Last updated: July 2026
Is AI voice cloning actually legal?
There’s no single US federal law yet. It depends heavily on where you are and whether the person consented.
Whether AI voice cloning is legal depends entirely on two things: where you are, and whether the person being cloned consented. Cloning your own voice with your own permission is a different legal situation than cloning someone else’s voice without asking, and the law is only just starting to catch up to that distinction.
Why is this suddenly a real legal question?
Voice cloning used to be a research curiosity. Now it’s cheap, fast, and has already been used in documented fraud cases, including a 2019 case where a cloned executive voice led to a $243,000 wire transfer, and a 2021 case where a cloned voice contributed to a $35 million loss. We covered both in detail in our guide to AI voice scams. Lawmakers moved because real harm was already happening, not as a theoretical exercise.
US states with specific voice-cloning laws
As of July 2026, six states have passed laws that specifically address AI voice cloning or digital voice replicas, rather than relying only on older publicity-rights law, per Recording Law’s state-by-state deepfake and voice cloning tracker.
| State | Law | Effective | What it does |
|---|---|---|---|
| Tennessee | ELVIS Act | 2024 | First US law to explicitly extend right-of-publicity protection to AI-generated voice clones, with criminal and civil remedies |
| California | Civil Code §3344 | Ongoing, updated | Civil penalties for unauthorized commercial use of a person’s voice, among the strictest in the country |
| Washington | SSB 5886 | June 11, 2026 | Updates the state’s Personality Rights Act to cover forged digital likenesses, including voice, with damages up to $3,000 plus noneconomic harm |
| Arkansas | Act 159 | February 25, 2025 | Amends the Frank Broyles Publicity Rights Protection Act to explicitly cover AI reproduction of voice for commercial use |
| Pennsylvania | Act 35 of 2025 | 2025 | Creates a criminal offense, “digital forgery,” for creating or distributing a forged digital likeness, including audio |
| New York | Statutory right of publicity | 2025 | Recognizes a formal statutory right of publicity that applies to unauthorized digital replicas |
Notice how recent every one of these is. Four of the six were passed or took effect in 2025 or 2026. This is not settled law anywhere yet, it’s actively being written right now.
What applies everywhere else in the US?
Roughly 44 states have no voice-cloning-specific statute as of this writing. That does not mean voice cloning is unregulated there. Three older legal frameworks still apply almost everywhere:
- General right-of-publicity law. Most states already protect a person’s name, likeness, and voice from unauthorized commercial use, even without an AI-specific update. Courts are increasingly willing to read “voice” broadly enough to include an AI clone of it.
- Fraud and wire fraud statutes. If a cloned voice is used to steal money, that’s fraud regardless of whether the state has a dedicated AI law. This is the legal basis prosecutors have used in cloned-voice scam cases so far.
- FTC authority. The Federal Trade Commission can act against deceptive AI voice practices under its existing consumer protection authority, and ran a public Voice Cloning Challenge specifically to develop better detection and prevention tools.
In practice, the honest summary is: unauthorized voice cloning for fraud or impersonation is very likely illegal almost everywhere in the US already, through older law, even in states without a dedicated statute. What the new state laws add is clarity, specific penalties, and, in some cases, criminal liability rather than only civil remedies.
It’s worth sitting with why right-of-publicity law was written for a world that no longer quite exists. Most state publicity statutes were drafted decades ago, often in response to a celebrity’s photo or catchphrase being used in an ad without permission, a single static use, easy to point to, easy to stop. Voice cloning breaks several of that framework’s built-in assumptions at once: the “use” can be generated fresh for every call rather than copied from an existing recording, the infringing audio can sound indistinguishable from a real, spontaneous conversation rather than a obviously staged advertisement, and the harm often isn’t commercial exploitation of fame at all, it’s fraud against someone with no public profile whatsoever. Courts applying 1970s-era publicity law to 2026-era voice cloning are, in a real sense, translating a statute into a technical reality its authors could not have anticipated, which is exactly why the six states above chose to write new, purpose-built language instead of trusting the old text to stretch far enough.
The EU AI Act’s real deadline: August 2, 2026
Article 50 of the EU AI Act, as detailed by the official EU Artificial Intelligence Act explainer site, creates a transparency requirement specifically for AI-generated or manipulated audio, image, video, and text, deepfakes, in the law’s own terminology. Two separate obligations apply. Providers of tools that generate synthetic audio must build in machine-readable marking so the output can be detected as AI-generated. Deployers, the businesses and individuals actually using those tools, must disclose when someone is encountering AI-generated or altered content.
There’s a narrower carve-out for content that’s evidently artistic, satirical, or fictional, where the disclosure requirement is reduced rather than eliminated. For anything that could plausibly be mistaken for a real person’s actual voice, the disclosure obligation applies in full. These transparency rules start applying on August 2, 2026, which is days away as of this writing, making this one of the most immediately relevant legal deadlines in this entire space.
What is likely to happen next?
Four of the six state laws above passed in 2025 or 2026 alone. That pace is the real signal here, not the current count. State legislatures often move in clusters once one state’s law survives its first real court test, and several more states reportedly have similar bills under consideration as of mid-2026. Assume this list of six states will look outdated within a year or two, not because this guide is wrong today, but because the underlying law is moving fast. If you operate in multiple states, it’s worth checking back rather than treating any specific list as permanent.
The same applies internationally. We can verify the EU’s Article 50 deadline directly because it’s a single, dated, binding rule. We have not independently verified specific voice-cloning statutes in the UK, Canada, or other major markets as of this writing, and we’d rather say that plainly than imply broader international coverage than we’ve actually confirmed.
