US, Tennessee·Voice and likeness·Laws and bills·First US state law protecting performers and musicians from unauthorized AI voice cloning; bars producing a sound recording or audiovisual work that clones a voice or likeness without consent.
Extends Tennessee right-of-publicity to AI voice clones and digital replicas; cloning a person's voice or likeness without consent is barred, with developer and distributor liability when a tool's primary purpose is producing a particular identifiable individual's voice or likeness.
ElevenLabs (industry)·Voice·Provider consent policies·Gatekeeps voice cloning behind self-ownership plus verification; the acceptable-use policy bans non-consensual cloning of identifiable individuals and impersonation.
Prohibits cloning any voice other than your own; Professional Voice Cloning requires identity verification, and the voice owner must create and verify the clone on their own account before sharing it, rather than a third party cloning with consent.
US, federal·Voice and likeness·Laws and bills·First federal-level voice-and-likeness right under active consideration; distinct from tn-elvis-act by operating at the national level and by adding platform-liability and preemption provisions the state law lacks.
Creates a federal intellectual-property-style right for individuals over their own voice and visual likeness against unauthorized AI-generated digital replicas; makes producers and distributors of unauthorized replicas civilly liable, holds platforms liable if they knowingly host them, and would set a uniform federal standard alongside state right-of-publicity laws such as Tennessee's ELVIS Act.
US, Washington·Voice and likeness·Laws and bills·Extends a pre-existing state property-right statute (not a right-of-publicity statute like Tennessee's) to AI-generated audio and video specifically, including real-time forgeries; distinct remedy structure from tn-elvis-act's right-of-publicity model.
Amends Washington's existing Personality Rights Act (a property right in name, voice, signature, photograph, and likeness) to add 'forged digital likeness,' a digitally altered image or audio, persistent or transmitted in real time, that misrepresents a person and could deceive others into believing it is genuine. Infringement carries a statutory civil penalty plus actual damages, injunctions, destruction of infringing material, and attorneys' fees.
US, Pennsylvania·Voice and likeness·Laws and bills·Criminal mechanism (fraud/injury intent, misdemeanor-to-felony escalation) rather than a civil right-of-publicity claim like tn-elvis-act; explicitly reaches voice-cloning scam scenarios such as a cloned grandchild's voice used to solicit money.
Creates a new criminal offense, digital forgery, at 18 Pa.C.S. 4101.1: producing or distributing an AI-generated forged image, video, or audio (including a cloned voice) of a real person with intent to defraud or injure. Base offense is a first-degree misdemeanor, escalating to a third-degree felony when used in a fraud scheme.
US, California·Likeness·Laws and bills·Civil, non-consensual-intimate-imagery cause of action targeting AI deepfake-porn services and facilitators specifically, distinct from tn-elvis-act's commercial voice/likeness right-of-publicity focus and from criminal-fraud statutes like Pennsylvania's.
Amends Civil Code 1708.86 to give a depicted individual a civil cause of action against anyone who creates or discloses digitized sexually explicit material (AI-altered or substantially altered nude/sexual depictions) without consent, and extends liability to a person who knowingly facilitates or recklessly aids or abets such conduct, including operators of dedicated deepfake pornography services.
US, Maryland·Likeness·Laws and bills·Election-disinformation-specific deepfake statute with a state-agency correction and takedown role, distinct from tn-elvis-act's commercial right-of-publicity focus; targets political and electoral deepfakes rather than entertainment or fraud use.
Requires the State Administrator of Elections to respond to credible reports of election misinformation or disinformation, including deepfakes, by publicly correcting the record and, where warranted, seeking court orders for removal; separately makes it a criminal misdemeanor to knowingly or with reckless disregard create, use, or disseminate a deepfake to produce materially false election information, with carve-outs for satire, parody, and disclosed broadcast or print use.
US, New York·Voice and likeness·Laws and bills·Consent-by-contract mechanism (mandatory representation and specificity in licensing) for entertainment-industry digital replicas, distinct from tn-elvis-act's tort remedy and from the criminal or election statutes here; addresses the labor-negotiation stage rather than after-the-fact enforcement.
