A new track by Drake and The Weeknd is going viral on social media and is bound to top the charts. The song “Heart on my Sleeve” details the emotional trauma of The Weeknd’s breakup with pop star Selena Gomez in a tear jerker that extends just over two minutes. The song has been hailed as the “banger of the weekend” and was viewed more than 8.5 million times on TikTok. The only problem—the song is generated entirely by artificial intelligence and not one bar was written, rapped, or produced by either Drake or The Weeknd.                

Several generative AI models now provide the ability to simulate your favorite artist’s voice and lyrical tempo to create a track that sounds indistinguishable from the real thing. All a user needs to do is record reference vocals, add musical background, and voila – it now sounds like it was written by Drake. Or The Weeknd. Or Kanye West. Or, really, anyone in the music industry that the user chooses. This video provides a great example of just how easy it is to create a song that sounds like it was written by Kanye West when in fact it was created in 20 minutes by a lyrically-challenged Youtuber.

The results are shocking and have significant implications for the music industry. Several artists have voiced their displeasure for these impersonations and music giants are now pushing streaming services to ban music created by AI models with some success (the Drake/Weeknd song has been removed from all streaming services). But legally, how can the artists fight back?

I. Copyright

Copyright protection is unlikely to help artists whose voice and musical style are simulated by AI models. The models are trained by the artist’s vocals but do not reproduce any of those vocals when creating the AI-generated song. There is nothing “copied” by the AI models and therefore no copyright to infringe.

II. Data Privacy

Data privacy may eventually create a path to protection but this will not come easy. The creators of the AI models are unlikely to be contractually bound to restrictions on the use of the voices in songs they own, at least not yet. And state or federal laws are unlikely to restrict use of publicly available songs on privacy grounds.

III. Right of Publicity

The most likely source of legal relief may be found in the artist’s right of publicity. A “right of publicity” can be described as a trademark for a person’s celebrity. “In cases involving confusion over endorsement by a celebrity plaintiff, ‘mark’ means the celebrity’s persona.” White v. Samsung Elecs. Am., Inc., 971 F.2d 1395, 1400 (9th Cir. 1992).

The concept was first created when a company released an advertisement with a robot that was meant to imitate TV star Vanna White. Id. White successfully litigated the dispute and was found to own a right in her “persona”, which was later defined as “the aspect of a person’s character that is displayed to or perceived by others.” Roberts v. Bliss, 229 F. Supp. 3d 240, 249 (S.D.N.Y. 2017). And so the “right of publicity” was born!

Courts later went on to limit the right of publicity to extend only to the celebrity’s name, image, or “a symbol or device such as a visual likeness, vocal imitation, or other uniquely distinguishing characteristic.” Waits v. Frito-Lay, Inc., 978 F.2d 1093, 1110 (9th Cir. 1992). The term likeness “does not include general incidents from a person’s life.” Matthews v. Wozencraft, 15 F.3d 432, 438 (5th Cir. 1994). Nor does it extend to “any attribute of an individual.” Martin v. Living Essentials, LLC, 160 F. Supp. 3d 1042, 1046 (N.D. Ill. 2016), aff’d, 653 Fed. Appx. 482 (7th Cir. 2016). So is a celebrity’s voice protected?

One case appears to be directly on point. The case of Waits v. Frito-Lay, Inc. held a celebrity’s right of publicity extends to “a symbol or device such as a visual likeness, vocal imitation, or other uniquely distinguishing characteristic.” 978 F.2d 1093, 1110 (9th Cir. 1992). In this case, the court held singer Tom Wait’s right of publicity extended to his voice, because “[a] voice is as distinctive and personal as a face.” Id. at 1100.

Here, the Drake/Weeknd/Kanye AI models are creating indistinguishable simulations of the real artist’s voice. They are profiting off the celebrity of these artists by imitating the artist’s voice, even though the model does not reproduce any collected snippets of the artist’s voice. The end result is the type of imitation that Waits holds to be violative of the right of publicity.

Generative AI will certainly alter the arts in a manner never seen before. All may not be lost for the music industry, but look for artists to focus more on their celebrity than their artistic talent when monetizing their skillset. Of course, there’s nothing stopping these artists from using the same AI models to provide cleaner versions of their own songs or avoid the costs of recording an album from scratch. If a Youtuber can create a song in 20 minutes, why couldn’t an established musician?