AI Voice Cloning Laws: The ELVIS Act and the States That Followed (2026)

Seven states now give a person a property right against an unauthorized AI clone of their voice, led by Tennessee's ELVIS Act (Tenn. Code Ann. § 47-25-1101 et seq.), effective July 1, 2024. No federal voice-cloning law exists yet; one, the NO FAKES Act, is pending in Congress.

Information last verified on 2026-08-14. This article has not yet been reviewed by a licensed lawyer.

Jurisdiction scope: This article compares the seven state AI voice-cloning and digital-replica statutes enacted since 2024 (Tennessee, California, Illinois, Montana, Arkansas, Washington, and Pennsylvania) and the pending federal NO FAKES Act. It does not restate Tennessee's ELVIS Act section by section; for that depth, see Tennessee Deepfake Laws. For the broader right-of-publicity framework across all 50 states, including postmortem terms outside the AI context, see Right of Publicity Laws.

In November 2025, the estate of Johnny Cash sued Coca-Cola under Tennessee's ELVIS Act (M.D. Tenn. No. 3:25-cv-01373), the law written to police AI voice clones. The twist: the case does not allege AI at all. It is the clearest illustration yet of how far the ELVIS Act's definition of «voice» reaches, and it sits at the start of a fast-moving, still mostly untested area of law.

The ELVIS Act: How Tennessee Became the Archetype

Tennessee's Ensuring Likeness, Voice, and Image Security (ELVIS) Act took effect July 1, 2024, as a 2024 amendment (Public Chapter 588, from HB 2091) to the state's existing 1984 right-of-publicity statute, Tenn. Code Ann. § 47-25-1101 et seq. It is not a freestanding new law; «ELVIS Act» is now that statute's official short name. The amendment defines «voice» broadly: «a sound in a medium that is readily identifiable and attributable to a particular individual, regardless of whether the sound contains the actual voice or a simulation of the voice of the individual» (§ 47-25-1102). Every individual has a property right in the use of their name, photograph, voice, or likeness «in any medium in any manner» (§ 47-25-1103(a)). The law reaches beyond the person who makes an unauthorized clone: § 47-25-1105(a) also imposes liability on anyone who distributes an algorithm, software, tool, or other technology whose primary purpose is producing an individual's unauthorized voice or likeness, reaching cloning-software vendors directly.

Two provisions of the 2024 amendment go beyond what casual coverage of the ELVIS Act typically describes. First, § 47-25-1106(f) gives a record label or other party holding an exclusive personal-services contract or exclusive distribution license standing to bring an enforcement action in its own right, alongside the artist. Second, the fair-use exemption in § 47-25-1107, covering news, commentary, scholarship, satire, and parody, lost its blanket status: it is now expressly qualified «to the extent protected by the First Amendment», and a defendant loses the exemption if it «had knowledge or reasonably should have known» the use was unauthorized, a constructive-knowledge standard that is a meaningfully lower bar than the prior actual-knowledge test.

The postmortem right lasts a minimum of 10 years after death and continues as long as the voice or likeness stays in commercial use, ending only after two consecutive years of non-use (§ 47-25-1104(b)(2)). That structure is why a long-dead artist's estate, such as Elvis Presley's own, can hold the right indefinitely as long as a catalog stays commercially available.

This is a summary, not the full statute. For the complete section-by-section breakdown, including the Class A misdemeanor criminal exposure and the penalties table, see Tennessee Deepfake Laws.

The States That Followed the ELVIS Act

Six more states enacted their own AI voice or digital-replica law within about two years of the ELVIS Act, though none copy Tennessee's tool-distributor liability provision outright, and one, Pennsylvania, uses an entirely different legal structure. California passed two separate bills on the same day, one protecting deceased performers and one protecting living performers. Illinois amended its existing Right of Publicity Act. Montana, Arkansas, and Washington added AI-specific language to their own publicity or personality-rights statutes. Pennsylvania is the outlier: it created a criminal fraud offense rather than a civil property right.

Pennsylvania's approach is the one genuine structural outlier among the seven. Instead of expanding a civil property right the way Tennessee, California, Illinois, Montana, Arkansas, and Washington all did, Act 35 makes knowingly creating or disseminating an AI voice or image forgery to defraud or injure someone a criminal offense, a first-degree misdemeanor that becomes a third-degree felony in a fraud scheme. A defendant faces prosecution rather than a civil damages claim, and there is no equivalent to the ELVIS Act's tool-distributor liability.

Illinois's HB 4875 was signed by Governor Pritzker on August 9, 2024, as Public Act 103-0836, amending the Illinois Right of Publicity Act (765 ILCS 1075, §§ 5, 20, 30, and 35), effective January 1, 2025. It adds a digital-replica definition covering an AI-generated voice, image, or likeness a reasonable person would believe is genuine, and it extends liability to anyone who materially contributes to, induces, or otherwise facilitates a violation with reason to know of it, reaching distributors as well as direct users. More detail is at Illinois Deepfake Laws on this site.

