The ELVIS Act: Tennessee Makes 'Voice' a Protected Property Right Against AI Cloning

This is a legislation post: it analyzes an enacted statute, not a court decision. On March 21, 2024, Tennessee Governor Bill Lee signed the Ensuring Likeness, Voice, and Image Security Act of 2024 — the “ELVIS Act” — which took effect July 1, 2024. Enacted as House Bill 2091 (Rep. William Lamberth) and Senate Bill 2096 (Sen. Jack Johnson) and passed unanimously (93-0 in the House, 30-0 in the Senate), the law amends Tennessee’s Personal Rights Protection Act of 1984, codified at Tenn. Code Ann. §§ 47-25-1101 et seq., and adds criminal provisions in Title 39. Its signature move is to make an individual’s voice an expressly protected attribute alongside name, photograph, and likeness — and to reach not just those who publish an unauthorized AI clone but those who distribute the tools designed to make one. It is the first statute of its kind in the United States, and Nashville’s music industry built it.

At a glance

• Statute: Ensuring Likeness, Voice, and Image Security Act of 2024 (ELVIS Act), HB 2091 / SB 2096
• Signed: March 21, 2024, by Gov. Bill Lee; effective: July 1, 2024
• Amends: Tenn. Code Ann. §§ 47-25-1101 et seq. (Personal Rights Protection Act of 1984), plus criminal provisions in Tenn. Code Ann. Title 39, ch. 14
• Key addition: “voice” becomes an expressly protected attribute, defined to include simulated or AI-generated reproductions readily identifiable as a particular individual
• Tool liability: reaches anyone who distributes an algorithm, software, or tool whose primary purpose is producing an unauthorized replica of a person’s voice or likeness
• Remedies: civil action with injunctive relief and damages; criminal violation as a Class A misdemeanor

What the law actually changed

Tennessee’s 1984 Personal Rights Protection Act — itself a product of litigation over Elvis Presley’s estate — already protected the use of a person’s name, photograph, and likeness, and made that right descendible after death. The ELVIS Act keeps that architecture but adds three things.

First, it elevates voice to a named, standalone protected attribute and defines it broadly enough to capture a “simulation” of a person’s voice — the precise output of a voice-cloning model. Before the amendment, a plaintiff trying to stop an AI voice clone had to argue that a synthetic voice counted as a “likeness,” a strained fit. The Act removes that ambiguity.

Second, it confronts the AI supply chain. The most significant doctrinal innovation is liability not only for the person who publishes an unauthorized replica, but for anyone who makes available a “technology, service, or device” — an algorithm, software, or tool — the “primary purpose or function” of which is producing an unauthorized simulation of an individual’s voice or likeness. That reaches developers and distributors of voice-cloning products, not just end users, a meaningfully broader target than traditional publicity law.

Third, it adds a criminal dimension. A violation can be prosecuted as a Class A misdemeanor, giving the state, not just private plaintiffs, a role in enforcement.

The First Amendment guardrails

A statute that restricts the use of a person’s voice and likeness inevitably collides with the First Amendment, and the drafters knew it. The Act preserves exceptions modeled on fair-use and newsworthiness principles: uses in connection with news, public affairs, sports broadcasts, and uses that constitute fair use — including comment, criticism, scholarship, satire, or parody — are not actionable. The statute also generally exempts representations in audiovisual works and advertising where the use is otherwise protected expression.

The tension is real and unsettled. The tool-liability provision in particular invites a First Amendment challenge: a general-purpose voice synthesizer has substantial protected uses, and the line between a tool whose “primary purpose” is unauthorized cloning and one merely capable of it is the kind of distinction that has tripped up secondary-liability regimes before. Industry groups, including TechNet, warned during drafting that the language was broad. Whether courts read the “primary purpose” limiter narrowly enough to survive scrutiny is a question the early enforcement actions will test.

How it compares to the publicity baseline

The ELVIS Act should be read against, not in place of, ordinary right-of-publicity law. California’s statutory right of publicity (Cal. Civ. Code § 3344) and its common-law analog already reach voice misappropriation in some circumstances — the classic sound-alike cases established that decades before generative AI. What Tennessee did was make the protection explicit, descendible, and aimed squarely at the technology rather than only at the publication. The descendibility point matters enormously in a state whose marquee assets are the personae of deceased recording artists; an heir or estate can invoke the Act, not merely a living performer.

The Act is also a template. Its passage accelerated parallel efforts elsewhere — California’s AB 1836 and AB 2602, Illinois amendments, and the proposed federal NO FAKES Act all share the ELVIS Act’s core instinct that voice and digital replicas need express statutory treatment. Tennessee got there first, which means its drafting choices, and any judicial gloss on them, will influence how the rest of the country frames the problem.

Open questions

The statute is young and largely untested in court, so several questions are genuinely unresolved. How will courts construe “primary purpose or function” for tool liability — will it require a product designed for infringement, or merely one predictably used that way? How does the Act interact with federal copyright preemption when the cloned voice appears in a fixed sound recording? Will the broad “voice” definition sweep in tribute performers, impressionists, and sound-alikes who have long operated under the older publicity regime? How will Tennessee courts reconcile the fair-use and parody carve-outs with the tool-liability provision, which targets the instrument rather than any particular expressive output? And because the right is descendible, how far back does it reach — can the estate of a long-deceased artist invoke it against historical-style AI recreations?

Implications

• For AI voice developers: Distributing a product whose primary purpose is cloning identifiable voices now carries direct exposure in Tennessee — including criminal exposure — independent of what any user does with it.
• For artists and estates: Voice is now an express, descendible property interest in Tennessee. Estates of deceased performers have a concrete statutory hook against AI recreations.
• For platforms: Hosting or facilitating voice-cloning tools targeted at identifiable people is a new risk vector; “primary purpose” framing makes the design and marketing of a tool relevant evidence.
• For the national landscape: As the first enacted law of its kind, the ELVIS Act is the de facto reference text other legislatures and the drafters of the NO FAKES Act are working from.

Frequently asked questions

Is the ELVIS Act a court ruling? No. It is an enacted Tennessee statute — legislation, not litigation. It amends the existing Personal Rights Protection Act and creates new civil and criminal liability; how courts will apply it is still developing.

Does it ban AI voice tools outright? No. It targets tools whose “primary purpose or function” is producing unauthorized replicas of a person’s voice or likeness, and it preserves exceptions for news, commentary, criticism, parody, and other fair uses. General-purpose tools with substantial legitimate uses are not automatically covered, though the boundary is untested.

Does it protect ordinary people or only celebrities? It protects every individual’s voice and likeness, not just the famous. As a practical matter the highest-value claims will come from recording artists and public figures, but the statutory right is not limited to them.

Authorities and sources

• ELVIS Act overview and legislative history: ELVIS Act (Wikipedia); Gov. Lee signing announcement (TN.gov).
• Statutory text and redline: Personal Rights Protection Act redline showing ELVIS Act changes (Rothman’s Roadmap, PDF); Tenn. Code Ann. tit. 47, ch. 25, pt. 11 (Justia).
• Analysis: Latham & Watkins, “The ELVIS Act” (PDF); Manatt client alert; Skadden, “Tennessee Law Addresses Proliferation of Deepfakes”.