What does the Tennessee ELVIS Act cover?

The Tennessee Ensuring Likeness Voice and Image Security (ELVIS) Act, effective July 1, 2024, expands Tennessee's existing right-of-publicity statute to explicitly cover AI-generated replicas of individuals' voices and likenesses. It prohibits using AI to clone, replicate, or simulate a person's voice or likeness for commercial purposes without their consent. The Act applies to any person — not just celebrities or musicians — and creates a private right of action for violations.

Does the ELVIS Act apply to businesses outside Tennessee?

Tennessee's ELVIS Act can reach businesses outside the state if they produce, distribute, or transmit AI-generated voice or likeness content that affects Tennessee residents or whose unauthorized use involves Tennessee-based individuals. Right-of-publicity laws generally apply based on the domicile of the person whose rights are infringed, not where the business operates. A California-based AI voice company that clones a Nashville musician's voice without consent is exposed under the ELVIS Act regardless of where the company is located.

What penalties can businesses face under the ELVIS Act?

The ELVIS Act creates a civil right of action. Successful plaintiffs can recover actual damages (lost licensing fees, harm to reputation), unjust enrichment from the defendant's unauthorized use, and statutory damages of up to $10,000 per violation — whichever is greater. Knowing violations can result in punitive damages and attorney fee awards. There is no cap on actual damages or unjust enrichment recovery, which means a widely distributed unauthorized AI voice clone of a recognizable artist could generate significant damages.

Does the ELVIS Act restrict AI training on voice data?

The ELVIS Act covers the use, production, and distribution of AI-generated voice replicas without consent — not explicitly the act of training AI models on voice data. However, the practical effect is similar: if training on a person's voice data enables production of an unauthorized clone, the downstream use is still unlawful. Several pending state bills go further and explicitly address training data consent for voice and likeness AI. Businesses should treat voice training data provenance as a legal risk area even where current law focuses on outputs rather than inputs.

Tennessee ELVIS Act: AI Voice Cloning Compliance for Businesses 2026

What the ELVIS Act Actually Does

The Ensuring Likeness Voice and Image Security (ELVIS) Act amended Tennessee's existing Personal Rights Protection Act to explicitly address AI-generated replicas. Before the ELVIS Act, Tennessee's right-of-publicity law protected voice and likeness in general terms — but it wasn't written for a world where AI can clone a person's voice from a few seconds of audio.

The Act adds "voice" as an explicitly protected attribute alongside name, photograph, and likeness. It creates liability for any person who:

Uses or causes the use of an AI system to clone a person's voice
Publishes, performs, distributes, transmits, or otherwise makes available an AI replica
Does so for commercial purposes without written consent of the individual

The law targets the chain of commercialization — not just the person who trains the model, but anyone who distributes or profits from the unauthorized AI voice replica. A platform that hosts user-generated AI voice clones, a marketer who uses a cloned voice in an ad, and the developer who built the cloning tool without consent-verification can all face liability.

Who the Law Protects — and It's Not Just Musicians

The ELVIS name is a deliberate tribute to Elvis Presley's home state, and the law was heavily backed by the music industry. But the statute itself protects any individual — Tennessee resident or not — whose voice is cloned by AI without consent for commercial use.

That means liability exposure applies beyond the obvious music AI scenarios. Voice cloning risks exist in:

High-Risk AI Voice Applications

⚠AI music generation tools that clone artist styles or voices
⚠Synthetic voice assistants trained on named individuals
⚠AI podcast tools that clone host voices for content
⚠Deepfake audio for entertainment or promotional content
⚠AI dubbing tools that replace actors' voices with clones

Moderate-Risk Voice AI Scenarios

⚠Text-to-speech tools that allow custom voice cloning from uploads
⚠Call center AI trained on recordings of real employees
⚠AI voiceover tools that scrape public speaker audio
⚠Customer service bots cloned from recorded support calls
⚠Marketing voice content using cloned spokesperson voices

Consent Is the Only Safe Harbor

The ELVIS Act's core protection is simple: you cannot clone someone's voice with AI for commercial purposes without their written consent. That consent must be specific — a general terms-of-service agreement that mentions "your data may improve our services" is not the same as explicit written consent to voice cloning.

What valid consent looks like under the ELVIS Act:

Explicit voice cloning authorization

The consent form must specifically describe AI voice cloning — what it means, how the clone will be used, and for what purposes. 'We may use your voice for AI training' is better than silence, but the best practice is explicit language: 'We will create an AI replica of your voice and use it to [specific purpose].'

Commercial use scope must be defined

If the AI clone will be used commercially — in ads, paid content, licensed products — that scope must be defined in the consent. Consent for 'personal use features' doesn't transfer to commercial distribution of the clone.

Survivors' rights apply

Tennessee's right-of-publicity law extends to deceased individuals for 10 years after death (extendable by registration). Elvis Presley's estate — and the estates of other artists — can enforce the ELVIS Act against unauthorized AI replicas after death. If your AI voice product uses deceased artists' voices, you need estate authorization.

