The voice is mimicked, not copied. The law has no clean answer for that.
In an age when a machine can conjure a voice without ever borrowing a single note, Taylor Swift has turned to one of law's oldest instruments — the trademark — to draw a boundary around something as intimate and irreplaceable as the sound of a human being speaking. By filing 'sound marks' with the United States Patent and Trademark Office, Swift and her legal team are not merely protecting a brand; they are asking the law to reckon with a question it has never formally answered: can a living person's voice, in its essence, belong to them in the eyes of the state? The answer, still unwritten, may reshape how artists and institutions navigate the frontier between identity and imitation for generations to come.
- Generative AI can now clone a celebrity's voice without touching a single copyrighted file — and existing law has no clean remedy for that.
- Swift's filing of spoken-word 'sound marks' is being described by trademark attorneys as genuinely unprecedented, with no court ruling yet to validate or reject the theory.
- Her broader portfolio of over 300 U.S. trademark applications signals a deliberate, systematic effort to legally fortify every dimension of her public identity against synthetic reproduction.
- Actor Matthew McConaughey has filed parallel applications, suggesting the entertainment industry is quietly treating trademark law as a forward hedge against AI — not just a branding tool.
- The filings now sit with the USPTO, and if they survive challenge, courts could be forced to decide whether a living person's voice is a protectable mark — a ruling that would expand intellectual property law without a single act of Congress.
Two audio clips and a photograph — that is what Taylor Swift submitted to the United States Patent and Trademark Office last Friday, and intellectual property lawyers are already calling the filings something genuinely new in American trademark law.
The clips are promotional: in one, Swift announces her album and directs listeners to Amazon Music Unlimited; in the other, she promotes a release date and encourages fans to pre-save on Spotify. The image shows her on stage holding a pink guitar in a sequinned outfit associated with her Eras tour. The image follows familiar celebrity trademark strategy. The audio does not.
Trademark law has long recognized 'sound marks' — the NBC chime, the MGM lion's roar — but registering a living person's spoken voice as a protectable trademark has never been tested in court. Attorney Josh Gerben described the filings as 'specifically designed' to shield Swift from AI-related threats. Her company, TAS Rights Management, already holds more than 300 U.S. trademark applications, a portfolio that one WIPO lawyer has called a deliberate strategy to reinforce Swift's identity across every public dimension.
The AI problem driving these filings is structural. Copyright protects existing recordings from being copied, but it cannot stop a generative model from synthesizing a convincing new audio clip that sounds like a specific person without ever touching a protected file. The voice is mimicked, not copied — and the law, as written, has no answer for that.
Trademark law might. If Swift's voice is registered as a mark, she could argue in court that AI-generated audio resembling that sound constitutes infringement. Whether a judge would agree remains entirely open. Swift is not alone: Matthew McConaughey has filed comparable applications, suggesting that parts of the entertainment industry are already treating trademark registration as a legal hedge against a technology flooding platforms with synthetic likenesses.
If these filings hold up, courts — not legislators — will have quietly expanded how intellectual property law functions in the AI era. If they fail, the gap between what the technology can do and what the law can address will remain as wide as ever. Swift's lawyers have not commented. The applications are now with the USPTO, and eventually, perhaps, the courts.
Two audio clips and a photograph — that's what Taylor Swift submitted to the United States Patent and Trademark Office last Friday, and the filings are already drawing attention from intellectual property lawyers who say they represent something genuinely new in American trademark law.
The clips are promotional in nature. In one, Swift's voice announces her album 'The Life of a Showgirl' and directs listeners to Amazon Music Unlimited. In the other, she promotes the same record's release date of October 3rd and encourages fans to pre-save it on Spotify. The image is a stage photograph: Swift holding a pink guitar, dressed in a sequinned outfit — a look that became closely associated with her recent Eras tour.