Adds General Obligations Law 5-302: any contract for the creation and use of a digital replica of a performer's voice or likeness must specifically describe the intended use, and the individual must be represented by independent legal counsel or a labor organization when negotiating the license, with terms stated clearly in a separately signed writing.
US, federal·Likeness·Laws and bills·Federal criminal-plus-platform-takedown mechanism focused on nonconsensual intimate imagery including AI-generated content, distinct from tn-elvis-act's civil right-of-publicity model and from the NO FAKES Act's broader voice and likeness IP-right approach; the only entity here with an active federal enforcement agency and a fixed 48-hour takedown clock.
Criminalizes publishing or threatening to publish nonconsensual intimate imagery, including AI-generated nonconsensual intimate imagery (deepfakes), via an interactive computer service; requires covered online platforms to implement a notice-and-removal process so a valid victim request results in takedown of the image and known identical copies within 48 hours, enforced by the FTC.
US, Louisiana·Voice and likeness·Laws and bills·Election-deepfake disclosure/labeling mechanism paired with an intent-based criminal penalty, distinct from md-election-deepfake-act's state-agency correction-and-takedown model: Louisiana's law is a disclosure mandate enforced by fine/imprisonment on the ad's creator rather than an administrator-driven public-correction process, and its legislative findings are the only source language here that names 'voice' alongside image and likeness.
Amends the Louisiana Election Code (R.S. 18:1463) to bar any electioneering communication that portrays a candidate or an elected official subject to recall and that, through AI, is created, altered, or digitally manipulated to falsely appear to a reasonable observer to be an authentic record of that person's actual speech or conduct, or to falsely replace an individual's likeness with a candidate's likeness, unless the communication carries a clear and understandable AI-use disclosure. The legislative findings clause explicitly cites protecting the public from being deceived by 'an image, likeness, or voice of a candidate or elected official subject to recall that is created using artificial intelligence.' Violating this section with intent to injure a candidate's reputation or deceive a voter is punishable by a fine of not more than two thousand dollars, imprisonment for not more than two years, or both; a media entity with no input into or control over a paid political ad is exempt.
US, California·Voice and likeness·Laws and bills·Extends an existing criminal false-personation statute to AI digital replicas of voice or likeness, distinct from ca-deepfake-pornography-act (a civil, sexual-imagery-specific cause of action under Civil Code 1708.86) and from tn-elvis-act's civil right-of-publicity tort: SB 1111 works through the general criminal impersonation code rather than creating a new publicity right or a porn-specific civil claim.
Amends California Penal Code Section 540 (criminal false personation) to clarify that false impersonation includes the use of a digital replica with the intent to impersonate another, where a digital replica is defined as a computer-generated, highly realistic electronic representation that is readily identifiable as the voice or visual likeness of an individual, whether the individual never performed at all or the fundamental character of an actual performance was materially altered.
US, Montana·Voice and likeness·Laws and bills·Frames the right as a transferable property right (like IP) rather than tn-elvis-act's right-of-publicity tort model, and conditions distributor/tool-provider liability on 'actual and specific knowledge' of primary unauthorized purpose rather than Tennessee's broader developer/distributor standard; remedy is actual damages plus profits, with no statutory minimum damages figure in the enrolled text.
Creates a property right, for Montana residents (or those domiciled in Montana at death), in their name, voice, and visual likeness that is freely transferable, licensable, and descendible, does not expire on death, and is exclusive to the individual during their lifetime and to their rights-holders for 20 years after death. A person is liable for actual damages plus profits from unauthorized use if, without consent, they (a) distribute a tool or service with actual and specific knowledge that its primary purpose is producing unauthorized digital voice or visual depictions of a particular identified individual for commercial use, or (b) intentionally publish or distribute an unauthorized digital voice or visual depiction for commercial use with actual and specific knowledge it is unauthorized. Exemptions apply to news-gathering, commentary/criticism/parody, an individual playing themself in an audiovisual work (unless intended to and does create a false impression of an authentic recording), and fleeting or negligible use. Civil actions must be brought within 4 years of discovery.
