Deepfake & AI Voice Cloning Laws by State (2026)

Deepfake laws vary widely across the United States. As of 2026, at least 45 states have enacted some form of law covering sexual deepfakes, election deepfakes, or AI voice cloning, with the federal TAKE IT DOWN Act (signed May 19, 2025) providing a national baseline for nonconsensual intimate deepfakes. This guide covers all three legal buckets for every state and DC.

What Is a Deepfake, and What Laws Apply?

A deepfake is media (video, audio, or still images) in which artificial intelligence replaces or synthesizes a real person's face, voice, or likeness to make them appear to say or do something they never did. The term comes from the deep learning techniques used to train these models.

Three distinct bodies of law now govern deepfakes, and states have moved at different speeds on each.

Bucket 1: Sexual deepfakes (NCII) and AI-generated child sexual abuse material. Nonconsensual intimate image (NCII) laws, sometimes called revenge-porn statutes, were originally written for real photographs and videos. Most states have now amended them to cover AI-generated synthetic content. AI-generated CSAM (depictions of minors that were never filmed) is separately covered by federal law (18 U.S.C. 2256) and by most state CSAM statutes.

Bucket 2: Election and political deepfakes. At least 28 states require disclosure or prohibit distribution of deepfake political content near elections. These laws face recurring First Amendment challenges; courts must weigh election integrity against the right to create satirical or parody content.

Bucket 3: AI voice cloning and digital likeness (right of publicity). A growing number of states have extended their right-of-publicity laws to cover AI-generated replicas of a person's voice or visual likeness used for commercial or harmful purposes. Tennessee's ELVIS Act is the most-cited model.

Many situations touch more than one bucket. An AI-generated deepfake of a candidate used in a political ad raises both election-law and likeness-rights issues. A voice clone used to commit fraud may trigger TCPA rules, state identity-theft statutes, and the FTC's impersonation rule simultaneously.

Federal Deepfake Law

TAKE IT DOWN Act: Enacted

The most significant federal development is the TAKE IT DOWN Act, signed into law on May 19, 2025 (Public Law 119-12). It is the first federal law specifically targeting nonconsensual intimate deepfakes.

The Act makes it a federal crime to knowingly publish nonconsensual intimate visual depictions of adults or minors, expressly including AI-generated "digital forgeries." Penalties reach two years in prison for adult victims and three years when minors are depicted. Critically, platforms must remove flagged content within 48 hours of receiving notice from a victim; the compliance deadline for that takedown infrastructure was May 19, 2026. The FTC enforces platform compliance.

The TAKE IT DOWN Act matters most in the nine states (including Missouri and New Mexico) that still lack adult NCII deepfake laws of their own. For those victims, the federal law is now the primary avenue for platform removal and criminal referral.

FCC AI Robocall Ruling: Enacted

In February 2024, the FCC issued Order 24-17 declaring that AI-generated voices in robocalls constitute "artificial" voices under the Telephone Consumer Protection Act (47 U.S.C. 227). This means voice-clone robocalls to any phone without prior express written consent are illegal nationwide. The ruling was triggered directly by the January 2024 fake-Biden robocall sent to New Hampshire voters ahead of the primary, a call the FCC later issued a $6 million fine for (finalized September 2024).

Federal CSAM Law: Enacted

Under 18 U.S.C. 2256(8)(B) (as amended by the PROTECT Act of 2003), federal law covers computer-generated or AI-generated images that are indistinguishable from real minors engaged in sexually explicit conduct. There is no First Amendment defense for indistinguishable material under the post-Ashcroft v. Free Speech Coalition framework. This applies even in states that have not explicitly amended their own CSAM statutes.

FTC Impersonation Rule: Enacted for Businesses

The FTC's Rule on Impersonation of Government and Businesses (16 C.F.R. Part 461, effective April 1, 2024) prohibits deceptive impersonation of government entities and businesses using AI voice cloning. The extension of the rule to individual impersonation remains a proposed rulemaking (NPRM) and is not yet final.