Consent can be revoked

Individuals can revoke consent to voice cloning. Build a revocation mechanism and a process for removing or retiring voice clones after revocation. Continuing to use an AI voice clone after consent revocation is a fresh violation.

Platform Liability: When Your Tool Enables Others to Clone

One of the most significant aspects of the ELVIS Act for AI companies is that it reaches platforms that enable others to create unauthorized voice clones — not just the person who directly clones the voice. If your tool provides the technical capability for users to clone someone else's voice and distribute the result commercially without the target's consent, you face exposure.

This is different from Section 230 protection, which generally immunizes platforms from liability for user-generated content. The ELVIS Act's liability attaches at the platform level when the platform knowingly enables the specific activity of voice cloning. Providing an AI voice cloning tool without consent-verification mechanisms may not qualify for the same Section 230 pass that a generic user content platform would receive.

Best practice for voice AI platforms: implement a consent verification step before any user can clone a third party's voice. Require uploaders to attest that they have written consent from the voice owner before enabling commercial use of the clone.

The ELVIS Act in the Context of the Growing State AI Voice Law Landscape

Tennessee was the first state to pass a law specifically targeting AI voice cloning, but it won't be the last. As of mid-2026, multiple states have enacted or are advancing similar protections:

Tennessee (ELVIS Act)

In Effect Since July 2024

AI voice and likeness cloning without consent. Private right of action. Covers all individuals. Post-mortem protection 10+ years.

California (AB 2602, AB 1836)

In Effect 2025

AB 2602 requires explicit consent for AI replicas in film/TV. AB 1836 restricts digital replicas of deceased performers. Applies to entertainment contracts and digital performance use.

New York (S7368)

Advancing 2026

Proposed right-of-publicity protections for AI voice and likeness. Would require consent for commercial AI replicas and cover post-mortem rights for public figures.

Federal (NO FAKES Act)

Proposed — Pending

Would create a federal right against AI replicas of voice and likeness without consent. Would override state-level patchwork but has faced lobbying opposition from tech and open-source advocates.

Compliance Checklist for AI Voice Businesses

If You Build AI Voice Cloning Tools

☐Add explicit written consent flow before enabling voice clone creation
☐Require users to attest to ownership or authorization of voice samples
☐Block commercial distribution of clones without verified consent
☐Implement revocation mechanism for consent withdrawal
☐Audit your training data for third-party voice samples without consent
☐Add legal review for any pre-recorded celebrity or artist voice features

If You Use AI Voice in Marketing or Content

☐Verify any AI voice used in ads or content has written consent on file
☐Don't use AI voices that sound like celebrities without licensing
☐Audit voiceover vendors for their consent practices
☐Add a voice-clone prohibition clause to your AI content vendor contracts
☐Train marketing teams on what constitutes an unauthorized AI likeness
☐Get estate authorization for any deceased public figure voice use

Frequently Asked Questions

We're building a voice AI app where users clone their OWN voice. Does the ELVIS Act apply?

If users are cloning their own voice for their own personal use, there's no third-party consent issue. The ELVIS Act focuses on unauthorized commercial use of someone else's voice. Self-cloning for personal use is generally fine. The risk arises if your platform lets users commercially distribute self-clones that could be confused for others, or if the platform itself commercializes user voice data in ways not covered by consent.

We fine-tuned a TTS model on public YouTube videos of various speakers. Are we exposed?

Potentially yes. Using public audio without the speaker's consent for AI training that enables voice cloning is the exact scenario the ELVIS Act targets — particularly if the trained model could reproduce those individuals' voices commercially. 'Public' doesn't mean 'consent to AI cloning.' This is one of the highest-risk training data scenarios in the current legal environment.

Does the ELVIS Act have a fair use or parody exception?

The ELVIS Act includes a limited exception for news, public affairs, sports broadcasts, documentaries, and commentary or criticism. Parody and satire are generally protected under First Amendment principles, though the line between parody and unauthorized commercial use is litigated case-by-case. A clearly labeled satirical AI impersonation is different from using an AI voice clone to sell products.

What should our voice AI terms of service include?

At minimum: explicit prohibition on creating AI clones of third parties without their written consent; user attestation that they have authorization for any voice samples they upload; clear statement that users bear sole liability for unauthorized voice cloning; a mechanism to report unauthorized clones; and a consent revocation flow for users who have provided voice samples. Have a lawyer review — generic TOS may not be sufficient to limit platform liability under the ELVIS Act framework.

Consent Is Infrastructure, Not a Checkbox

The ELVIS Act is already law — it's been enforceable since July 2024. And it's the leading edge of a wave of state and federal legislation targeting AI voice and likeness without consent. The music industry got there first, but the framework protects everyone.

If your business involves generating, distributing, or enabling AI voice content, the question isn't whether consent matters — it's whether your product has the infrastructure to capture, verify, store, and honor it. Building that infrastructure now is far cheaper than defending a lawsuit from an artist whose voice you cloned without asking.