HeyGen (industry)·Voice and likeness·Provider consent policies·Distinct from elevenlabs-voice-clone-policy by offering a contractual waiver tier: enterprise accounts with a signed indemnity agreement can skip consent collection entirely, rather than every clone requiring the ElevenLabs model of self-account verification by the voice owner.
HeyGen's Avatar Consent API defines three escalating consent mechanisms for digital-twin avatars: Level 1, the subject records a short consent statement on camera through HeyGen's hosted consent page, available to all customers; Level 2, a pre-recorded consent video is submitted instead, limited to enterprise whitelisted accounts; Level 3, consent collection is waived entirely for enterprise accounts that have signed an indemnity agreement. Consent applies only to digital-twin avatars; photo avatars and prompt-to-avatar characters, which depict no real identifiable person, do not require it.
Resemble AI (industry)·Voice·Provider consent policies·Distinct from elevenlabs-voice-clone-policy by setting a verbal-consent standard for cloning a third party's voice, rather than ElevenLabs' self-account model that requires the voice owner to create and verify the clone on their own verified account.
Resemble AI's Terms of Service state Resemble may require consent from the individual or third party whose voice is being cloned, and that this consent needs to be verbal unless otherwise stated by Resemble. Users also warrant they own their content or have the necessary licenses, rights, consents, and permissions to grant Resemble the license to use it.
Synthesia (industry)·Voice and likeness·Provider consent policies·Distinct from elevenlabs-voice-clone-policy by pairing an explicit legal-age gate with an informed-consent standard covering both voice and likeness together, rather than ElevenLabs' account-verification mechanism scoped to voice cloning only.
Synthesia's Customer Terms of Service, in the section on Responsibilities To, and For, Authorized Users, require that when a customer authorizes the creation or use of a Custom Avatar using an individual's voice or likeness, that individual must be over the applicable statutory legal age and must have provided free and informed consent.
Descript (industry)·Voice·Provider consent policies·Distinct from elevenlabs-voice-clone-policy by structuring consent as a contractual representation-and-warranty tied to a defined 'Consenting Speaker' term with an express prohibition on deceptive consent statements, rather than ElevenLabs' account-based self-verification and ownership gate.
Descript's Terms of Service require that by reading the Consent Statement or otherwise submitting Training Audio, the user consents and represents and warrants they have the right to use the Consenting Speaker's AI Voice and have obtained the consent of all Consenting Speakers. The terms expressly prohibit submitting a third party's unauthorized voice recordings as training audio, or submitting deceptive or unauthorized Consent Statements.
US, New Jersey·Voice and likeness·Laws and bills·Distinct from ny-digital-replica-contracts-act (which governs pre-use licensing contracts for entertainment-industry replicas) and tn-elvis-act (a right-of-publicity tort) by creating a standalone civil cause of action with a fixed $1,000-per-violation penalty cap, an explicit label/watermark safe harbor, and a graduated remedy scheme that shields a minor defendant from damages on a first violation.
Creates civil liability for publishing, distributing, transmitting, or otherwise making available a 'realistic representation,' a highly realistic electronic representation created using a GenAI system that is readily identifiable as an individual's voice or visual likeness (whether the individual never performed at all or the performance's fundamental character was materially altered), without authorization or consent from the individual or, for a minor, a parent or legal guardian. Remedies are injunctive relief plus the greater of actual damages and reasonable attorneys' fees, or a civil penalty not to exceed $1,000 per violation. Content bearing a clear and conspicuous label or watermark identifying it as GenAI-produced is exempt. Where the offender is a minor, the prevailing plaintiff is limited to injunctive relief for a first violation only.
US, Ohio·Voice and likeness·Laws and bills·Distinct from tn-elvis-act and mt-name-voice-likeness-property-act by amending a pre-existing Ohio persona statute to drop its commercial-value gate rather than creating a new property or publicity right from scratch, and by pairing that change with a separate deepfake-recording prohibition rather than a developer/distributor liability standard.
