The trademark office just became the new frontline in the AI identity wars, and Swift's team knows the legal system isn't built for this fight.
The Summary
- Taylor Swift's team filed trademark applications for two spoken phrases — "Hey, it's Taylor Swift" and "Hey, it's Taylor" — including audio clips of her voice from a recent album promotion
- IP attorney Josh Gerben says these trademarks are "specifically designed to protect Taylor from threats posed by artificial intelligence" as celebrities lose control over their voices and likenesses
- The move could give Swift legal standing against AI deepfakes, but it's uncharted territory for trademark law
- This escalates years of AI imitation controversies surrounding Swift, but the legal strategy may be a long shot
The Signal
Swift isn't the first celebrity to face AI copycats, but she might be the first to weaponize trademark law this specifically against voice cloning. TAS Rights Management filed the applications with actual audio recordings attached, turning her voice itself into protected intellectual property. Not just the words. Not just the likeness. The literal sound waves.
Here's why this matters: traditional IP law wasn't built for generative AI. Copyright protects creative works. Right of publicity protects against unauthorized commercial use of your identity. But neither cleanly addresses AI models trained on your voice, spitting out infinite variations that sound exactly like you but saying things you never said.
"Trademark law is being retrofitted into an AI defense weapon because the other legal tools are too slow."
The trademark angle is clever and desperate in equal measure. Gerben's assessment that this is an AI-specific defensive strategy reveals the broader problem. Celebrities and their legal teams are scrambling because:
- Voice cloning models can train on minutes of audio, not hours
- Deepfake audio is nearly indistinguishable from real recordings
- Current right-of-publicity laws vary wildly by state and weren't written for synthetic media
But trademark protection requires proving consumer confusion in commerce. Swift would need to show that when someone hears an AI-generated "Hey, it's Taylor," they believe it's actually her endorsing a product. That's a narrow needle to thread. AI voice clones aren't usually selling sneakers. They're making social media content, spreading misinformation, or creating synthetic pornography.
The Verge notes this may be a long shot, and they're right. Trademark office examiners will ask: is a greeting phrase distinctive enough to warrant protection? Does Swift use this phrase in commerce consistently? Can you trademark the sound of your own voice saying hello? These are questions the legal system isn't equipped to answer yet, because the technology moved faster than the law ever could.
The Implication
Watch for more celebrities to file similar applications in the next six months. Swift's team is probing for a legal framework that doesn't exist yet, and whether this works or not, it maps the boundary of what's possible under current law. If you're building AI voice or avatar tools, this is a warning shot. The big names are lawyering up, and they're coming after the platforms that enable this stuff, not just the individual users making deepfakes. For everyone else, this is a reminder that our digital identities are becoming separable from us, and the tools to protect them are being invented in real time by whoever can afford the best attorneys.