The copy and the original are no longer the same thing.
In an age when artificial intelligence can conjure a voice or a face without touching a single protected original, Taylor Swift has turned to an old legal instrument — the trademark — to draw new boundaries around her identity. Filing applications for two audio recordings and a performance photograph through her company TAS Rights Management, Swift is testing whether trademark law can do what copyright and publicity rights have so far failed to accomplish: close the gap that AI deepfakes have quietly slipped through. The move is untested in court and unproven in outcome, but it speaks to a broader human reckoning with what it means to own oneself in a world where likeness can be synthesized without consent.
- AI deepfakes have already weaponized Swift's voice and image — fake ads, fabricated political endorsements, and explicit content have made the threat viscerally real, not theoretical.
- Existing Right of Publicity laws were written for a world where copying meant reproducing an original, leaving a legal void that AI-generated synthesis exploits with impunity.
- Swift's legal team is attempting something courts have never ruled on: registering a spoken voice as a trademark, a strategy trademark attorney Josh Gerben calls genuinely novel territory.
- The filings are precise and deliberate — two specific audio clips promoting her album and a single onstage image chosen for its immediate, unmistakable recognizability.
- Actor Matthew McConaughey has already secured similar approvals, suggesting a quiet industry movement toward trademarking identity before legislators can act.
- If Swift's applications survive scrutiny, they could produce landmark case law that reshapes how the entertainment world — and the courts — define personal identity in the AI era.
Taylor Swift has filed trademark applications with the US Patent and Trademark Office covering two audio recordings of her voice and a photograph of herself performing onstage — a legal move that trademark experts say is designed to combat AI-generated deepfakes. The filings, submitted through her company TAS Rights Management and surfaced publicly by trademark attorney Josh Gerben, are specific by design: one clip promotes her album on Amazon Music Unlimited, another announces its release date on Spotify, and the image captures her onstage in a sequined outfit with a pink guitar — a visual that is immediately, unmistakably her.
The strategy targets a gap that AI has been quietly exploiting. Right of Publicity laws give celebrities recourse against unauthorized commercial use of their name or image, but those laws were built for a world where copying meant reproducing an original. When an AI model synthesizes a voice or generates a likeness without drawing on any single protected recording or photograph, it can slip through that legal net. Trademark law, Gerben argues, may offer a different kind of protection — guarding not just a specific recording but the distinctive mark itself against uses that cause confusion or dilute its value.
Applying that logic to a spoken voice is untested ground. Courts have not yet ruled on whether a person's voice can function as a trademark, which means Swift's filings could eventually produce case law that shapes how the entertainment industry thinks about identity in the AI age. She is not alone in pursuing this path — actor Matthew McConaughey has had similar applications approved and has spoken openly about establishing clear boundaries around consent and attribution.
For Swift, the threat is not abstract. Her voice and image have already appeared in fake advertisements, fabricated political endorsements, and explicit AI-generated content she never consented to. Whether her trademark applications are approved and whether they hold up under legal challenge remains uncertain — but the filings signal that celebrities are no longer waiting for legislators to catch up. They are reaching for the tools already available and bending them toward new purposes.
Taylor Swift has filed trademark applications with the US Patent and Trademark Office covering two audio recordings of her voice and a photograph of herself performing onstage — a legal maneuver that trademark experts say is aimed squarely at the growing threat of AI-generated deepfakes.
The filings, submitted on a Friday and made public by trademark attorney Josh Gerben on his blog the following Monday, list Swift's company TAS Rights Management as the owner of the protected material. A spokesperson for Swift had not responded to press inquiries by Monday, and the attorneys named on the filings were similarly quiet.
The two audio clips are specific and deliberate. In one, Swift's voice promotes her album 'The Life of a Showgirl,' directing listeners to Amazon Music Unlimited. In the other, she announces the album's October 3 release date and encourages fans to presave it on Spotify. The image she is seeking to trademark shows her onstage in a sequined outfit, pink guitar in hand — a visual that is immediately, unmistakably her.
Gerben, who has been watching this space closely, described the filings as purpose-built defenses against artificial intelligence. The logic is straightforward: existing law already gives celebrities some recourse through what are called Right of Publicity statutes, which restrict unauthorized commercial use of a person's name or image. But those laws have a gap that AI has been quietly exploiting. When a model is trained to mimic a voice or generate a likeness, it doesn't necessarily copy any single protected recording or photograph — it synthesizes something new. That distinction has left celebrities legally exposed in ways that weren't anticipated when the current frameworks were written.
Trademark law, Gerben argues, may help fill that gap. A registered trademark doesn't just protect a specific recording; it protects the mark itself — the distinctive sound, the recognizable image — against uses that could cause confusion or dilute its value. Applying that logic to a celebrity's spoken voice is, as Gerben noted, genuinely novel territory. Courts have not yet weighed in on whether a person's voice can function as a trademark in this way, which means Swift's filings could eventually produce case law that shapes how the entertainment industry — and the law — thinks about identity in the AI era.
Swift is not the first public figure to pursue this kind of protection. Actor Matthew McConaughey has had similar applications approved, and he spoke openly about the reasoning in January, telling the Wall Street Journal that the goal is to establish clear boundaries around ownership, with consent and proper attribution becoming standard practice in an AI-driven world.
For Swift, the stakes are particularly high. Her voice and image have been used without her consent in a wide range of AI-generated content — fake advertisements, fabricated political endorsements, and explicit imagery that she never agreed to and that caused significant public outcry. The deepfake problem is not hypothetical for her; it has already arrived.
The photograph she is trademarking is telling in its specificity. By protecting not just any image of herself but a particular visual — the jumpsuit, the pose, the guitar — her legal team is staking out ground that could be used to challenge AI-generated images that evoke her likeness without reproducing any single protected photo. It's a strategy built for an age when the copy and the original are no longer the same thing.
Whether the applications are approved, and whether the resulting trademarks hold up under legal challenge, remains to be seen. But the filings signal something larger: that celebrities and their legal teams are no longer waiting for legislators to catch up with technology. They are reaching for the tools already on the shelf and bending them toward new purposes. If Swift's approach survives scrutiny, it could become the template others follow.
Notable Quotes
These filings are specifically designed to protect Taylor from threats posed by artificial intelligence.— Trademark attorney Josh Gerben
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
The Hearth Conversation Another angle on the story
Why trademarks? She already has copyright over her music.
Copyright protects specific recordings. AI doesn't copy recordings — it learns from them and generates something new. That's the gap.
So the law as written just... doesn't cover this?
Not cleanly. Right of Publicity laws offer some protection, but they were written before anyone imagined a machine could synthesize a voice from scratch.
And a trademark on a voice would work differently?
In theory, yes. A trademark protects the mark itself — the distinctive sound — not just a particular recording. It's a broader claim.
Has anyone actually won a case using this approach?
Not yet. Gerben was explicit about that. Trademarking a spoken voice is untested in court. Swift's filings could be the test case.
What about the photo? That seems more straightforward.
It's actually quite specific — a particular outfit, pose, guitar. The idea is to protect a visual identity, not just a single image, against AI-generated likenesses that evoke her without copying anything.
McConaughey did something similar?
He did, and his applications were approved. He framed it as building a perimeter around ownership — consent and attribution as the baseline expectation.
If this works for Swift, does it work for everyone?
That's the real question. A celebrity with Swift's resources can pursue this. Most people can't. The precedent matters, but so does who gets to use it.