out of committee is a step, not a finish line. A bill still needs to pass the full Senate and the House and be signed. Until that happens, the answer to the question “has the NO FAKES Act become law?” is no, and you should confirm the current status on Congress.gov before relying on any description of it, including ours.
The Copyright Office’s recommendations show the shape such a law might take. It proposed a federal “floor” of protection nationwide while letting states keep offering additional protection. That means even if a federal law passes, state rules would likely continue to matter, so a contract written to the stricter standard remains the safer approach.
Speech protections and the limits of voice rights
Voice rights are not unlimited. Parody, commentary, news reporting and other expressive uses raise First Amendment questions, and any serious law has to balance them. The Copyright Office recommended that federal law address those concerns expressly through a balancing framework, to avoid overbreadth, rather than relying only on narrow, categorical exemptions.
This matters for creators as much as for performers. A satirical sketch using an impression is not the same as a fake endorsement, and lawyers will look at context, labeling and commercial purpose. Human judgment on those lines is hard to automate, which is why we agree with the argument for human review of AI-generated content before anything synthetic is published.
When you need consent or a license
As a working rule, get written permission whenever any of these is true:
You are building a voice model from a real person’s recordings.
The output is recognizable as a specific person, even if no one says the name.
The use is commercial: advertising, branded content, product demos, training courses sold to others.
The audience could reasonably believe the person endorsed or said the words.
The performance will be reused in a project other than the one it was recorded for.
Permission for the original session does not automatically extend to any of these. A person who sold you an hour of narration did not necessarily sell you a reusable digital twin.
The same logic applies to likeness and music. If your campaign also touches faces or soundtracks, see our notes on model releases for AI and ads and on licensing AI music safely for U.S. campaigns.
What a voice-use agreement should specify
A strong agreement leaves little to interpretation. Whether you are the client or the performer, check that it addresses each of the following in plain language.
Clause | What to specify | Why it matters |
Replica scope | Whether a synthetic model or digital replica may be created at all | Silence invites a dispute over implied rights |
Permitted uses | Exact projects, media, products and territories | Prevents reuse in ads or categories the performer never agreed to |
Prohibited uses | Political content, adult content, impersonation, sensitive categories | Protects reputation beyond the paycheck |
Payment | Fee for the session, fee for the replica, and fees for each new use | Separates one-time work from ongoing value |
Duration | Start date, end date, renewal terms | A voice license without a term can run indefinitely |
Approval | Who reviews new scripts or outputs and how fast | Keeps control with the person whose voice it is |
Storage and deletion | Where the model lives, who can access it, what happens at the end | Limits leaks and unauthorized reuse |
Disclosure | Whether and how audiences are told the voice is synthetic | Reduces consumer protection risk |
Legal advice | Whether each side had a chance to consult counsel | Helps if a vague clause is later challenged |
Drafting this in your own words is a job for an entertainment or media attorney in your state, particularly if a union agreement or a New York or other state digital replica rule could apply to the performer. Ask them to review the replica and duration clauses first, because those are where disputes tend to start.
A practical checklist for brands and producers
Map every voice in the project. Narration, characters, music vocals, call center audio, social clips.
Record where each voice came from. A signed agreement, a vendor license, or a stock library with clear terms.
Read the vendor’s terms. Many text-to-speech tools restrict commercial or cloned use; confirm what your plan covers.
Label synthetic audio internally and, where appropriate, publicly. Our write-up on synthetic media policy, C2PA and takedowns shows how to build that process.
Keep call and outreach use separate. Anything delivered by phone should be reviewed against FCC rules before launch.
Set a takedown path. If someone objects, you need a named person and a deadline to respond.
Re-check state law before each new campaign. The landscape is moving, and a spot cleared last year may need another look.
For teams working in production, the same discipline applies to the pipeline. Our look at AI workflows for video production teams explains where to keep people in the loop, and our piece on generative AI in branding argues for rules rather than ad hoc prompts.
Our perspective: voices are craft, not raw material
We came to this topic as producers. Every voice we have recorded, from a quiet documentary narration to a high-energy spot, carried a performer’s choices: breath, timing, a smile you can hear. That is craft built over years, and we think of it as the emotional heartbeat of a piece, not as data to be harvested.
So our approach to synthetic audio is simple. We treat consent as part of the creative brief, not as paperwork at the end. We build projects with a curated network of collaborators, and we want those collaborators to know exactly how their work will be used and to be paid fairly when a new use appears. Synthetic tools can be wonderful for temp tracks, accessibility, translation drafts and ideas that would never get a budget. They are at their best when a person’s voice is used with that person’s blessing.
We also believe clients benefit from this stance. A campaign with a clean rights chain can run anywhere, be reused without fear and survive scrutiny. A shortcut that saves a day in the studio can cost months later.
— Vain.
Working with Vain. on audio and synthetic media
If you are planning a campaign that involves voice, whether human talent, synthetic narration or a mix, our team can help with the creative and the production plan. We offer video, audio, animation and strategy services for brands, artists and organizations in New York City and beyond, and we can help you map consent, approvals and usage into the project from day one.
Tell us what you are making, who the voices are, and where it will run, and we will help you build a process you can defend.
FAQ
Is AI voice cloning illegal in the US?
Not in itself. Liability depends on the use: commercial exploitation of a recognizable voice without consent, deceptive impersonation, or automated calls can all create legal exposure under state publicity and privacy laws, contracts or FCC rules. Consent in writing is the cleanest protection.
Has the NO FAKES Act become law?
Not as of the most recent reporting we found. The Senate Judiciary Committee advanced S. 4591 on June 18, 2026, but a bill must still pass both chambers and be signed by the President. Check Congress.gov for the current status.
Can I own my voice?
You cannot copyright a voice as such, but you can often control its commercial use through state right-of-publicity law and through the contracts you sign. The Copyright Office has recommended broader federal protection that covers everyone, not only famous people.
Do I need permission to use a synthetic voice in an ad?
If the voice imitates or is trained on a real person, yes, get written permission that covers replicas, the specific uses, payment and duration. If you use a licensed stock synthetic voice, confirm the vendor’s terms allow commercial use and that the voice was created with consent.
What should I do if my voice was cloned without permission?
Save evidence such as links, recordings and dates, then speak with an attorney in your state about publicity, privacy or contract claims. Report deceptive robocalls to the FCC and ask the hosting platform for a takedown.
Sources
Recommended
Comments