On September 25th, 2025, Spotify released a statement, committing to removing AI deception and impersonation from their platform.  This includes “vocal deepfakes,” where AI software impersonates an artist’s voice to be used in another producer’s song. Notably, Spotify doesn’t declare war on producers that use AI (without copying others) to make their own music.  

Vocal deepfakes have been seeping into streaming platforms for awhile. This past July, a song called “Heart on My Sleeve” was released, clearly mimicking The Weeknd and Drake. Before it got taken down, it was streamed millions of times.

Most music platforms don’t actually do something about AI use in music. Tidal, a streaming platform known for its high pay per stream, has just teased its own AI tools for its built-in upload function. AI-generated bands pervade platforms like YouTube and Amazon Music with little oversight. In fact, the algorithm ends up recommending them. 

It brings up an interesting question - can we apply any current copyright laws to AI production? Additionally, can we prevent AI deepfakes of other’s singing voices? 

AI Producers & Music Copyright

Under 17 U.S.C. § 114-115, amended by the Music Modernization Act of 2018 (MMA), digital streaming services require licenses for three types of rights: public performance, music composition reproduction (referred to as “mechanical”), and sound recording digital performance. These rights can belong to different people. A songwriter can receive the mechanical rights associated with writing the composition, but the band playing the song can hold the performance rights that give them access to the master recording. A record label can also own the copyright to the master recording separate from the person who made the song. 

Does producing the song with an AI give that software the mechanical rights to your music? No. This precedent comes from a relatively recent case discussed in a previous article I’ve published regarding AI art & piracy. The software’s creator also can’t claim your music - otherwise, it’d spell doom for everyone using any type of software to create music. 

Article I, Section 8, Clause 8 of the U.S. Constitution, known as the Intellectual Property Clause, is interpreted to require an original work to receive copyright protection. The Supreme Court goes further in the case Feist Publications Inc. v. Rural Tel. Serv. Co., 499 U.S. 340 (1991). Delivering the majority opinion, Justice O’Connor cites two main components of originality: work created independently (without copying from others) and a minimal level of creativity. 499 U.S. 358. The amount of effort you put in isn’t relevant.

This can go two ways. You can have a “black box” dilemma: since the individual doesn’t know exactly how the AI processes information to make a song, there isn’t a minimal level of creativity to be met (see this law review article by Zach Naqvi at pages 39-40). Alternatively, you can view AI as a tool to recognize the author’s vision, which would satisfy both requirements for originality. 

How Copyright Law (Doesn't) Combat Deepfakes

Another problem is that your voice isn’t what’s being copyrighted. 

You can file a copyright claim if the AI composition includes copyrighted material, which would require directly investigating the AI software. For AI-generated pictures, you may have an easier time with that. For music, as evidenced by the Ed Sheeran case and many more, you have to prove the infringement goes beyond the fundamental building blocks. That means similarities can be ignored if it’s reasonably certain someone else could come up with a similar pattern of notes and lyrics independently. 

However, it’s possible to seek remedies for copying one’s voice through the right of publicity. The right of publicity is a doctrine that prevents individuals from “unauthorized commercial use of an individual’s name, likeness, or other recognizable aspects of one’s persona,” according to the Legal Information Institute. Yes, this is now a crossover episode with NIL rights. 

The right of publicity is only recognized at common law or by statute at the state level. That means the court systems of each respective state recognizes a distinct right to publicity and/or the legislature enacts a new law doing the same. Some even support a post-mortem right to publicity after the author’s death. In Florida, for instance, the right of publicity extends 40 years after the author’s death. There’s even case law that suggests it applies to voices, too. See: Neva, Inc. v. Christian Duplications Intern., Inc., 743 F. Supp. 1533 (1990).

Not even this remedy is infallible. Writing for the Fordham Intellectual Property, Media & Entertainment Law Journal, J.D. Candidate Prachi Patel states:

“...because the standard for violation of the right of publicity is that the recording leads the audience to believe that the person performing is the actual plaintiff it is possible that the common law right of publicity claim may be avoided if users inform the audience that an artificial voice is being used because the audience would then know that it is not actually the person performing.”

A simple disclaimer could be enough to defeat a right of publicity claim. If that isn’t already bad enough, right of publicity protects commercial use of name, image and likeness; any non-commercial uses could be rendered invalid. 

Is There Nothing We Can Do?

The topic is too new for any persuasive case law or statutes to help inform us. Right now, copyright law at the state and federal levels aren’t saying “no,” but neither are they affirming it. 

At some point, we need to realize the damage AI is doing to the music industry. Our copyright laws are too outdated to handle the ongoing piracy of other artists’ musical talents. We need legislation to provide proper remedies under the law for AI deepfakes and blatant copies of others’ music. Changing copyright law to include protections for voices may come with unintended consequences. Nonetheless, it would build a stronger foundation for courts to penalize those who maliciously use AI to pass off someone else’s likeness for their music. 

The heart and soul of the music industry is humans expressing their emotions to others. You connect with an artist for what they represent and how their art makes you feel. It’s not something you can automate.