What makes these filings unusual isn't the image, which follows a well-worn path in celebrity trademark strategy. It's the audio. Trademark law has long recognized what are called 'sound marks' — think the NBC chime or the MGM lion's roar — but attempting to register a living celebrity's spoken voice as a protectable trademark is something courts have not yet been asked to evaluate. Josh Gerben, a trademark attorney and founder of Gerben IP, described the filings as 'specifically designed' to shield Swift from AI-related threats, and noted that this particular application of trademark registration has no legal precedent to lean on.
The filings sit inside a much larger portfolio. Swift's company, TAS Rights Management, is listed as the owner on more than 300 trademark applications in the United States alone — a body of filings that intellectual property lawyer Leticia Caminero of the World Intellectual Property Organization has described as a deliberate strategy to reinforce Swift's brand across every dimension of her public identity.
But the AI angle is what gives these latest filings their edge. Copyright law, the traditional instrument artists reach for when their work is imitated, has a structural problem in the age of generative AI: it protects existing recordings from being copied, but it cannot stop a model from synthesizing an entirely new audio clip that sounds convincingly like a specific person without ever touching a single protected file. The voice is mimicked, not copied. The law, as written, has no clean answer for that.
Trademark law might. In theory, if Swift's spoken voice is registered as a mark, she could argue in court that any AI-generated audio resembling that registered sound — or any synthetic image of her in a sequinned outfit with a guitar — constitutes an infringement of her trademark rights. Whether a judge would agree is another matter entirely. The theory has not been tested.
Swift is not alone in exploring this territory. Actor Matthew McConaughey has filed comparable applications in recent months, also seeking trademark protection for his voice and likeness. The parallel efforts suggest that at least some corners of the entertainment industry are treating trademark registration not just as a branding tool but as a forward-looking legal hedge against a technology that is already generating realistic videos of well-known performers and flooding streaming platforms with AI-produced music.
The broader stakes are significant. If these filings survive legal challenge and courts begin to recognize celebrity voices as protectable trademarks, it would mark a meaningful expansion of how intellectual property law functions — one driven not by legislation but by individual artists testing the boundaries of existing statutes. If the filings fail, or go untested, the gap between what AI can do and what the law can address will remain wide open.
Swift's lawyers have not commented publicly. The applications are now in the hands of the USPTO, and eventually, perhaps, the courts.
Citações Notáveis
Attempting to register a celebrity's spoken voice is a new use of trademark registration that has not been tested in court before.— Josh Gerben, trademark attorney and founder of Gerben IP
AI technologies now allow users to generate entirely new content that mimics an artist's voice without copying an existing recording, creating a gap that trademarks may help fill.— Josh Gerben, Gerben IP
A Conversa do Hearth Outra perspectiva sobre a história
Why does it matter that these are 'sound marks' specifically? Couldn't she just sue under copyright?
That's exactly the gap she's trying to close. Copyright protects a recording — a fixed, existing thing. But AI can generate a new voice clip that sounds like Taylor Swift without ever touching one of her actual recordings. Nothing gets copied, so copyright has nothing to grab onto.
So trademark is a workaround?
More like a different tool for a different problem. Trademark protects identity — the signal that tells the public who something comes from. If her voice is a registered mark, the argument becomes: this AI clip is confusingly similar to a protected identifier, not that it copied a protected work.
Has anyone won a case like this before?
No. That's the honest answer. Gerben was explicit about it — registering a celebrity's spoken voice as a trademark is untested in court. These filings are essentially a legal experiment.
What about the image she filed — the one with the pink guitar?
That's more conventional territory. Image trademarks for celebrities exist. But pairing it with the voice clips suggests a coordinated strategy: lock down the whole recognizable package, not just one piece of it.
Matthew McConaughey is apparently doing something similar?
Yes, which is interesting. When two high-profile names independently pursue the same novel legal theory around the same time, it starts to look less like individual caution and more like a coordinated industry response.
What happens if the USPTO rejects the applications?
Then the theory dies quietly, at least for now. But if they're approved and someone challenges them in court, that's where the real precedent gets made — or doesn't.
Is there any legislation moving to address this more directly?
The source doesn't say, and that silence is telling. Right now, artists are improvising with existing law because Congress hasn't caught up. These filings are what filling a vacuum looks like.