A system can sound like Taylor Swift without copying a single second of her music.
In an age when artificial intelligence can conjure a convincing likeness from nothing, Taylor Swift has turned to an old legal instrument — the trademark — to stake a claim over something as intimate and irreplaceable as her own voice. By filing applications for two audio recordings of herself speaking, she is testing whether the law can be stretched to cover not just brand sounds, but the human voice itself. No court has yet been asked to answer that question, and the answer, when it comes, will matter far beyond one artist's catalog.
- For years, fake versions of Taylor Swift have circulated online — endorsing products she never touched, appearing in images she never consented to, and lending her voice to causes she never supported.
- AI voice-cloning has exposed a dangerous gap in copyright law: a system can now generate audio that sounds exactly like a real person without copying a single second of their actual recordings, leaving existing protections effectively toothless.
- Swift's trademark filings — covering two spoken promotional clips and a stage photograph — represent a novel legal gambit, attempting to use 'confusing similarity' standards to challenge convincing imitations, not just direct copies.
- The strategy is untested: sound trademarks exist for brand jingles and corporate chimes, but no court has ever been asked whether a celebrity's spoken voice can be owned in the same way.
- Swift is part of a growing wave — Matthew McConaughey, Scarlett Johansson, Tom Hanks, and over 700 artists have all pushed back against AI misuse, signaling that the entertainment industry is moving toward a collective legal reckoning.
Taylor Swift has spent years watching fabricated versions of herself circulate online — endorsing products she never touched, appearing in compromising images, and seemingly backing political candidates she never supported. On Friday, she filed trademark applications for two audio recordings of her own voice, both promotional clips tied to her album 'The Life of a Showgirl,' along with a photograph of herself performing onstage. Together, the filings represent something genuinely new: a major celebrity attempting to use trademark law as a direct shield against AI-generated imitation.
The legal logic is precise. Copyright protects recorded work, but AI voice-cloning can now produce audio that sounds unmistakably like a real person without copying a single second of their actual recordings — leaving copyright with nothing to grab onto. Trademark law operates differently, on a standard of 'confusing similarity.' If a fake voice is close enough to fool a reasonable listener, a legal claim may follow. Trademark attorney Josh Gerben, who surfaced the filings publicly, noted that while sound marks have existed for decades — MGM's lion roar, NBC's three-note chime — registering a celebrity's spoken voice is territory no court has yet been asked to navigate.
Swift is not alone in confronting this problem. Matthew McConaughey filed a similar suite of trademarks in January, becoming the first major celebrity to do so. Scarlett Johansson has battled unauthorized use of her likeness and voice by AI companies. Tom Hanks and Bryan Cranston have raised public alarms. More than 700 artists backed a campaign called 'Stealing Isn't Innovation,' demanding tech companies stop training generative AI on copyrighted material without consent. YouTube, meanwhile, announced a partnership with talent agencies to streamline deepfake removal requests.
Whether Swift's filings will hold is a question only the courts can answer — and they haven't been asked yet. But if the theory survives legal scrutiny, the playbook for protecting celebrity identity in the age of generative AI may look very different within a few years, and a manufactured voice may finally have somewhere to be challenged.
Taylor Swift has spent years watching fake versions of herself spread across the internet — hawking cookware she never endorsed, appearing in sexually suggestive images she never posed for, even seeming to back a presidential candidate she never supported. On Friday, she moved to close some of the legal distance between herself and those fakes, filing trademark applications for two audio recordings of her own voice.
The clips are promotional in nature. In one, she announces her album 'The Life of a Showgirl' and directs listeners to Amazon Music Unlimited. In the other, recorded in a slightly lower register, she promotes the same album's presave on Spotify, noting an October 3 release date. A third application covers a photograph of her performing onstage in one of her signature sequined bodysuits, pink guitar in hand. Together, the filings represent something genuinely new: a major celebrity attempting to use trademark law as a shield against AI-generated imitation.
Trademark attorney Josh Gerben, who first surfaced the applications publicly, was direct about what makes this unusual. Sound marks have existed for decades — MGM's lion roar, NBC's three-note chime, the Pillsbury Doughboy's laugh are all registered trademarks. But those protect brand sounds, not human voices. Attempting to register a celebrity's spoken voice as a trademark, Gerben wrote, is territory that no court has yet been asked to navigate.
