Is AI Voice Impersonation Legal? The Laws Explained

AI voice impersonation sits at the intersection of powerful technology and legal territory that courts and legislatures are still working to map. Whether you are a content creator, a developer, or just curious after hearing a convincing deepfake clip, the question of legality matters — and the answer is genuinely complex. This post gives you a factual overview of the legal areas that govern AI voice impersonation: right of publicity, fraud, defamation, copyright, consumer-protection rules, and the wave of new deepfake legislation moving through US states and the EU.

This post is general information only, not legal advice. Laws vary significantly by country and US state, change frequently, and their application to any specific situation depends on facts a lawyer would need to evaluate. Nothing here should be taken as a definitive ruling on any specific case. If you face a real legal question, consult a qualified attorney in your jurisdiction.

TL;DR

AI voice impersonation is not automatically illegal, but it is not automatically legal either — purpose, consent, and jurisdiction are everything.
Right-of-publicity laws protect everyone’s voice from commercial exploitation without consent, not just celebrities.
Using a cloned voice to deceive, defraud, or harass someone is criminal in virtually every jurisdiction.
The US has no single federal AI-voice law yet, but state laws are multiplying rapidly; the EU AI Act is already in force.
Parody and satire offer some protection, but not a blanket pass.
Cloning your own voice or a voice you have explicit written permission to use is the lowest-risk path.

What Does “AI Voice Impersonation” Actually Mean?

Before getting into law, it helps to be precise about the technology. AI voice impersonation covers two related but distinct capabilities.

The first is voice conversion — transforming a speaker’s voice in real time so it sounds like a different person. This is what a real-time voice changer does. The second is voice synthesis — generating new audio in a target person’s voice from text or other input, sometimes called AI voice cloning. Both technologies use neural models trained on audio data, and both can produce results that are difficult for human listeners to distinguish from the real person.

The legal questions differ somewhat between the two. Real-time voice conversion used privately, as a character voice among friends, sits in different territory than synthesizing a politician’s voice saying words they never said and publishing it online. Much of the law written so far focuses on the latter — deepfake audio intended to deceive — but the principles underlying right-of-publicity and fraud law apply to both.

The Right of Publicity: Protecting Your Voice as Property

What is the right of publicity?

The right of publicity is a legal doctrine that gives individuals control over the commercial use of their identity — including their name, image, likeness, and voice. It developed in US common law and has been codified by statute in around half of US states. In Europe, equivalent protections exist under personality rights frameworks in civil law countries.

The right of publicity is not limited to celebrities. Anyone — a private individual, a minor, a deceased person (in many states) — can hold this right. The key question is whether someone used your identity for commercial benefit without your consent.

How does it apply to AI-cloned voices?

When a company or individual uses an AI-cloned version of a real person’s voice to advertise products, generate revenue on content platforms, or otherwise exploit that voice commercially, without the person’s permission, right-of-publicity claims are almost certainly available. Courts have found voice claims viable since at least the 1988 Bette Midler case against Ford Motor Company, in which the Ninth Circuit held that imitating a distinctive voice for advertising was actionable even without copying a specific recording.

AI cloning makes this doctrine far more practically relevant. Before, a human impersonator willing to take the risk was the main threat. Now, anyone with a laptop can produce a highly convincing likeness. The underlying legal principle — you cannot commercially exploit someone’s identity without consent — has not changed, but the scale of potential violation has.

States with particularly strong right-of-publicity statutes include California (the California Celebrities Rights Act extends protection 70 years post-mortem), New York, Texas, Illinois, and Tennessee. If your target audience is in any of these states, their laws may apply even if you are based elsewhere.

Fraud and Impersonation: When Deception Becomes Criminal

Using an AI-cloned voice to deceive someone — into sending money, revealing credentials, or taking action they would not otherwise take — is not a civil matter. It is criminal fraud. Voice fraud using AI is already being prosecuted under existing wire fraud statutes (18 U.S.C. § 1343) and identity theft laws, without waiting for AI-specific legislation.