DEFIANCE Act and NO FAKES Act: Pending, Not Law

Two significant federal bills have been introduced but are not enacted as of September 2026:

The DEFIANCE Act (S.1837 / H.R.3562, 119th Congress) would create a federal civil cause of action for victims of sexual deepfakes with liquidated damages of $150,000 (rising to $250,000 if the conduct involved actual or attempted sexual assault, stalking, or harassment). An earlier version passed the Senate in July 2024 but died in the House before the end of the 118th Congress; the reintroduced S.1837 passed the Senate again by unanimous consent on January 13, 2026 and is now pending in the House. See our DEFIANCE Act explainer for current status.

The NO FAKES Act (S.4591 / H.R.8915, 119th Congress) would create a federal right of publicity against unauthorized AI digital replicas of a person's voice or visual likeness. The Senate Judiciary Committee ordered it reported with a substitute amendment on June 18, 2026 and reported it on June 24, 2026, placing it on the Senate Legislative Calendar. It has not passed either chamber; the House companion remains in the House Judiciary Committee.

Deepfake Laws by State

The table below shows the enacted status of each of the three legal buckets for all 50 states and DC. Each state name links to the full state guide. "Yes" means an enacted law exists. "No" means no state law; federal law may still apply.

Northeast states

The table below covers 9 states in the Northeast.

Midwest states

The table below covers 12 states in the Midwest.

Southern states

The table below covers 16 states in the South, plus the District of Columbia.

Western states

The table below covers 13 states in the West.

AI Voice Cloning Laws

The ELVIS Act: The National Model

Tennessee's Ensuring Likeness Voice and Image Security Act (the ELVIS Act, Tenn. Code Ann. 47-25-1101 et seq., effective July 1, 2024) was the first state law to extend a right of publicity specifically to AI simulations of a person's voice. "Voice" is defined as a "readily identifiable sound" whether it is the person's actual voice or a simulation of it. The law reaches not just users of unauthorized voice replicas but also anyone who knowingly distributes a tool whose "primary purpose or primary effect" is to produce such unauthorized replicas.

Enforcement is civil and criminal: victims may seek injunctions, impoundment, actual damages plus profits, and attorney fees under Tenn. Code Ann. 47-25-1105. A Class A misdemeanor criminal penalty applies under 47-25-1108. Postmortem rights extend 10 years after death.

At least seven states have enacted comparable protections since 2024. Montana (HB 513, effective January 1, 2026) created property rights over name, voice, and visual likeness with damages up to $50,000 per incident. Arkansas (Act 159, effective February 25, 2025) amended its Frank Broyles Publicity Rights Protection Act to explicitly cover AI reproduction of voice for commercial purposes, one of the first to add a specific AI amendment to an existing right-of-publicity statute. Washington (SSB 5886, effective June 11, 2026) updated the state's Personality Rights Act (RCW 63.60) to cover "forged digital likenesses" including AI voice, raising the civil penalty to $3,000 plus noneconomic damages. Our Washington digital likeness explainer covers that law in detail.

Pennsylvania took a different approach with Act 35 of 2025 (18 Pa.C.S. 4101.1, effective September 5, 2025). Rather than a right-of-publicity model, Pennsylvania created a criminal offense called "digital forgery": creating or distributing a forged digital likeness (including audio of a person's voice) as genuine with intent to defraud or injure. It is a first-degree misdemeanor escalating to a third-degree felony in fraud schemes.

Several states now also treat using AI to impersonate someone for money as its own crime. Arizona, for example, amended its criminal impersonation statute in 2025 (A.R.S. § 13-2006(A)(4)) to cover using a computer-generated voice, image or video of another person to defraud. Those fraud-focused laws, and the federal statutes prosecutors use, are covered in deepfake fraud and impersonation laws; scam phone calls that use a cloned voice are covered in are AI voice scam calls illegal. When the cloned voice belongs to a relative asking for bail or ransom money, our guide to grandparent and family emergency scams explains how to verify the call before paying.

