INTRODUCTION
Generative artificial intelligence can now recreate a person's voice, face, and mannerisms with unsettling accuracy including the voices and faces of people who are no longer alive. Podcasts have used AI to script and perform new material “in the style of” deceased comedians. Fans have used voice-cloning tools to produce new songs in a departed artist's timbre. Families have experimented with AI chatbots trained on a late relative's text messages. This practice, sometimes called AI “resurrection,” sits at the intersection of intellectual property, the right of publicity, privacy law, and the law of estates and the legal system is only beginning to catch up.
BODY
The clearest recent illustration is the dispute over comedian George Carlin. In January 2024, Carlin's estate sued the creators of the Dudesy podcast after they released an hour-long special, generated with AI, that mimicked Carlin's voice and comedic style[1]. The estate argued the special amounted to unauthorized exploitation of Carlin's likeness and voice for commercial gain. The case ended up settling about two months later the podcast creators agreed to pull the special down and accept a permanent injunction, meaning they can never use Carlin's voice, image, or likeness again without getting written consent from his estate first. Good outcome for the family, I'd say, but kind of a letdown from a law-nerd perspective, because a settlement means we never actually got a court to rule on the merits. So the real legal question underneath all of this is still just out there, unresolved.
And that question, annoyingly, depends entirely on state right-of-publicity law, which is all over the place. Only about half the states even recognize a postmortem right of publicity in the first place basically, the idea that an estate gets to control commercial use of a dead person's name, voice, or likeness and even in the states that do recognize it, how long it lasts and how far it stretches varies a ton. Part of the issue is these statutes are old. They were written back when the biggest concern was someone slapping a dead celebrity's photo on a t-shirt, not a machine generating an entirely new, fake voice recording from scratch. So there's this pretty obvious gap where AI-made "digital replicas" just don't fit cleanly into laws that were never built to think about them.
Tennessee moved first to close that gap. Its Ensuring Likeness Voice and Image Security (ELVIS) Act, signed on March 21, 2024, and effective July 1, 2024, amends the state's 1984 Personal Rights Protection Act to add “voice” defined broadly enough to cover an AI-simulated voice alongside name, photograph, and likeness[2]. It was described as the first law in the country written specifically to address AI voice cloning and audio deepfakes[3]. Notably, the Act does not stop at the person who makes the unauthorized clone: it also creates civil liability for anyone who knowingly publishes or distributes such content, and even for companies that distribute a tool whose primary purpose is producing an identifiable person's simulated voice or likeness language that could reach AI developers and hosting platforms, not just end users[4]. Protection under the Act extends to deceased individuals as well as the living, folding directly into existing postmortem publicity rights.
Other states are following suit, with Illinois, Kentucky, and others introducing comparable bills, and a federal proposal the DEEPFAKES Accountability Act still pending in Congress. Until Congress actually gets around to passing something federal (if that ever happens), we're basically stuck with a patchwork. The same exact AI-generated recreation of a dead performer could be flat-out illegal if it happens in Tennessee, but totally fine or at least in a weird gray zone in a state that never bothered to recognize a postmortem right of publicity in the first place. Which, when you sit with it, is kind of a wild result: the legality of the exact same conduct depends entirely on which state line you happen to be on.
It gets messier too, because U.S. copyright law just doesn't have the concept of "moral rights" the way a lot of other countries do that personal, almost emotional connection a creator is supposed to have to their own work. So if a family is upset about how their late relative's persona is being used, they usually can't just point to copyright and say "that's not okay." They don't have that option. Instead they're stuck picking through whatever publicity, privacy, or defamation theories their particular state happens to recognize and as we've seen, that's a pretty hit-or-miss list depending on where you live.
CONCLUSION
So, what's really protecting you at this moment? To be honest, it's probably not a law or a contract. From my understanding, lawyers who specialize in estate planning and entertainment are now advising their clients to take matters into their own hands. They're telling them to clearly state their AI-likeness rights in their wills, trusts, and licensing agreements, rather than relying on existing publicity laws to cover it. When you think about it, this approach makes sense - why would you leave something as important as this to a patchwork of old laws that weren't even written with AI in mind, when you could just explicitly state your wishes? It's like, why take the risk and hope that the current laws will somehow magically apply to this new technology, when you could just be clear and direct about what you want to happen.
To be honest, I don't think this will remain a small issue in estate planning for much longer. The technology for cloning voices and synthesizing videos is becoming more affordable and realistic almost every month. As a result, the question of who actually controls a person's digital presence after they're gone is going to become a big problem that everyone will be talking about, not just lawyers who work with celebrities. At some point, governments outside of Tennessee will have to deal with this issue and make some decisions, because they won't be able to just ignore it and let the courts figure it out. They'll be under a lot of pressure to take action and come up with some rules, rather than just waiting to see what happens in court.
Reference
[1]Main Sequence, Ltd. v. Dudesy, LLC, No. 24-00711 (C.D. Cal. filed Jan. 25, 2024); discussed in “The King is Back (in the Digital Era): The ELVIS Act, Generative AI and Right of Publicity,” Proskauer Rose LLP, proskauer.com/blog (2024).
[2]Ensuring Likeness Voice and Image Security Act, Tenn. Code Ann. § 47-25-1101 et seq. (2024); “The ELVIS Act: Tennessee Shakes Up Its Right of Publicity Law and Takes On Generative AI,” Latham & Watkins LLP, lw.com (Apr. 2024).
[3]“First-of-Its-Kind Artificial Intelligence Law Addresses Deep Fakes and Voice Clones,” Holland & Knight LLP, hklaw.com (2024); Tennessee Governor's Office, “Tennessee First in the Nation to Address AI Impact on Music Industry,” tn.gov (Jan. 10, 2024).
[4]“The ELVIS Act: Tennessee Shakes Up Its Right of Publicity Law and Takes On Generative AI,” Latham & Watkins LLP, lw.com (Apr. 2024) (summarizing secondary-liability provisions).