Taylor Swift Trademarks Her Voice and Likeness in Novel Legal Shield Against AI Deepfakes

No copy, no copyright violation — that gap is what Swift is trying to close.
AI can mimic a voice without copying any existing recording, leaving copyright law with a structural blind spot.

In an age when artificial intelligence can conjure a person's voice and face from nothing, Taylor Swift has turned to an unlikely shield: trademark law. Through her company TAS Rights Management, Swift filed applications with the U.S. Patent and Trademark Office to claim legal ownership over specific audio recordings of her voice and a photograph of her stage persona — a strategy legal observers describe as genuinely novel and largely untested. The move reflects a broader reckoning unfolding across culture and law, as public figures grapple with the unsettling reality that their identities can now be fabricated, distributed, and weaponized without their consent. What Swift is asking, in essence, is for the legal system to grow fast enough to meet the technology that has already outpaced it.

  • AI-generated deepfakes have repeatedly placed Swift's voice in fake ads, her image in explicit content, and her name on political endorsements she never made — and the problem is accelerating.
  • Copyright law, long the guardian of recorded music, has a structural blind spot: AI can replicate a voice without copying a single existing recording, leaving celebrities legally exposed.
  • Swift's team filed three trademark applications — two promotional audio clips and one stage photograph — in a deliberate attempt to build a legal wall around her identity using an untested framework.
  • Trademark attorney Josh Gerben flagged the filings publicly, noting that registering a living person's spoken voice as a trademark has never been tested in court, making the outcome deeply uncertain.
  • Actor Matthew McConaughey has pursued similar filings and spoken openly about the need for consent and attribution as baseline expectations in the AI era, signaling a growing industry push.
  • If Swift's applications survive legal scrutiny, they could hand every public figure a new instrument for defending their identity — making the stakes of these pending filings far larger than one artist's brand.

Taylor Swift has long fought to control her music and her masters. Now she is staking a claim on something more elemental: the sound of her own voice and the image of her own face.

Through her company TAS Rights Management, Swift filed three trademark applications with the U.S. Patent and Trademark Office — two audio clips and one photograph. The audio recordings are promotional in nature, featuring her voice directing fans to stream her album 'The Life of a Showgirl' on Amazon Music Unlimited and presave it on Spotify. The photograph shows her onstage in a sequined outfit, pink guitar in hand — an image closely associated with her public persona. Trademark attorney Josh Gerben flagged the filings on his blog, and legal observers quickly recognized the strategy as something new.

The need is not abstract. Deepfake videos have placed Swift's voice in fabricated advertisements. Manufactured images have put her in explicit scenarios without her consent. AI-generated political endorsements have circulated under her name. The harm is ongoing and, by most accounts, growing.

What makes the legal approach notable is the gap it is trying to close. Copyright law protects recorded music, but AI can study a voice and generate entirely new audio that sounds like that person without copying a single existing file — no copy, no violation. Trademark law, traditionally the domain of logos and slogans, may offer a different avenue. By registering specific audio clips and a visual, Swift's team could argue that AI-generated content designed to evoke her voice or appearance constitutes trademark infringement, even without reproducing any protected original.

Swift is not alone in this effort. Matthew McConaughey has had comparable filings approved and has spoken publicly about establishing consent and attribution as baseline expectations in an AI-saturated world. But the legal infrastructure to enforce such claims remains largely unbuilt, and courts have not yet ruled on whether a spoken voice qualifies as a protectable trademark.

The applications are pending. Whether they survive scrutiny is an open question — trademark law was not designed with synthetic media in mind. But if they do, the precedent could extend well beyond one artist, offering every public figure a new legal tool for defending their identity against a technology that has already learned to wear their face.

Taylor Swift has never been shy about protecting what's hers — her music, her masters, her brand. Now she's staking a claim on something more fundamental: the sound of her own voice and the image of her own face.

On Friday, Swift's company TAS Rights Management filed three trademark applications with the U.S. Patent and Trademark Office — two audio recordings and one photograph. The move, first flagged publicly by trademark attorney Josh Gerben on his blog Monday, is being read by legal observers as a deliberate attempt to build a legal wall around her identity in an era when artificial intelligence can conjure a convincing version of almost anyone.

