By Laura T. Geyer and Andrew J. Costa
As a technology, general access to generative AI is now so comprehensive that efforts to regulate its various forms are horses long gone as the barn burns swiftly to the ground. Fortunately for those with specific concerns, some aspects of AI abuse fall squarely within existing law, while other aspects demand expansion of law to explicitly cover what slips through existing rubrics. Laws regarding “personal rights” including the unauthorized uses of a person’s name, image, and likeness are embodied in state law, and so, naturally reflect the primary policy concerns of those states. Expansions frequently arise from a state’s specific interest communities’ responses to existential challenges, and so it is no surprise that the first act directed at the human voice would arise from Tennessee, one of the USA’s greatest creative musical centers. Tennessee’s ELVIS (Ensuring Likeness Voice and Image Security) Act, which goes into effect July 1, 2024 has been billed as the first state law specifically written to protect musicians from the unauthorized use of their voices through audio deepfakes and voice cloning – although its reach certainly extends beyond musicians.
The ELVIS Act expands Tennessee’s already robust structure (itself enacted in 1984’s “Personal Rights Protection Act” (PPRA)) in response to demands from the Elvis Presley estate) for allowing individuals (and their estates) to take civil action to protect unauthorized uses of their name, image, and likeness by: (a) introducing new types of prohibited/covered material and conduct, in particular with regard to unauthorized (including AI-generated) use of the human voice; (b) expanding claims against such unauthorized use to allow third-party recording companies to make them; (c) expanding and adding to existing civil and criminal penalties including making it easier to make claims against third party hosts for third-party content; and (d) narrowing and clarifying “fair use” and like exceptions which might otherwise allow such use. In this article, we take a deep dive into the ELVIS Act by first analyzing the changes it makes to existing name, image, likeness (and now voice) law, before we address some new or lingering open-ended questions that in our view would be most appropriately remedied by a federal statutory scheme for protection of rights of publicity and privacy including the name, image, likeness, and voice rights that comprise the new Act.
Adding Voice to the Mix
The ELVIS Act expands existing Tennessee law (§§ 47-25-1101, et seq.) protecting use of a person’s name, image, and likeness to include a person’s voice. “Voice” is defined broadly as “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.” This expansion is a natural extension of existing laws, the heart of which is preventing people from doing anything that identifies another person without their consent – for many performing musicians, voice is as key an identifier as a photograph or name. Nobody familiar with the work of Bob Dylan would fail to instantly recognize that gravelly voice, and, as with use of his photo in, say, an advertisement, would assume his approval or sponsorship of the subject of the advertising. Moiya McTier of the “Human Artistry Campaign” said of the Act: “Fittingly named after one of the world’s most iconic voices, the ELVIS Act marks a history-defining moment – protecting us all from irresponsible and unethical AI…. The life’s work and irreplaceable contributions of the creative community to our culture deserve safeguards that allow AI technology to be used responsibly without violating anyone’s rights or appropriating their art.”
While it enjoyed overwhelming music industry support from powerhouse organizations like the RIAA, SAG-AFTRA, and the Songwriters Association of America, it did not meet universal acclaim. The Motion Picture Association raised concerns that the law as expanded could fatally limit filmmakers’ ability to portray living or (recently departed) people and events – for example in documentaries, especially of more controversial figures. Broader concerns raised by others included worries that it could put cover bands out of business and impair the careers of soundalike voice artists. Lawmakers made clear in debates, hearings, and related proceedings as the law moved through the legislature that these were not intended targets of the law. Once a law is enacted, legislative history (particularly in hearings) becomes relevant to interpretation of a law, but so long as the plain language is potentially problematic for these groups, it remains to be seen how the law will be enforced.