Montana's HB 513, Arkansas's Act 159, Washington's SSB 5886, and Pennsylvania's Act 35 all trace to the enacted bill text and are covered in more depth on this site's own Arkansas, Washington, Montana, and Pennsylvania deepfake law pages, alongside California's.

Where the Federal Government Stands: The NO FAKES Act

No federal law against AI voice cloning exists yet. The closest bill, the NO FAKES Act of 2026 (S.4591, sponsored by Sen. Chris Coons, introduced May 20, 2026), would create a federal intellectual-property right against an unauthorized digital replica of a person's voice or visual likeness. The Senate Judiciary Committee reported it out with an amendment in the nature of a substitute on June 24, 2026, and it is now on the Senate's Legislative Calendar (Calendar No. 446). It has not passed the Senate, has not been taken up by the House, and is not law.

S.4591 replaces an earlier, stalled version, the NO FAKES Act of 2025 (S.1367), which never advanced past committee referral. Under the official committee summary, the federal right would be licensable but not assignable during a person's life, would not expire at death, and would come with a notice-and-takedown process modeled on copyright law, plus penalties for a knowingly false takedown notice.

The bill's summary states it would «expressly preempt state laws related to digital replicas except for causes of action under state statutes or common law in existence as of January 2, 2025». The ELVIS Act took effect July 1, 2024, several months before that cutoff. As currently drafted, that means a federal NO FAKES Act would not wipe out Tennessee's law, or the similarly timed California AB 2602; it would grandfather laws that predate the cutoff rather than override them. That reading depends on the bill's language staying the same through further amendment, and on the bill actually becoming law, neither of which is guaranteed. The recordinglaw.com explainer on the Senate committee vote covers the bill's mechanics in more depth.

Is This Just for Musicians? Practical Questions Voice-Cloning Law Raises

No. The ELVIS Act's core right belongs to «every individual» under § 47-25-1103(a); it is not limited to musicians, actors, or other public figures. A private person whose voice is cloned without consent, for a scam robocall or a harassment campaign, has the same civil claim in principle that a recording artist does. In practice, the disputes that have generated press attention so far all involve public figures or estates, because those are the conflicts visible enough to draw a lawsuit and coverage, not because the statute itself is limited to them.

Whether an AI-generated cover song or parody is lawful under the ELVIS Act turns on the fair-use exemption in § 47-25-1107. A clearly labeled parody or a satirical use is likely exempt, since the exemption covers comment, criticism, and parody «to the extent protected by the First Amendment». A realistic AI «cover» that is marketed or presented as if it were an authentic, previously unreleased recording by the artist is a different matter: the audiovisual-work carve-out inside that exemption is void if the work creates the false impression that it is an authentic recording in which the individual participated. The line is whether the work misrepresents itself as real, not whether it uses AI.

Has Anyone Actually Been Sued Under the ELVIS Act?

One case, and it is still pending. The John R. Cash Revocable Trust sued Coca-Cola in the U.S. District Court for the Middle District of Tennessee on November 25, 2025 (No. 3:25-cv-01373, before Judge Waverly D. Crenshaw Jr.). Per the complaint itself, the estate alleges Coca-Cola used a Johnny Cash sound-alike singer in a television advertisement in a way that led viewers to believe they were hearing Cash's actual voice, without consent. The complaint, filed primarily under the Lanham Act's false-endorsement provision with the ELVIS Act claim brought alongside it, seeks an injunction pulling the ad plus compensatory damages, Coca-Cola's profits, and attorney's fees, all in amounts to be determined at trial; it does not plead a specific dollar figure. (The complaint's only reference to $75,000 is the boilerplate statement that the amount in controversy exceeds that sum, the minimum needed to be in federal court on diversity grounds; that is not a damages demand.) A CourtListener search for other ELVIS Act litigation, including a direct citation search for § 47-25-1105, turned up no other filed cases as of this writing.

The case is notable for what it is not: an AI case. Its core allegation involves a human performer, not a generative-AI voice clone, pled under a statute written for the AI era. That fits the ELVIS Act's own «voice» definition, which reaches «the actual voice or a simulation of the voice» alike, so a skilled human sound-alike can trigger the same claim an AI clone would. No court has yet ruled on the merits of an ELVIS Act claim, AI or otherwise, so how the statute will actually be applied and interpreted remains untested.

Related Resources

• Tennessee Deepfake Laws covers the ELVIS Act in full statutory depth, including its criminal penalties and fair-use exemptions.
• Right of Publicity Laws covers the general 50-state right-of-publicity framework, including postmortem terms outside the AI context.
• Deepfake & AI Voice Cloning Laws by State is the full 50-state hub for sexual, election, and voice-cloning deepfake law.
• California Deepfake Laws, Illinois Deepfake Laws, Montana Deepfake Laws, Arkansas Deepfake Laws, Washington Deepfake Laws, and Pennsylvania Deepfake Laws cover each state's own AI voice or likeness statute in full.
• NO FAKES Act Advances Out of Senate Committee covers the pending federal bill's committee vote and mechanics in more depth.

---

Last updated: 2026-08-14.