Taylor Swift Is Trademarking Her Voice. That Shouldn’t Have to Exist.

Taylor Swift’s company filed three trademark applications on April 24. Two cover her spoken voice — the specific phrase “Hey, it’s Taylor Swift” and its abbreviated version “Hey, it’s Taylor.” The third covers a photograph of her onstage during the Eras Tour in a specific outfit. The filings went through TAS Rights Management. The reason, per every outlet that covered it: artificial intelligence. Not a licensing dispute. Not a business deal gone wrong. A technology that can replicate someone’s voice from training data and produce it on demand, without consent, at scale.

The trademarks are worth understanding because they represent something legally new. You can trademark a name, a logo, a phrase. Trademarking a human voice as a defense against AI impersonation hasn’t been properly tested in court yet — there’s no clear precedent for how it holds against a model trained on recordings rather than a direct copy of any specific track. What Swift is doing is planting a legal flag in territory that isn’t mapped, and hoping the courts eventually rule that it means something. That’s not a knock on the strategy. It’s just what the strategy is.

The reason this has to exist is worth sitting with. Earlier this year, sexually explicit deepfakes of Swift circulated widely enough that the story reached mainstream news and forced a policy conversation that resolved nothing in particular. AI voice cloning has matured to a point where the average listener cannot reliably distinguish a synthetic version from the real thing. And copyright law — which covers recorded music — doesn’t obviously cover a voice generated from scratch by a model that was trained on recordings without copying any one of them directly. That’s the gap Swift is trying to address. The law hasn’t caught up, so she’s using the closest available tool.

She’s not the first. Matthew McConaughey filed similar trademarks in January, and more high-profile names will follow. The core legal move is the same for all of them: arguing that the sound of you is a brand asset — a frame courts have some framework for — rather than arguing that your identity itself carries rights independent of its commercial value, which courts have far less framework for. It’s a clever move. But calling it a workaround is also accurate, and accuracy matters here, because the next question is whether workarounds will be enough — or whether the actual fix requires legislation, platform liability standards, and a court system that currently has essentially no case law on AI impersonation at scale.

There’s also an enforcement problem the coverage mostly skips. Filing a trademark application is relatively fast and cheap. Monitoring every platform, podcast service, voice assistant, and social media feed for unauthorized uses of your voice — at the scale AI synthesis operates — is not. Trademark enforcement requires active policing by the rights holder, not a platform-triggered takedown. There’s no DMCA equivalent here: no standardized notice-and-takedown framework that activates automatically when a violation appears. Swift can file the paperwork, but actually stopping the specific fake that goes viral on a Tuesday requires identifying it, issuing a legal demand, and waiting for a platform to respond. Platforms have no legal obligation to police trademark violations the way they do copyright claims. The AI moves faster than that process does. The trademark is the right tool. It just isn’t fast enough for the problem it’s being asked to solve.

We covered the platform side of this already — 75,000 AI-generated tracks landing on streaming services every day, artists losing royalties to synthetic versions of their sound that no platform has the infrastructure to catch at scale. That piece was about what happens when the machinery moves faster than the gatekeepers. This is about what happens when a specific artist with significant legal resources tries to build a moat around her own identity. Both stories are true. They just describe different parts of the same problem.

The question this filing raises isn’t really about Taylor Swift. It’s about what happens to every artist who can’t afford to file trademarks in multiple jurisdictions, hire a team to monitor violations, and sustain litigation against a technology sector that moves faster than the courts that are supposed to regulate it. Swift has resources that most artists don’t. She’s using them intelligently. But “intelligent use of resources” is a solution to a different problem than the one that actually exists, which is that the legal infrastructure for protecting human identity from synthetic reproduction is not there yet — and the industry building the tools to do the replication has no particular incentive to wait for it to arrive. So artists with the means will do exactly what Swift did: find the closest available legal tool and file the paperwork. The artists without the means will not. The gap between those two groups is the actual story.

Sources: Variety · CNN Business · NPR