Indiana (IC 32-36-1) and Nevada (NRS 597.770) have older right-of-publicity statutes that explicitly list "voice" among the protected personality elements, making them applicable to voice cloning for commercial use even without a specific AI amendment.

Illinois extended its Right of Publicity Act (765 ILCS 1075) via HB 4875 (signed August 2024, effective January 1, 2025) to add "digital replica," "generative artificial intelligence," and "unauthorized digital replica" as defined terms, prohibiting distribution of works containing unauthorized digital replicas while providing safe harbors for cloud providers lacking actual knowledge.

Illinois also has BIPA. The Biometric Information Privacy Act (740 ILCS 14/) requires written consent before a private entity collects or uses biometric identifiers, including voiceprints. BIPA's $1,000-to-$5,000 per-violation damages have driven some of the largest privacy-related class actions in the country and apply to any AI voice-data collection from Illinois residents.

FCC Robocall Rule

The FCC's February 2024 ruling (Order 24-17) that AI-generated voices in robocalls are "artificial" under the TCPA applies nationwide. Marketers, political groups, and fraudsters alike are now barred from using cloned voices in phone calls without prior express consent.

Election Deepfakes and the First Amendment

At least 28 states have enacted laws restricting deepfakes in election communications. The approaches split between two models: disclosure (requiring a prominent statement that content was AI-generated) and prohibition (banning deceptive deepfakes outright, usually within a window before an election). Prohibition-only laws carry higher First Amendment risk.

California's AB 2839 (2024) was the most ambitious election deepfake ban, but on August 29, 2025 a U.S. District Court held it unconstitutional in Kohls v. Bonta (E.D. Cal. No. 2:24-cv-02527). The court found the law a content-based, speaker-based restriction subject to strict scrutiny; it accepted that California has a compelling interest in election integrity but held that AB 2839 was not the least restrictive means of serving it, and that no portion of the statute could be severed to save it.

Read the remedy carefully. The order grants the plaintiffs summary judgment and permanently enjoins the Attorney General and Secretary of State "from enforcing AB 2839 against the named Plaintiffs." It does not by its terms erase AB 2839 from the California codes or enjoin enforcement against anyone else, so treat the statute as held unconstitutional by a district court rather than repealed. California's earlier, narrower AB 730 (2019) disclosure-only law survived scrutiny.

Minnesota's Minn. Stat. 609.771 (enacted 2023, effective August 1, 2023) prohibits spreading a deepfake within 90 days before an election with intent to damage a candidate. It has faced First Amendment challenges; as of February 2026, a preliminary injunction was denied and the litigation continues. The state's election deepfake law remains in force pending appeal.

Texas enacted the first election deepfake law in the nation in 2019 (Tex. Election Code 255.004), but a subsection was later ruled unconstitutionally overbroad. The core prohibition on creating and publishing a deepfake video within 30 days of an election with intent to injure a candidate was upheld.

Louisiana's governor vetoed its election deepfake bill in July 2024 on First Amendment grounds. Hawaii's 2024 election deepfake law (Act 191, SB 2687) was enacted but permanently enjoined on January 30, 2026 by a federal court as an unconstitutional restriction on political speech (Babylon Bee v. Hawaii). The pattern suggests that broad prohibition models (particularly those lacking a clear "intent to deceive" or narrow temporal-window requirement) face the most constitutional vulnerability.

Pure disclosure requirements, by contrast, have generally survived challenge. Utah (SB 131), Colorado (HB 24-1147), Oregon (SB 1571), and Wisconsin (Act 123) all use disclosure models.

What to Do If You Are the Victim of a Deepfake

Step 1: Request platform removal. Under the TAKE IT DOWN Act, platforms must remove flagged nonconsensual intimate images (including AI-generated deepfakes) within 48 hours of a verified victim notice (effective May 19, 2026). File notices directly through each platform's reporting channel, see how to file a Take It Down Act request for the exact steps, and reference the Act by name if the platform is slow to respond.