Several high-profile cases have involved AI voice clones impersonating executives to authorize wire transfers (sometimes called “CEO fraud” or “vishing”). Financial losses in these cases have reached millions of dollars per incident. Prosecutors treat these as serious federal crimes.

Even when no financial fraud is involved, using an AI voice to impersonate a specific person in a way that deceives a third party — posing as a law enforcement officer, claiming to be a doctor, representing yourself as a business executive you are not — can violate impersonation statutes that exist in every US state and most countries.

The intent to deceive is the key element. Transforming your own voice for entertainment, or using a fictional voice character that bears no resemblance to a real person, is categorically different from crafting a recording to make listeners believe a real individual said something they did not.

Defamation: False Statements of Fact

Defamation law — libel for written content, slander for spoken — prohibits publishing false statements of fact that harm someone’s reputation. A convincing AI voice clone of a public figure saying something defamatory (for example, confessing to a crime they did not commit) is just as actionable as a written false statement, and likely more damaging because audio carries emotional weight that text does not.

For public figures in the US, the First Amendment requires proof of “actual malice” — knowledge of falsity or reckless disregard for truth — under the New York Times Co. v. Sullivan standard. For private individuals, the bar is lower. In many countries, defamation law is stricter than in the US, and the burden may fall on the defendant to prove truth.

Labeling a deepfake audio as satire does not automatically shield it. If a reasonable listener would take the content as a genuine statement by the real person, a defamation claim can survive. Context, disclaimer placement, and how clearly the content is framed as fictional all matter.

Copyright: What Protects a Voice (and What Does Not)

A human voice, as a biological characteristic, is not copyrightable. You cannot own a copyright in the sound of your voice any more than you can own the shape of your face. However, this does not mean voice cloning is copyright-free territory.

Recorded performances are typically protected. If you copy a commercially released audio recording to train an AI model, the underlying sound recording copyright (held by the label or distributor) and the copyright in the underlying musical composition may both be infringed. Several ongoing lawsuits challenge whether scraping audio for AI training constitutes copyright infringement or falls under fair use; courts have not yet reached a definitive answer.

Derivative works and passing off. Even without copying a specific recording, using an AI voice to produce content that imitates a distinctive artistic style closely enough to be passed off as genuine work by that artist could give rise to claims under both copyright (if specific protected expression is reproduced) and unfair competition law.

The copyright landscape here is actively being contested. The US Copyright Office has issued guidance stating that AI-generated content lacks human authorship and may not be copyrightable, but that does not resolve questions about training data or the rights of the people whose voices are used. Expect this area to evolve significantly in the next few years.

Consumer Protection and FTC Rules

The US Federal Trade Commission (FTC) has authority over deceptive trade practices under 15 U.S.C. § 45. Using an AI voice to impersonate a real person — a celebrity, a business executive, a government official — in commercial communications is a deceptive practice under FTC standards. The FTC finalized a rule in 2024 specifically targeting AI impersonation of individuals in addition to companies, closing a gap that existed in earlier guidance.

The FTC rule makes it unlawful to use AI to impersonate real people in commerce in a way that could harm consumers. Violations can result in civil penalties. The rule also potentially creates private rights of action for injured parties.

At the state level, consumer-protection statutes in most states contain similar prohibitions on deceptive commercial practices. California’s Consumer Legal Remedies Act and Illinois’ Consumer Fraud and Deceptive Business Practices Act, for example, have been used in cases involving false endorsements and deceptive identity use.

Emerging Deepfake Legislation in the United States

The most targeted legal development in this space is the wave of state and federal legislation specifically addressing AI-generated audio and video of real people.