Amends Ohio's existing persona/right-of-publicity statute and adds new deepfake-recording prohibitions. Per the Ohio House's own bill title, the act 'amends sections of the Revised Code to make changes to the law relating to the unauthorized use of an individual's persona and to prohibit certain unauthorized deepfake recordings.' Secondary reporting (not independently confirmed against enrolled text) describes the bill as removing the prior requirement that a persona have commercial value before a claim can arise, and as amended in the House Technology and Innovation Committee in May 2026 to narrow liability to those who knowingly share a deepfake they know to be false, while treating internet providers as distribution channels rather than producers.
US, California·Likeness·Laws and bills·Distinct from ca-deepfake-pornography-act (a civil cause of action under Civil Code 1708.86 for disclosure or creation of AI-altered sexual depictions generally) by operating through the criminal extortion and minor-contact statutes specifically for deepfake images used as leverage against minors, rather than creating a standalone civil claim available to victims of any age.
Amends California's contact-with-a-minor and extortion statutes to explicitly include threats, intimidation, or coercion made using an AI-generated deepfake image as a means of committing the offense. The bill's text at Section 518(c)(2)(A) covers threats made with an image 'including an image of an intimate body part that is produced by artificial intelligence and depicts the identity of another minor.' It also creates a related offense for a minor who coerces or uses another minor to harm, groom, harass, or exploit a third minor. Where the perpetrator is a minor ineligible for diversion services under Welfare and Institutions Code Section 654, the offense is a misdemeanor.
US, Pennsylvania·Likeness·Laws and bills·Distinct from pa-digital-forgery-act (Act 35 of 2025, a fraud/injury-intent criminal offense covering forged voice, image, video, or audio broadly) by amending Pennsylvania's existing nonconsensual-intimate-image statute specifically, dropping the harassment-intent element for the base offense, and defining 'artificially generated intimate image' as its own statutory term scoped to sexual imagery rather than the broader 'digital forgery' concept.
Amends 18 Pa.C.S. 3131 (unlawful dissemination of intimate image) to add a new offense for disseminating an 'artificially generated intimate image' of an individual, replacing the prior narrower 'artificially generated sexual depiction' term. The bill defines the new term verbatim as: an image that (1) 'appears to authentically depict an intimate image of an individual that did not occur in reality' or (2) 'was produced substantially by technical means, including artificial intelligence or photo editing software, rather than by the ability of another person to physically impersonate the individual.' The amendment also removes the prior blanket requirement that dissemination be done 'with intent to harass, annoy or alarm another' for the base offense, while adding intent-based aggravating factors (harassment, harm, profit, or sexual gratification) that elevate grading to a first-degree misdemeanor; consent of the depicted person remains a defense.
Google (industry)·Voice and likeness·Provider consent policies·Distinct from heygen-avatar-consent-tiers and synthesia-custom-avatar-consent, which both build a third-party consent workflow (webcam statement, pre-recorded video, or waiver) so a customer can clone someone else's voice or likeness. Google's policy is affirmatively scoped to the account holder's own likeness only, and it geofences the feature out of the EEA/UK/Switzerland rather than relying on a uniform worldwide consent standard.
Lets a Google Account holder create a personal AI avatar of their own face and voice for use in Gemini Omni video generation inside Google Vids, gated by an on-device capture and verification step, an 18-and-over age requirement, and a domain-level admin on/off control; every generated clip carries an invisible SynthID watermark. Personal avatars are restricted to the account holder's own likeness, and the feature is withheld entirely from the EEA, Switzerland, and the UK.
ElevenLabs (industry)·Voice·Provider consent policies·Distinct from elevenlabs-voice-clone-policy, which gates who may create a clone in the first place (self-ownership plus verification). This policy instead addresses post-hoc detection and provenance of audio that has already been generated, letting any listener verify a clip's AI origin after the fact.
Embeds an inaudible SynthID digital watermark, built with Google DeepMind, directly into ElevenLabs-generated audio. Paired with a free public ElevenLabs Audio Detector tool so a third party can check whether a given clip was generated by ElevenLabs.
US, New Jersey·Likeness·Laws and bills·Distinct from take-it-down-act (federal, FTC-enforced 48-hour notice-and-removal window) and from ca-deepfake-pornography-act (a standalone civil cause of action under California Civil Code 1708.86) by mandating platform cooperation specifically with nonprofit hash-matching and removal initiatives, rather than creating a fixed federal takedown clock or a new civil claim; its two-track nonconsensual definition (real recording without consent, or 'deceptive audio or visual media' created without consent) is the operative hook bringing AI deepfakes into an otherwise intimate-imagery statute.
