Taylor Swift files AI identity trademarks

Taylor Swift is trying a new legal move against AI impersonation. Instead of waiting for Congress or the courts to settle what counts as an illegal deepfake, she filed three trademark applications with the U.S. Patent and Trademark Office on April 24 through TAS Rights Management. Two cover short clips of her voice. One covers a specific stage image tied to the Eras Tour look. The point is simple — create cleaner legal hooks before fake audio and fake endorsements spread even further. ### What exactly did she file? The filings appear to cover two spoken phrases in Swift’s own voice — “Hey, it’s Taylor Swift” and “Hey, it’s Taylor” — plus a visual mark built around a recognizable onstage image of Swift holding a pink guitar while wearing a multicolored bodysuit and boots. That matters because trademark law can protect source identifiers — things that make consumers think “this came from her” — even when copyright law is a messier fit for voice clones or AI-generated lookalikes. ### Why use trademark law for an AI problem? Because trademark law is really about confusion. If an AI ad, fake endorsement, cloned greeting, or synthetic promo makes people think Swift approved it, that starts to look like classic trademark territory. Copyright is less tidy here — especially when the output is new content that imitates a style, a voice, or a persona without directly copying a specific recording or photo. Swift’s move basically says: if the law won’t cleanly stop imitation, maybe it can still stop misleading commercial use. ### Is this the same as owning your face? Not quite. A trademark is narrower than a general “right to be yourself.” It protects specific marks used in commerce. So this does not mean Swift now owns every version of her voice or every image that resembles her. The catch is that a registered mark can still become a useful weapon if someone uses something close enough to confuse fans, buyers, listeners, or advertisers. That’s why the exact phrases and the specific visual look matter. ### Why are people calling this a template? Because Matthew McConaughey tried something very similar earlier this year. His filings included marks tied to his voice and signature persona, including the famous “alright, alright, alright” line. Lawyers quoted in coverage of Swift’s case see the same strategy at work — lock down recognizable identity markers as trademarks, then use those registrations to challenge AI misuse faster and more concretely. If that works for stars, athletes, podcasters, and creators will notice. ### Why now? The pressure has been building for more than a year. AI tools can now clone voices and generate convincing celebrity images cheaply and at scale. Swift was already a flashpoint in the deepfake debate after nonconsensual fake images of her spread online in 2024, which helped push lawmakers and platforms to talk more seriously about synthetic media harms. These filings land in that unresolved gap — the tech got good fast, but the legal tools are still patchy. ### Will this actually stop deepfakes? Not by itself. Trademark claims work best when there is commercial use and likely consumer confusion. Random viral fakes, parody, anonymous posts, and offshore content can still be hard to police. But this move could make certain cases easier — fake ads, fake endorsements, fake branded audio, and other uses where someone is clearly trading on Swift’s identity. Think of it less as a universal shield and more as a better lock on the doors most likely to get kicked in first. ### What’s the bottom line? Swift is not solving AI law. She’s stress-testing a workaround. If these applications mature into registrations and prove useful in enforcement, they could become a playbook for famous people trying to protect the parts of themselves AI can now mimic almost perfectly.