Liability/Exposure
Name, Image, Likeness, and now Voice Rights (or colloquially NIL+V) are generally limited in time for periods that vary from state to state. At one extreme, in some states, once a person is dead, any causes of action they might have made evaporate instantly. Others have very long periods, like California’s 70 years or Oklahoma’s 100 years. Tennessee is a standout in having no limitation on the duration of these rights so long as the estate continually exploits the celebrity’s attributes. This means that so long as anyone is managing it, the Elvis Presley Estate will be able to enforce his personal rights in his name, image, likeness, and voice well into the future. So at least from the NIL+V standpoint, Elvis is absolutely alive and well! Differing state NIL+V laws however mean that enforcement is complicated as it’s possible to pursue offenders in multiple forums.
The Act further extends liability beyond a direct offender to parties at all points in the chain by which an offending imitation reaches anyone (including the sort of third-party platforms which would, to the extent a use would otherwise violate U.S. Copyright law, be entitled to rely on the safe harbor/take down rubric of the Digital Millennium Copyright Act). These include:
- any party who “[p]ublishes, performs, distributes, transmits, or otherwise makes available to the public an individual’s voice or likeness, with knowledge that use of the voice or likeness was not authorized by the individual”;
- anyone who “[d]istributes, transmits, or otherwise makes available an algorithm, software, tool, or other technology, service, or device, the primary purpose or function” of which “is the production of a particular, identifiable individual’s photograph, voice, or likeness, with knowledge” that the use “was not authorized by the individual”.
You can see that generative AI developers (like OpenAI and Google) and those who host and/or publish unauthorized AI-generated content third-party content (like TikTok or YouTube) alike are in the crosshairs of the Act. Whether such parties will (or can) have the degree of knowledge requisite to be liable under the Act, however, remains to be seen.
New Claimants
Prior to the Act’s enactment, only an affected individual person could make a claim under state law for misuse of their own name, image, or likeness. Now under the ELVIS Act, such claims may include AI-generated or real-life imitations of a person’s voice. The ELVIS act has also added a new group of claimants: third-party record companies may bring an action on behalf of one of their exclusive licensees either as recording artist or as licensee for distribution of sound recordings featuring the individual’s voice.
Stable Remedies
The ELVIS Act leaves most existing remedies untouched. Injunctive relief to prevent or restrain the unauthorized use of their name, photograph, voice, or likeness remains available, as well as impounding or destruction of all materials made or used in violation of their publicity rights along with recovery of actual damages and any additional profits attributable to that infringement. The law also retains a highly unusual Tennessee criminal law which makes some violations of the law (as expanded) a Class A misdemeanor carrying penalties of up to a year in jail and/or fines up to $2,500.00. Criminal liability for anything can be a strong deterrent, and even though such prosecutions would likely be rare, the possibility could have a chilling effect on parties making arguably fair use of covered material, to the detriment of creative endeavor.
Narrowed Exemptions
In enacting this new law, the state of Tennessee also took the opportunity to curtail existing exemptions.
First, before ELVIS, Tennessee’s advertising exemption provided owners or employers of advertising media such as newspapers, magazines, and radio stations with a safe harbor for unauthorized use unless they had explicit knowledge that their use of NIL was unauthorized. This kind of “explicit knowledge” standard protects innocent offenders – at least as to use under state law – where the offender didn’t know of the misuse. The new standard is significantly more burdensome. Now, liability attaches where an entity or person had “knowledge or reasonably should have known of the unauthorized use.” This means innocent offenders won’t be protected if they simply didn’t know their use was unauthorized; said another way, owners and employers of media now have, to the extent reasonable, a duty to make sure they have received authorization to use NIL+V content, including if that consent comes (as it usually will) through a third party (as in the case of advertising companies). This kind of standard also arguably means that even the simple expedient of requiring a content provider to represent that it was not in violation of any third party rights would not be sufficient to protect it from liability.