Is there a landmark court case testing this right now?
The clearest sign that this law is still being written in real time is Lehrman v. Lovo, Inc., a case moving through the US District Court for the Southern District of New York, covered in a legal alert from law firm Fred Law. Two professional voice actors, Paul Skye Lehrman and Linnea Sage, allege that AI company Lovo obtained recordings of their voices through Fiverr in 2019 and 2020, under the stated purpose of internal research, then used the recordings to create commercial synthetic voice clones without further permission. One of the plaintiffs says he discovered his own cloned voice narrating a podcast episode he had never recorded.
The lawsuit brings sixteen separate claims, including breach of contract, Lanham Act violations, Copyright Act claims, and New York right-of-publicity claims. On July 10, 2025, the court partially granted and partially denied Lovo’s motion to dismiss, meaning some claims survived and are proceeding. Whatever the eventual outcome, this case is likely to shape how courts handle consent, scope of use, and voice ownership in future disputes, well beyond just this one company.
What does this actually mean for you?
If you’re an individual whose voice might be cloned without permission, you likely already have some legal recourse through right-of-publicity or fraud law, even outside the six states above, and stronger, clearer protection if you’re in one of them.
If you’re a business using voice AI, consent is the deciding factor in nearly every framework here, state law, EU law, and basic ethical practice all converge on the same requirement. We covered how legitimate platforms build in consent verification specifically because of this in our comparison of AI and human voice actors, and how disclosure requirements already apply to AI voice agents in customer-facing calls in our explainer on how voice agents work.
If you operate across state lines or serve EU users, the safest practical baseline is to treat the strictest applicable rule as your floor, obtain clear consent before cloning any real person’s voice, and disclose AI-generated audio rather than trying to track which specific jurisdiction applies to which user. Trying to apply a different, looser standard state by state is a real operational burden for very little practical benefit, since the strictest rule you’re subject to sets your actual floor anyway.
If you’re a content creator using AI voice tools for narration, ads, or videos, cloning a real person’s voice without permission carries real legal exposure even if you never intended fraud. Using a licensed synthetic voice, or your own cloned voice, sidesteps this entirely. The distinction that matters legally is whose voice it is and whether they said yes, not how the audio was technically produced.
Sources
- Recording Law, “Deepfake & AI Voice Cloning Laws by State (2026),” recordinglaw.com/us-laws/deepfake-laws, retrieved 2026-07-31
- Wikipedia, “ELVIS Act,” en.wikipedia.org/wiki/ELVIS_Act, retrieved 2026-07-31
- EU Artificial Intelligence Act (official explainer site), “Transparency Rules: A Practical Guide to Article 50,” artificialintelligenceact.eu/transparency-rules-article-50, retrieved 2026-07-31
- Fred Law, legal alert on Lehrman v. Lovo, Inc., fredlaw.com, retrieved 2026-07-31
We verified these against the specific facts cited above at time of writing. We have not independently cross-checked each source against a second, fully independent publisher for every individual statute date, so treat the state-by-state table as a strong starting point for your own verification, not a substitute for checking the actual statute text or consulting an attorney.
Frequently Asked Questions
Is this page legal advice?
No. This is general, publicly sourced information as of July 2026, not legal advice for your specific situation. Consult a licensed attorney for that.
Is it illegal to clone AI voices in most US states?
There’s no single answer. Six states have specific voice-cloning statutes as of 2026. The other roughly 44 rely on general right-of-publicity and fraud law, which often still applies to unauthorized cloning even without a dedicated AI statute.
Can I legally clone my own voice?
Cloning your own voice with your own consent is a fundamentally different legal situation than cloning someone else’s without permission. Consent is the central factor across nearly every framework covered here.
What is the ELVIS Act?
Tennessee’s ELVIS Act, passed in 2024, was the first US state law to explicitly extend right-of-publicity protection to AI-generated voice clones, with both criminal and civil remedies.
What does the EU AI Act require for AI voices?
Article 50 requires providers of synthetic audio tools to build in machine-readable marking, and requires deployers to disclose when content is AI-generated. These rules start applying August 2, 2026.
Does using AI voice cloning for fraud count as a crime even without a specific state law?
Yes. Fraud and wire fraud statutes apply regardless of whether a state has passed an AI-specific voice cloning law. This is the legal basis used in documented cloned-voice fraud cases so far.
Are there real court cases testing these laws right now?
Yes. Lehrman v. Lovo, Inc. in the Southern District of New York is a closely watched case where two voice actors allege their voices were cloned commercially without proper consent. A court ruling in July 2025 allowed several of their claims to proceed.
Will more states pass voice-cloning laws soon?
Likely, based on the pace so far. Four of the six state laws covered here passed in 2025 or 2026 alone, and several other states reportedly have similar bills under consideration.
Do these laws cover countries outside the US and EU?
We haven’t independently verified specific voice-cloning statutes in the UK, Canada, or other major markets as of this writing, so we’re not claiming coverage we haven’t confirmed.
Laws in this area are changing quickly. This page reflects publicly available information as of July 2026 and is not a substitute for advice from a licensed attorney.
Richard Johnson
Richard Johnson is an AI specialist with over five years of experience guiding large organizations through AI adoption, across more than 100 customers. He founded CognitiveFuture to research and compare AI tools across design, development, writing, research, voice and business, cutting a crowded, fast-moving market down to the right choice for the job in front of you.
