Elvis Inspires First State AI Protections for Musicians’ Voices

Elvis impersonators have been ubiquitous for decades, but only when artificial intelligence gave the masses tools to replicate his iconic sound did his home state change its laws to explicitly protect his voice.

Tennessee’s Ensuring Likeness, Voice, and Image Security Act—ELVIS Act—was enacted March 21 and codifies a right to control commercial exploitation of one’s own voice if it’s recognizable and attributable to a particular person. The law came largely in response to a viral AI-made song mimicking the voices of rappers Drake and The Weeknd in spring 2023, which alarmed the music industry and raised a host of legal questions.

VIDEO: Can Laws Keep Up With AI?

The law, which went into effect July 1, could provide a sweeping template as industries and lawmakers confront the metastasizing capabilities of AI. While many new or expanded trademark, copyright, or publicity rights tend to incorporate tension with the First Amendment, the ELVIS Act’s extensive reach seems particularly set to test the bounds of free speech.

It went beyond barring exploitation of celebrities’ voices, expanding the reach of Tennessee’s prior right of publicity law—which was also pushed to protect Elvis’ estate but had been limited to advertising. It also creates liability for anyone who “makes available” tools whose “primary purpose” is replicating an individual’s voice or image without authorization.

“It’s written incredibly broadly. It’s hard to know what the limits on the right might be,” IP law professor Stacey Dogan of Boston University said. The language seems to rope in uses that are “more expressive than exploitative,” and it “appears to reach” tribute bands and other imitations of performers, he said.

Dogan added that he recognized the concern among artists that “AI could support actual performances,” which has led to the push for states to “take a stab at trying to prevent that.” But the ELVIS Act, she said, leaves most boundary-setting to the First Amendment and the courts.

“It’s not clear whether this state law is going to be interpreted by the court to have any internal limits,” she said.

Voicing Concern

The commercial value of mimicking voices isn’t new, from impersonators to cover band singers to sound-alike voices in ads.

As in other areas of IP and the broader economy, though, the power of AI has proven to be a game-changer. A January lawsuit against Will Sasso and a fellow podcaster targeted their AI-generated George Carlin routine—allegedly after the duo previously produced a routine featuring an AI-generated Tom Brady.

“Like with digital technology a generation ago, the fundamental difference is it makes these types of uses much easier and at a much higher quality,” IP law professor Jeremy Sheff said. “It puts sound reproductions in the grasp of just about anybody.”

That digital revolution hit the music industry hard, with labels’ revenues only recently recovering to pre-internet levels—and only achieving that by “vastly reducing the fractional share of artists,” Sheff said.

Elvis Presley’s estate provided the driving force behind Tennessee’s 1984 codification of the right of publicity, which extended control of “name, image and likeness” to a person’s heirs, if they claim it within 10 years of an individual’s death. Tennessee’s rights are potentially the most extensive in the nation; while Indiana and Oklahoma grant post-mortem rights lasting 100 years, Tennessee’s are perpetual unless heirs stop exploiting it for two years after the initial decade of protection.

Existing statutes or common law may have allowed for rightsholders of Elvis’ persona to block exploitation of AI versions of his voice—even if “likeness” isn’t interpreted to include voice.

A California law that included protection of “voice” was found in 1988 by the US Court of Appeals for the Ninth Circuit to be “of no aid” to actress and singer Bette Midler’s suit against Ford Motor Co. over its use of a sound-alike voice in a television ad. But the Ninth Circuit found misappropriation of her voice to be a common law tort. Four years later, it confirmed a jury’s finding on a similar claim by singer Tom Waits against Frito-Lay Inc. while also affirming a false endorsement verdict under federal law.

After the AI-generated song “Heart on My Sleeve” mimicking Drake and The Weeknd racked up millions of streams on music platforms last spring, Tennessee moved to codify artist protections. In January the governor announced the bill that would ultimately be enacted as the ELVIS Act.

“I think everyone on the creative side is going to be happy with it,” entertainment law professor John Simson of American University said of the legislation.

The law protects individuals’ “name, photograph, voice, or likeness in any medium” in advertising without consent, defining “voice” as one attributable to a particular individual even if it’s a “simulation.” It also creates liability for anyone who “publishes, performs, distributes, transmits or otherwise makes available to the public an individual’s voice or likeness” with knowledge that it was not authorized.

Going Further

Dogan noted that while the law specifically applies to “commercial exploitation” of a famous voice or likeness, it doesn’t explicitly limit the right to the advertising context as the previous version did and other states’ rights of publicity laws do. It keeps a carve-out for fair use in “connection with any news, public affairs or sports broadcast or account,” but added a new qualifier, “to the extent protected by the First Amendment.”

“Clearly you’re not going to stop Jimmy Fallon from impersonating someone,” Simson said, and “no one’s going to care” about tribute bands clearly identified as not associated with the original act.

Courts should still be able to easily get rid of obviously abusive cases, Dogan said, while noting that won’t necessarily preclude all frivolous litigation.

“Courts sometimes struggle with deciding First Amendment issues early in litigation, so it could make it easy for rightsholders to send cease-and-desist letters and make people back down because they don’t want to spend the money,” she said.

The ELVIS Act goes further than other laws by creating liability for making available “an algorithm, software, tool, or other technology, service, or device, the primary purpose or function of which is the production of an individual’s photograph, voice, or likeness ” without permission.

“I’ve not seen any effort by states to regulate the distribution of software as a means to protect the rights of publicity,” Sheff said. “There are First Amendment obstacles to” barring access to software.

He noted that the Second Circuit in 2001 shut down a free speech defense for code alleged to circumvent digital copyright measures meant to stop the copying of DVDs in Universal Studios v. Corley. The Supreme Court, meanwhile, declined to assess liability for making available another copying device—the VCR—in its 1984 decision in Sony v. Universal City Studios. Sheff said a constitutional analysis will depend on the nature of the particular law and software.

“Like any other tech, these things are tools that can be used for good or for ill,” Sheff said.

There are examples of permission being given for AI to mimic famous celebrities’ voices. Actor James Earl Jones licensed the rights to his archive for the purpose of replicating his iconic Darth Vader voice for Disney+ miniseries “Obi-Wan Kenobi,” entertainment attorney Simon Pulman of Pryor Cashman LLP noted. AI also reproduced Val Kilmer’s voice for 2022’s “Top Gun: Maverick,” as the actor lost the ability to speak years earlier after undergoing treatment for throat cancer.

The technology will “create a lot of work for the First Amendment and fair use of attorneys,” and, given the pure volume of content, courts will predictably find some infringement, Pulman said.

There’s “tension between short-term profitability and long-term consequences” as the entertainment industry charts a course for using the tools, he said. The “huge potential for abuse” of voice-replicating technology represents just a fraction of AI’s potential downside given all the other harms stemming from copyright theft, fraud, and deepfakes.

“This has gone from theoretical and science fiction fear to a very real and imminent perceived danger in a very short period of time,” Pullman said.