Step 2: Document everything. Take screenshots with timestamps of where the content appears and any communications from the person responsible. Documentation preserves evidence for criminal and civil proceedings.

Step 3: Report to law enforcement. If you are in a state with an enacted NCII deepfake law, file a report with local police or the state attorney general. Even in states without specific deepfake laws, federal law (TAKE IT DOWN Act for intimate images; 18 U.S.C. 2256 for CSAM) applies, and the FBI has jurisdiction over federal crimes.

Step 4: Pursue civil remedies. States including California, Colorado, Florida, Michigan, Minnesota, New Hampshire, New Jersey, New York, and others provide explicit private rights of action with statutory damages. Consult an attorney about the statute of limitations in your state; some run as short as one year from discovery.

Step 5: Contact the Cyber Civil Rights Initiative. The CCRI (cybercivilrights.org) provides crisis helpline support and can help identify legal resources in your state.

If the deepfake involves election interference, contact your state elections authority and, if the candidate is depicted, their campaign's legal team, as many states give candidates standing to seek emergency injunctive relief.

How Deepfake Laws Interact With Recording Laws

Deepfake creation and recording consent are distinct legal frameworks, but they intersect in important ways. Recording a conversation without consent may itself be illegal under your state's wiretapping statute; see United States recording laws by state for a full breakdown. But if someone uses an AI tool to clone your voice from a legally-made recording, the recording-consent question is separate from the voice-cloning question.

Meeting transcription tools and AI note-takers add another layer. If an AI meeting recorder captures your voice and that audio is later used to train a voice-cloning model, you may have claims under BIPA (if you're in Illinois), your state's right-of-publicity statute, or the FTC Impersonation Rule, regardless of whether the original recording was lawful. Our guide to AI meeting recording laws covers consent requirements for AI-assisted recording tools specifically.

For a broader look at how AI intersects with state law, see our AI laws by state hub, and for data privacy frameworks that often govern AI data collection, see data privacy laws by state.

Disclaimer: This page provides general legal information, not legal advice. Deepfake laws are changing rapidly; confirm current statutes with a licensed attorney or the official state legislature website for your jurisdiction.

Frequently Asked Questions

Are deepfakes illegal in the US?

Creating or distributing a deepfake is illegal in many circumstances but not all. The federal TAKE IT DOWN Act (signed May 19, 2025) makes it a federal crime to publish nonconsensual intimate deepfakes of any person. More than 40 states add their own criminal or civil penalties. Deepfakes used for clear satire or parody retain some First Amendment protection, but distributing a sexual deepfake of a real person without consent is illegal under federal law and the laws of most states.

What states have deepfake laws?

As of June 2026, at least 45 states have enacted at least one deepfake law. The most comprehensive coverage (all three buckets of sexual deepfakes, election deepfakes, and voice or likeness rights) exists in Tennessee, California, New York, Washington, New Jersey, and Montana. Missouri and New Mexico are the most notable states without an adult NCII deepfake law; Alaska, Georgia, Ohio, and Hawaii also lack most deepfake-specific protections.

Is it illegal to make a deepfake of someone?

It depends on the purpose. Creating a deepfake for clear satire or parody is generally protected speech. Creating a deepfake to distribute nonconsensual intimate images is a federal crime (TAKE IT DOWN Act) and a state crime in more than 40 states. Creating election deepfakes without disclosure is illegal in at least 28 states. Creating AI-generated child sexual abuse material is illegal under federal law (18 U.S.C. 2256) regardless of whether a real child was used.

Can you sue someone for making a deepfake of you?

Yes, in many states. California (Civ. Code 1708.86), Colorado (SB 25-288), Florida (Fla. Stat. 836.13), Michigan (PA 11/12 of 2025), Minnesota (Minn. Stat. 617.262), New Hampshire (RSA 507:8-j), New Jersey (P.L. 2025 c. 40), and New York (Civ. Rights Law 52-c) all provide explicit private rights of action. States with right-of-publicity statutes covering voice (including California, Indiana, Illinois, Nevada, and Tennessee) allow suits for unauthorized commercial use of AI-cloned voice or likeness.