Second, prior to ELVIS, existing Tennessee law set forth a broad exemption for uses of NIL in connection with “news, public affairs, or sports broadcast or account”, even when such uses otherwise would have been impermissible under the prior law. Now that exemption has been amended to apply only to the extent “such use is protected by the First Amendment to the United States Constitution.” In some ways this is not news. Then as now, any action under state law that would represent a violation of the First Amendment would be void as unconstitutional. However, by eliminating a blanket exemption and incorporating the First Amendment as the lens through which to determine liability, Tennessee is, in effect, importing an enormous (and often convoluted) body of First Amendment jurisprudence directly into a Tennessee lawsuit – whether that case involves a question of federal law, or not, and regardless of whether that case involves a so-called “state actor”. This legislative choice will no doubt further complicate litigation under this section while requiring state courts to grapple with arguably novel questions of First Amendment law more properly suited for a different forum.
Lingering Concerns and Implications
Jurisdiction: The history and enactment of the law take jurisdiction as a given, likely because claims based on personal rights such as name, image, and likeness typically involve conduct within the relevant state. However, it’s unclear how the Act will affect jurisdictional questions with respect to out of state defendants, and as between state and federal courts in Tennessee. For example, state “long arm statutes” generally allow courts within a state to have jurisdiction over out-of-state defendants if their conduct was directed towards the state at issue, or causes harm felt within that state. It’s unclear the extent to which out-of-state defendants – like tech companies in Silicon Valley – may be subject to liability in Tennessee under the ELVIS Act for conduct arguably directed towards the whole of the United States (rather than Tennessee specifically). While the Act is silent on this, presumably state common law should govern this determination. Similarly, lawsuits pursuant to the Act can also involve questions of both state and federal law (like copyright infringement which is always federal, and federal trademark infringement), and thus raise the issue of whether state or federal court is the proper forum. In mixed cases such as this, it seems likely that Tennessee federal courts would be the proper venue for suits with embedded ELVIS Act claims; but this is anything but clear. Should the U.S. Congress further enact a federal right of publicity scheme, as many proposed bills are meandering through Congress, its uncertain whether ELVIS Act claims would be dragged along into federal court, be preempted, or would need to be brought in parallel in a state court.
Pre-emption: To the extent a voice imitation would infringe copyright material, under the doctrine of pre-emption, the Copyright Act could pre-empt state claims that rest on the same facts. Whether preemptions will inevitably displace ELVIS claims with mixed questions of copyright law remains to be seen.
Fair Use: As noted above, certain exemptions under the Act have been replaced by a First Amendment “fair use” standard. In any court, to the extent that state action conflicts with the First Amendment, that can be the basis of an objection. But building that rubric into state law specifically seems likely to raise the burden for a target alleged offender to require not just state law expertise, but also significant constitutional law understanding.
“Knew or Should Have Known”: The revised law’s new “knew or should have known” standard for liability for a third-party host/carrier of objectionable content may require that such parties (whether YouTube or a local radio station) conduct or have in place mechanisms to represent some kind of diligence to ensure that a use is authorized. What will a party “ought to have known”? What must they ask for from content providers – proof of a license? Again, this remains to be seen.
Feelin’ “All Shook Up”?
If you’ve followed us this far and wonder “what does this all mean for me and my clients?”, as has become usual with all things AI, a few key principles jump out:
- If you’re using the voice or imitating the voice of an existing person, simply get their consent beforehand. And what’s more, don’t assume that just because your use is colloquially “non-commercial” that it’s permissible – for example, using an AI to impersonate Elvis singing Gangsta’s Paradise on TikTok (not to mention that it would also violate relevant Copyright Act sections) is still considered “commercial”.
- Make sure your creatives or marketing people are aware that using AI to make a soundalike or imitation for distribution of any kind is the opposite of a defense to violation of name, image, and likeness laws; and if they are going to do so, again, get all consents/permissions beforehand.
- Watch the law – NIL+V laws are coming into effect in other states and Federal laws are in the works which might require revision of compliance standards.
Finally, even if you operate outside of Tennessee, assume the ELVIS Act applies, especially if you operate in the music industry. Companies need to pay attention to laws in states with industry epicenters, and this is no less the case for California for film as it is Tennessee for music. Further, if the ELVIS Act is a precursor to a federal scheme, early compliance simply gets you ahead of the curve.