The legal logic behind the move is worth understanding. Copyright law has long been the primary tool musicians use to protect their recorded work. But AI voice-cloning technology has opened a gap that copyright cannot easily reach: a system can now generate entirely new audio that sounds like Taylor Swift without ever copying a single second of her actual recordings. There's nothing to infringe if nothing is copied. Trademark law, however, operates on a different standard — 'confusing similarity.' If a fake voice is close enough to fool a reasonable listener, that may be enough to bring a legal claim. By registering specific spoken phrases tied to her voice, Swift would have standing to challenge not just exact reproductions but convincing imitations.
Swift has filed hundreds of trademarks over the course of her career, protecting her name, her lyrics, her merchandise, and various elements of her public identity. But this appears to be the first time she has sought protection for the sound of her voice itself. The attorney listed on the applications, Rebecca Liebowitz, did not respond to a request for comment, and a spokesperson for Swift was similarly unavailable.
She is not alone in confronting this problem, and not quite the first to try this approach. In January, Matthew McConaughey became the first major celebrity to file a suite of trademarks covering images, video, and audio of himself — a move that AI experts noted could become a template for others. The broader celebrity community has been raising alarms about AI misuse for several years now. Scarlett Johansson's legal team demanded an AI app stop using her likeness in 2023, and she publicly called out OpenAI the following year for deploying a chatbot voice she described as eerily similar to her own — despite having turned down the company's request to use it. Tom Hanks flagged fake ads using his name and voice to promote medical products. Bryan Cranston raised concerns about OpenAI's Sora video tool and its ability to replicate likenesses without consent. In January of this year, more than 700 artists and creatives — including Johansson, Cate Blanchett, and Joseph Gordon-Levitt — backed a campaign called 'Stealing Isn't Innovation,' organized by a coalition of unions and artists' rights groups pushing tech companies to stop training generative AI on copyrighted material without permission.
On the platform side, YouTube last week announced a partnership with several talent agencies to give celebrities easier access to its deepfake detection tool, allowing them to request removal of unauthorized likenesses from the platform.
What Swift's filings ultimately accomplish will depend on courts that have not yet been asked the question. Trademark protection for a human voice is an untested theory, and the entertainment industry will be watching closely to see whether it holds. If it does, the legal playbook for protecting celebrity identity in the age of generative AI may look very different within a few years — and the line between a real voice and a manufactured one may finally have somewhere to go when it needs to be enforced.
Notable Quotes
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, trademark attorney who first reported Swift's filings
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
The Hearth Conversation Another angle on the story
Why trademark the voice specifically? Doesn't copyright already cover recordings?
It covers the recordings themselves — but not what AI can do with them. A model can learn to sound like Taylor Swift without ever sampling a single second of her actual music. Copyright has no grip on that.
So the trademark is meant to fill that gap?
That's the theory. Trademark law asks whether something is 'confusingly similar' — close enough to mislead a reasonable person. That standard might actually reach AI imitation in a way copyright can't.
Has anyone tried this before?
McConaughey filed something similar in January, covering images, video, and audio of himself. But no court has ruled on whether a celebrity's spoken voice can be trademarked. Swift's filings would be the first real test if challenged.
Why use promotional clips rather than, say, a song?
A song would be a copyright question. These are spoken-word phrases tied to her commercial identity — closer to how a brand sound works. MGM's lion roar is a trademark. She's essentially arguing her voice is a brand asset in the same way.
What does she actually gain if the trademark is granted?
Legal standing. Right now, if someone clones her voice to sell cookware or endorse a politician, the path to court is murky. A registered trademark gives her a cleaner basis to sue — and potentially to demand takedowns before a lawsuit is even necessary.
Is this something only someone like Swift can afford to do?
Practically, yes — for now. Filing and defending trademarks takes money and lawyers. But if the strategy works, it could push for broader legislative solutions that don't require each celebrity to build their own legal fortress.
What's the thing beneath the thing here?
It's a recognition that the law hasn't kept up. The technology to fake a person's voice convincingly is already widely available. The legal infrastructure to stop it is still being invented in real time.