Are AI voice clones illegal?

AI voice cloning without consent is illegal in several contexts. The FCC has ruled that AI voice-clone robocalls violate the TCPA nationwide. The FTC's Impersonation Rule bars AI voice cloning used to impersonate businesses or government agencies. States including Tennessee (ELVIS Act), California, Illinois, Indiana, Nevada, Montana, New Hampshire, New Jersey, New York, Pennsylvania, and Washington have laws that prohibit unauthorized commercial or harmful use of a cloned voice. Using a voice clone to commit fraud can also trigger general fraud and identity-theft statutes in every state.

What is the TAKE IT DOWN Act?

The TAKE IT DOWN Act (Public Law 119-12, signed May 19, 2025) is the first federal law specifically targeting nonconsensual intimate deepfakes. It makes it a federal crime to knowingly publish nonconsensual intimate visual depictions of adults or minors, expressly including AI-generated 'digital forgeries.' Platforms must remove flagged content within 48 hours of victim notice. The FTC enforces compliance. Criminal penalties reach two years in prison for adult victims and three years for minor victims.

What is the ELVIS Act?

The ELVIS Act (Ensuring Likeness Voice and Image Security Act, Tenn. Code Ann. 47-25-1101 et seq., effective July 1, 2024) is Tennessee's law extending the right of publicity to AI simulations of a person's voice. It was the first state law specifically written to address AI voice cloning. It protects any 'readily identifiable sound' including simulations of a person's voice, covers tools whose primary purpose is creating unauthorized replicas, and provides civil damages plus a Class A misdemeanor criminal penalty. Several states have modeled subsequent voice-cloning laws on the ELVIS Act.

Is deepfake porn illegal?

Yes, in most of the country. The federal TAKE IT DOWN Act criminalizes nonconsensual intimate deepfakes nationwide. More than 40 states add separate criminal and civil remedies. Notable penalties: Louisiana (10-30 years), Florida (felony plus civil damages minimum $10,000), Kansas (person felony), Nevada (Category D felony, 1-4 years), and Delaware (Class G felony). A handful of states (including Missouri, New Mexico, Alaska, Georgia, Ohio, and Hawaii) still lack adult NCII deepfake statutes, but federal law still applies in those states.

Are election deepfakes illegal?

Election deepfakes without proper disclosure are illegal in at least 28 states. Laws vary: some prohibit deceptive deepfakes outright (Minnesota, Michigan, Mississippi), while others require prominent disclosure that content is AI-generated (Colorado, Utah, Oregon, Wisconsin). All face ongoing First Amendment scrutiny. California's broad prohibition in AB 2839 was held unconstitutional in August 2025 in Kohls v. Bonta, which permanently enjoined enforcement against the plaintiffs in that case but left the statute on the books; its older disclosure-only law survived. Texas enacted the nation's first election deepfake law in 2019. About 22 states have no election deepfake law as of June 2026.

Is it illegal to deepfake a dead person?

It depends on the state and purpose. Several states extend right-of-publicity protections postmortem: California (70 years, Civ. Code 3344.1, amended in 2025 to cover AI digital replicas of deceased performers), New York (Civ. Rights Law 50-f, 40 years), Indiana, Nevada (50 years), and Tennessee (ELVIS Act, 10 years postmortem) all cover deceased persons. Texas protects deceased persons' voice as a property right for 50 years (Tex. Prop. Code ch. 26). Federal CSAM law applies equally to deepfakes of deceased minors. Outside of right-of-publicity states, the family of a deceased person may have fewer legal options.

Does Section 230 protect platforms that host deepfakes?

Section 230 (47 U.S.C. 230) does not immunize platforms from federal criminal law or from the TAKE IT DOWN Act's removal obligations. State criminal laws targeting the creators (not the platforms) of deepfakes are also not barred by Section 230. However, Section 230 may still limit state civil claims against platforms for third-party user content, meaning victims typically must pursue the creator, not just the hosting platform, in civil litigation.