The audio clips are promotional in nature. In one, Swift's voice announces her album 'The Life of a Showgirl,' directing listeners to Amazon Music Unlimited. In the second, she promotes the same record's release date of October 3rd and encourages fans to presave it on Spotify. The image she's seeking to trademark shows her onstage in a sequined outfit, pink guitar in hand — a pose and aesthetic that has become closely associated with her public persona.

Swift's likeness has been weaponized repeatedly in the AI era. Fabricated videos have placed her voice in fake advertisements. Manufactured images have put her in explicit scenarios she never consented to. Deepfake political endorsements have circulated under her name. The problem is not hypothetical — it has been ongoing and, by most accounts, accelerating.

The legal strategy Gerben describes is genuinely novel. Trademark law has traditionally been the domain of logos, brand names, and slogans — not the spoken voice of a living person. Copyright law has long protected recorded music, but it has a structural blind spot: AI systems can study a voice and generate entirely new audio that sounds like that person without ever copying a single existing recording. No copy, no copyright violation. That gap is precisely what Swift's team appears to be trying to close.

'Registering a celebrity's spoken voice is a new use of trademark registration that has not been tested in courts,' Gerben wrote. He noted that while right-of-publicity laws already offer some protection against unauthorized use of a person's likeness, trademark registration adds another layer — one that could give Swift's legal team additional grounds to pursue claims against AI-generated content that mimics her without her permission.

The photograph filing follows similar logic. By trademarking a specific visual — the sequined jumpsuit, the guitar, the stage pose — Swift's representatives may be able to argue that AI-generated images designed to evoke her appearance constitute trademark infringement, even if they don't reproduce any single protected photograph.

Swift is not the first celebrity to try this approach. Actor Matthew McConaughey has had comparable filings approved, and he spoke publicly about the reasoning in January, telling the Wall Street Journal that the goal is to establish clear ownership boundaries in a world where AI-generated content is becoming the norm. His framing — consent and attribution as the baseline expectation — captures what many in the entertainment industry are pushing for, even as the legal infrastructure to enforce it remains largely unbuilt.

Neither Swift's spokesperson nor the attorneys listed on the filings responded to requests for comment. The applications are pending, and whether they will survive legal scrutiny is an open question. Trademark law was not designed with synthetic media in mind, and courts have not yet weighed in on whether a spoken voice clip qualifies as a protectable mark in the way a logo or slogan does.

What Swift is doing, in essence, is asking the legal system to catch up with the technology — and doing it in a way that, if it succeeds, could hand every public figure a new tool for defending their identity. The outcome of these filings may matter far beyond one artist's catalog.

We want to create a clear perimeter around ownership, with consent and attribution the norm in an AI world.
— Matthew McConaughey, speaking to the Wall Street Journal in January
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.
— Trademark attorney Josh Gerben, writing on his blog
The Hearth Conversation Another angle on the story
Inventor

Why trademark instead of copyright? Isn't her voice already protected?

Model

Copyright protects specific recordings — the actual audio file. But AI doesn't need to copy a file. It can listen to thousands of hours of her voice and generate something new that sounds just like her. No copy, no infringement under copyright law.

Inventor

So the law has a hole in it.

Model

A significant one. And it's not hypothetical — fake ads, fake endorsements, explicit images. It's been happening to her for years.

Inventor

Has anyone actually won a case using this trademark approach?

Model

Not yet. Gerben was clear that trademarking a spoken voice is untested in courts. McConaughey got his filings approved, but approval and legal victory are different things.

Inventor

What makes a voice trademarkable? It seems like a stretch.

Model

Trademark law protects distinctive identifiers associated with a source. The argument is that Taylor Swift's voice is so recognizable it functions like a brand mark — you hear it and you know exactly who it is.

Inventor

And the photograph — why trademark a photo when you could just copyright it?

Model

The photo isn't really about the image itself. It's about the visual identity — the sequined outfit, the pose, the pink guitar. Trademarking that combination could let her pursue claims against AI images that evoke her look without reproducing any specific picture.

Inventor

What does she actually gain if these are approved?

Model

Another legal avenue. Right-of-publicity laws vary by state and have limits. A federal trademark registration gives her team a more consistent, nationwide tool to send takedown notices and pursue infringers.

Inventor

Could this set a precedent for ordinary people, not just celebrities?

Model

The framework could, in theory. But trademark registration requires demonstrating that something is a distinctive identifier in commerce. Most people don't meet that bar. This is, for now, a tool for the famous.

Contact Us FAQ