Jurisdiction	Law / Bill	Key Provisions
Tennessee	ELVIS Act (2024)	Civil and criminal liability for AI voice reproduction without consent; covers commercial use
California	AB 2602 (2024)	Requires consent for AI replicas of performers in entertainment contracts
California	AB 1836 (2024)	Restricts use of AI replicas of deceased performers
Texas	SCOPE Act	Prohibits AI-generated sexual deepfakes; broader provisions pending
Illinois	BIPA + pending AI bills	Biometric data protections apply to voice prints; AI-specific bills in committee
New York	Multiple pending	Personality rights expansion to cover AI-generated likenesses
Federal (US)	NO FAKES Act (proposed)	Would create federal right protecting voice and likeness from AI replication without consent
Federal (US)	DEFIANCE Act (2024)	Civil cause of action for non-consensual intimate deepfakes

The trend is clear: legislators are moving toward explicit protection, not away from it. The absence of a single federal AI-voice law does not mean the space is unregulated — it means it is regulated by a patchwork of statutes that is getting denser every legislative session.

The EU AI Act and European Personality Rights

The EU AI Act, which entered into force in August 2024 and is being phased in through 2027, takes a more comprehensive regulatory approach than US state-by-state statutes.

Under the EU AI Act, AI systems that generate synthetic audio, video, or text representing real people must include disclosures stating that the content was AI-generated. This applies to any content that could “appreciably resemble” real people. The obligation falls on the provider of the AI system and on deployers who use the system to produce content.

Certain uses — deploying an AI system to impersonate natural persons in ways that could deceive the public into believing they are interacting with or listening to that person — are classified as unacceptable-risk uses that are prohibited outright under Article 5 of the Act. Penalties for prohibited use can reach 35 million euros or 7 percent of global annual turnover.

Beyond the AI Act, EU member states have personality rights frameworks in their civil codes (France’s droit à l’image, Germany’s general personality rights under constitutional law, etc.) that protect against unauthorized commercial use of a person’s voice and likeness. These coexist with the AI Act’s requirements.

The UK, post-Brexit, is developing its own AI governance framework but has not yet passed legislation equivalent to the EU AI Act. However, UK courts can apply existing privacy, defamation, and tort law to AI voice cases.

Across virtually every legal framework discussed above — right of publicity, fraud law, consumer protection, the EU AI Act — the single clearest factor that determines legality is consent. Explicit, informed consent from the person whose voice is being cloned or impersonated changes the legal picture dramatically.

“Explicit” means the person understood what they were agreeing to. “Informed” means they knew how the voice would be used. “Written” is strongly preferred because verbal consent is hard to prove. And consent given for one purpose does not transfer to a different purpose — if someone agrees to have their voice used in one podcast, that does not authorize using it in a political advertisement.

For commercial voice work, standard industry practice increasingly includes specific AI-use clauses in talent contracts. The absence of such a clause is not the same as prohibition — but it does mean the scope of rights is unclear, which creates legal exposure.

What About Using a Voice Changer Personally?

This is where many people find themselves — they want to use a voice changer to play a character in a game, sound different on Discord, protect their privacy in online calls, or create content with a distinctive voice personality. This use is fundamentally different from the scenarios above.

Using a real-time voice changer to modify your own voice for entertainment, roleplay, privacy, or creative expression does not raise right-of-publicity concerns because you are not exploiting anyone else’s identity. It does not raise fraud concerns as long as you are not using it to deceive people in ways that harm them. It does not raise defamation concerns because you are not making false statements about real people.

The legal risk in personal voice-changer use is narrow and specific: if you deliberately impersonate a specific real individual in a context where others genuinely believe they are speaking to that person, and that deception causes harm, you have crossed the line. The technology does not create the problem; the intent to deceive does.

Tools like VoxBooster are built for legitimate use cases — voice transformation, custom AI voice creation from your own voice, and privacy protection. They are used by streamers, content creators, gamers, and people with accessibility needs. The legal principles above apply to how you use any tool, not to the tool itself.

