The Wild Legal Battle Over AI Voice Clones
(Read time: 3 Minutes)
By Ollie Coull
Imagine scrolling through TikTok and hearing a brand-new, unreleased track by your favorite artist. The vocals are flawless, the beat is perfect, and the lyrics are incredibly sharp. The song racked up millions of streams overnight, but there is one massive catch: the artist never stepped foot in a recording studio. They had no idea the song even existed. It was entirely generated by a fan using a highly sophisticated AI voice model trained on the artist’s existing music. When the record label sends a furious takedown notice, they discover a shocking loophole in the legal system: in many jurisdictions, you cannot copyright a human voice. As generative AI advances at a terrifying speed, courts are scrambling to answer a foundational question: who owns your voice when a machine can replicate it perfectly?
To understand how someone can legally steal your voice, you have to look at the boundaries of traditional copyright law. Historically, copyright protects physical expressions of ideas—things like written lyrics, musical compositions, and specific sound recordings. But copyright does not protect a person's physical characteristics, such as the unique timber of their vocal cords. If an AI creator trains a model on your songs to generate a completely new melody with your "voice," they haven't technically copied your audio files; they have just taught a computer how to mimic you. The real legal chaos starts when lawyers try to stretch old laws to cover this brand-new digital identity theft, forcing a messy clash between copyright, trademark, and the "right of publicity."
In the United States, the primary weapon against AI voice cloning is the right of publicity—the legal right of an individual to control the commercial use of their name, image, and likeness. This was famously tested decades before AI in the landmark case Midler v. Ford Motor Co., where Ford hired a sound-alike singer to mimic Bette Midler’s voice for a commercial after she refused to do it herself. The court ruled in Midler's favor, declaring that a voice is as distinctive and personal as a face, and deliberately counterfeiting it is a tort. Today, stars like Scarlett Johansson are using these exact legal precedents to battle AI companies who build "custom" voices that sound suspiciously identical to them.
Over in the UK and commonwealth jurisdictions, the legal battle is much harder because there is no federal "right of publicity." Instead, lawyers have to rely on an ancient, clunky common-law concept called "passing off." To win a passing off case, an artist has to prove that the AI creator misrepresented the clone as the real artist, causing genuine commercial damage. But if an AI creator explicitly labels a song as "AI Drake" or "Drake Clone," they aren't technically tricking the public into thinking Drake sang it. This leaves artists incredibly vulnerable, prompting governments to draft urgent new legislation to protect digital identity as a fundamental human right.
At the end of the day, the legal system is facing a massive identity crisis. If the law makes voice protection too strict, it could stifle creative expression, parody, and the development of cutting-edge technology. But if it leaves the loophole wide open, it risks destroying the creative industry entirely, turning human performers into free data for corporate algorithms. As the courts scramble to catch up, the music and tech industries are learning a brutal truth: in the digital age, your own voice might no longer belong to you.
References (APA 7)Midler v. Ford Motor Co., 849 F.2d 460 (9th Cir. 1988). Haliczer v. OpenAI OpCo, LLC, F. Supp. 3d (S.D. Cal. 2024). Irvine v. Talksport Ltd, EWHC 367 (Ch) (2002).