Requires a social media company, on request, to cooperate with any nonprofit organization that seeks to remove nonconsensual intimate images or videos, and to immediately remove flagged material pending a review of consent. Section 3.c(2) separately defines an intimate image or video as nonconsensual when the depicted individual 'did not give express consent for the image or video to be created or disclosed' where it is 'deceptive audio or visual media,' defined elsewhere in the bill as a video, image, or audio representation, substantially dependent on technical means rather than physical impersonation, that realistically depicts speech or conduct of a person who did not in fact engage in it, so an AI-generated intimate deepfake falls within the platform-cooperation duty even though the depicted person never actually performed the act.
Authentic Interactions (industry)·Likeness·Industry and technical standards·An open standard specifically for consenting to or refusing AI recreations of a living or deceased person's likeness, distinct from provider-specific tiers like heygen-avatar-consent-tiers by being a portable, platform-agnostic directive rather than a single vendor's intake workflow, and distinct from the Human Consent Standard's machine-readable licensing-declaration model by centering on a personal consent or refusal directive paired with a takedown process.
A portable, lifetime-authored Directive lets a person specify permissions for AI recreations of their likeness; for those without a Directive, decisions follow existing legal frameworks for handling personal remains after death. Registering a refusal is free, requires no relationship with Authentic Interactions, and needs only identity verification; the registry stores no photos, voice data, or biometrics and cannot be searched by face or voice.
The Human Consent Foundation (industry)·Voice and likeness·Industry and technical standards·A machine-readable, declaration-based identity-consent protocol meant for AI systems to query directly, distinct from the Digital Likeness Directive's directive-plus-takedown model and from any single vendor's consent workflow; backed publicly by talent including George Clooney, Meryl Streep, and Tom Hanks and by CAA and the Music Artists Coalition.
An XML and JSON-LD extension to the Really Simple Licensing (RSL) standard that lets a rights holder declare machine-readable AI permissions for identity attributes (name, image, likeness, voice, movement, signature) through an hcs:right element scoped to subject="identity", using permits or prohibits terms so AI systems can programmatically check consent at scale. The draft explicitly states it must not be used to grant or facilitate AI training, generation, reproduction, adaptation, or emulation involving the identity of a minor.
Cartesia (industry)·Voice·Provider consent policies·Distinct from resemble-ai-voice-consent-tos (a verbal-consent standard) and elevenlabs-voice-clone-policy (self-account verification) by pairing a general explicit-consent requirement with a named political-candidate and government-official impersonation ban not present in either tracked policy.
Requires users to submit only their own voice and audio recordings, or those of others with explicit consent, and holds the user solely responsible for ensuring they have all necessary rights and consents for any voice input. The same Acceptable Use Policy separately prohibits using the service to impersonate another person, including celebrities, and bars political uses intended to mislead voters or impersonate political candidates or government officials.
Hume AI (industry)·Voice·Provider consent policies·Distinct from resemble-ai-voice-consent-tos and descript-overdub-consenting-speaker by gating clone creation itself behind a discrete, signed legal-agreement step at upload time, rather than a general terms-of-service representation-and-warranty clause.
Requires a user uploading an audio sample for voice cloning to fill out a legal agreement confirming they hold the necessary rights or consent to upload and clone that specific voice sample; separately, Hume's Acceptable Use Policy prohibits creating voice outputs that replicate another person without clear, affirmative consent.
ElevenLabs (industry)·Voice·Provider consent policies·Distinct from elevenlabs-voice-clone-policy, which gates self-service Professional Voice Cloning behind self-account verification by the voice owner, by creating a separate curated licensing channel for celebrity and historical-figure voices that only the rights holder, not the cloning user, can authorize.
A rights-holder-approval licensing marketplace where only verified talent or their authorized estate or rights-holder representatives can list a voice; a brand submits a request for a specific licensed voice, and the talent or rights holder individually reviews and decides whether to approve that specific proposed use, with no guarantee of approval.