Practical Guidelines for Staying on the Right Side of the Law

To summarize the practical implications without repeating the legal analysis:

Lower-risk uses:

Cloning your own voice for personal or commercial use with clear disclosure
Using fictional voice characters with no resemblance to real people
Parody or satire that is unambiguously labeled and clearly non-commercial
Voice transformation for privacy or creative expression among consenting participants

Higher-risk uses:

Cloning any real person’s voice without explicit written consent, regardless of purpose
Using a cloned voice commercially without the subject’s permission
Creating content that could be mistaken for a genuine statement by a real person
Using a voice clone to deceive anyone about the identity of who they are hearing

Steps to take before a commercial use:

Get explicit written consent specifying how the voice will be used.
Check the right-of-publicity statute in the state where the subject lives and where you operate.
Check the relevant laws in the country where your audience is located.
Add clear disclosure labeling to any AI-generated audio content.
Consult a lawyer if you have any doubt about a specific use case.

Frequently Asked Questions

Is AI voice impersonation illegal in the United States?

It depends on purpose, consent, and the state. Using a cloned voice to commit fraud is a federal crime. Using it to harm a public figure’s reputation without consent can trigger right-of-publicity lawsuits. Several states now have specific deepfake-voice statutes. There is no single federal AI-voice law yet, but enforcement is accelerating.

What is the right of publicity and how does it apply to voice cloning?

The right of publicity protects a person’s name, likeness, and voice from commercial exploitation without consent. In states like California and New York it covers everyone, not just celebrities. Using an AI-cloned voice of a real person to sell products or services without their permission is a textbook violation.

Is voice impersonation legal for parody or satire?

Parody and satire enjoy First Amendment protection in the US and similar free-expression carve-outs in other democracies, but the protection is not unlimited. If the content could be mistaken for a genuine statement by the real person, or if it is used commercially, the parody defense weakens considerably. Consult a lawyer before publishing.

Does the EU AI Act regulate AI voice impersonation?

Yes. The EU AI Act classifies certain AI systems that impersonate real people as high-risk or prohibited, depending on context. It also requires clear disclosure when AI-generated audio could deceive the public. Non-compliance carries fines up to 30 million euros or 6 percent of global turnover.

Can I use an AI voice changer to sound like someone else in a video game or on Discord?

Using a voice changer as a character voice or for light-hearted roleplay among friends is generally low-risk. The line is crossed when you impersonate a specific real person in a way that deceives others or damages their reputation. Purpose and context matter far more than the technology itself.

What is the ELVIS Act and which states have similar laws?

Tennessee’s ELVIS Act (2024) makes it a civil and criminal offense to use AI to reproduce a person’s voice without consent for commercial purposes. California, Texas, Illinois, New York, and Georgia have passed or introduced related legislation. The trend is toward more protection, not less.

Does cloning a voice infringe copyright?

A voice itself is generally not copyrightable, but a recorded performance often is. Reproducing a protected recording to train a model, or mimicking a distinctive artistic style to pass off imitations as genuine, can infringe copyright. The law here is still developing, and several major cases are currently in litigation.

Conclusion

AI voice impersonation is one of the genuinely difficult areas where fast-moving technology is testing the limits of existing law. The honest answer to “is it legal?” is: it depends — on what you do with it, who consented, where you and your audience are located, and whether you intend to deceive. Right-of-publicity doctrine, fraud statutes, defamation law, copyright, consumer-protection rules, and new deepfake-specific legislation all overlap in this space, and they are pointing in the same general direction: toward more accountability for non-consensual and deceptive uses.

This post is general information, not legal advice. The law in this area is changing rapidly. Courts are still working out how existing doctrines apply to AI-generated audio. New statutes are being passed and challenged. If you face a specific legal question — about a project, a contract, a potential claim — get qualified legal counsel in your jurisdiction.

For uses that are straightforwardly lawful — cloning your own voice, using fictional voice characters, transforming your voice for privacy or entertainment with clear disclosure — the legal landscape is much less fraught. These are exactly the use cases that tools like VoxBooster are designed to support.

Download VoxBooster and try the 3-day free trial — real-time voice transformation, AI voice cloning from your own voice, and a full soundboard, all running locally on Windows without sending your audio anywhere.

External references: US Copyright Office AI and Copyright guidance, FTC Impersonation Rule, EU AI Act official text, Wikipedia: Right of publicity