Since its release in April 2023, the song “Heart on My Sleeve,” featuring R&B singer The Weeknd and rapper Drake, has racked up millions of views and hundreds of thousands of streams. However, neither singer was involved in the creation of the track, which was uploaded by TikTok user Ghostwriter977. Ghostwriter977 claimed the song was made using generative artificial intelligence, which imitates Drake and The Weeknd by analyzing patterns in their performances in order to “sing” like them. Shortly after achieving viral success, “Heart on My Sleeve” was removed from streaming platforms such as Spotify and Apple Music, by request of Universal Music Group (UMG), which represents The Weeknd, Drake, as well as many other prominent musicians.
UMG framed the debate about using AI-generated voices as one that pits “the side of artists, fans and human creative expression” against “the side of deep fakes, fraud and denying artists their due compensation.” Since vocal AI engines are trained on recordings of real people, some argue that using these engines violates copyright and is unethical because the engine uses clips of artists without their consent. Furthermore, AI voices could be used to spread misinformation or misrepresent artists maliciously. The potential of this misrepresentation may jeopardize artists’ ability not only to capitalize off their talents, but also their right to control their public image. Such rights are sometimes protected by laws designed to protect artists’ financial interests, so that their image cannot be commercialized without their consent.
Record labels like UMG also have a financial incentive to worry about AI vocals, since such labels only own the recordings created by the artists, not the rights to the artists’ personas. Thus the artist may permit producers under other companies to utilize their voices without violating contracts with their labels.
On the other hand, the U.S. Copyright Office has ruled that the inherent qualities of a voice are not clearly defined enough to be copyrighted and it is actually the words expressed through them that carry copyrightable ideas. Some producers who use AI vocals might say that their music should be eligible for copyright, since the voices – and the lyrics that they sing – are being transformed or “created” by the producers. Furthermore, so long as songs featuring AI vocals are clearly marked as not being created by the actual person whom the voices are imitating, the threat of misrepresenting the artist may be reduced.
Some AI voice fans might even argue that voice-generative AI engines could benefit artists: AI voices could bring in revenue for artists who have lost the use of their voices through medical issues or simply wish to create more music without having to sing. The AI voices of actors or singers who have passed away could also be used to create new media, with permission from their estates.
–DISCUSSION QUESTIONS–
Has TikTok user Ghostwriter977 done something wrong? If so, is it harm to Universal, Drake and TheWeeknd, the public?
Should we ban all AI-created vocals?
Universal framed the debate as one of artists and fans of human expression against frauds looking to profit off of deep fakes. Is this accurate? How else might we frame the ethical issue?