Celebrities Like Taylor Swift Are Setting the Guardrails for the AI Age

Taylor Swift has spent years building one of the most legally fortified personal brands in entertainment. But her latest round of trademark filings ventures into previously uncharted territory — one shaped entirely by the rise of generative artificial intelligence. In April 2025, Swift's company submitted applications to trademark two audio recordings: one of her saying "Hey, it's Taylor," and another saying "Hey, it's Taylor Swift." The clips are deliberately simple. Their significance, however, runs much deeper.

The strategy, flagged by intellectual property attorney Josh Gerben of Gerben IP, hinges on what is known as a "sound mark" — a legal tool that has historically been used to protect jingles and sonic logos rather than human voices. NBC's famous three-note chime is one well-known example. The premise, if applied to Swift's case, would be that any AI-generated audio closely mimicking the timbre, cadence, and character of her voice could constitute an infringement of her registered intellectual property.

"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, Gerben IP

Under existing copyright law, reproducing a recorded song without permission is illegal. But copyright only protects fixed, original works. If an AI system generates a new audio clip that sounds like Taylor Swift without ever sampling a single second of her actual recordings, no copyright has technically been violated. Sound mark protection would potentially close that loophole — though, as Gerben notes, the approach remains entirely untested in court.

Swift is not alone in pursuing this path. Matthew McConaughey has filed similar trademark applications aimed at protecting the unique sound of his voice. Together, the moves signal a growing awareness among celebrities and their legal teams that the laws governing identity and likeness were written for a pre-AI world.

The Deepfake Crisis That Forced a Reckoning

The urgency behind these filings is not abstract. Celebrities — particularly women — have become prime targets for AI-powered impersonation, with consequences ranging from commercial misappropriation to outright abuse. Swift herself became the subject of a high-profile deepfake campaign in early 2024, when sexually explicit AI-generated images of her circulated widely on platforms including 4chan and X. The images spread rapidly before platforms began removing them, drawing widespread condemnation and renewed calls for federal legislation targeting nonconsensual synthetic media.

The incident was a stark illustration of just how far the technology has outpaced the law. Earlier generations of fake imagery required skilled graphic artists and significant time. Today, sophisticated deepfakes can be produced in seconds by anyone with an internet connection and the right prompts. The barrier to impersonation has collapsed.

"With support from leading talent agencies and management companies, including CAA, UTA, WME, and Untitled Management, we've worked to refine how likeness detection can best serve talent," YouTube said in a recent statement, announcing an expansion of its deepfake detection tools to Hollywood.

A Pattern of Pushback

The battles being waged go beyond individual filings. In 2024, OpenAI paused the rollout of a ChatGPT voice feature after actor Scarlett Johansson publicly accused the company of deliberately imitating her voice — a claim that landed with particular cultural weight given her performance as an AI chatbot in the 2013 film Her. OpenAI maintained that the voice was recorded by a different actor and that any resemblance was unintentional, but the company paused the rollout regardless.

Separately, the family of the Reverend Martin Luther King Jr. pressured OpenAI to remove representations of the civil rights leader from Sora, its AI video generation platform, before the service was shut down. The action highlighted another dimension of the problem: AI systems can not only mimic living celebrities, but also resurrect and potentially distort the legacies of historical figures.

The accumulation of these cases is beginning to resemble something larger than a series of individual disputes. Last year, more than 400 Hollywood executives and creators signed an open letter to OpenAI and Google, opposing the use of copyrighted creative work to train AI models without consent or compensation.

Will Protections Extend Beyond the Famous?

There is a pointed irony at the center of all this activity: the celebrities most capable of mounting legal and reputational challenges to AI companies are also among the least vulnerable to its most harmful effects in everyday terms. Swift's legal team, McConaughey's attorneys, the King family estate — these are parties with the resources and leverage to fight back. The vast majority of people targeted by deepfake abuse, manipulated audio, or identity theft have no such recourse.

Federal law has struggled to keep pace. While some states have enacted statutes targeting nonconsensual deepfakes — particularly in the context of sexual imagery — a comprehensive federal framework governing AI-generated likenesses remains absent.

The optimistic reading of the celebrity push is that high-profile cases create legal precedents and cultural pressure that eventually benefit everyone. A successful sound mark lawsuit brought by Taylor Swift could establish legal principles applicable far beyond the entertainment industry.

The pessimistic reading is that celebrity activism primarily secures an opt-out for the privileged, while the rest of the population is left to navigate an internet increasingly saturated with synthetic replicas of real people. The guardrails being built right now may end up resembling the velvet ropes outside an exclusive venue: effective barriers for those on the right side of them, invisible to everyone else.

What seems certain is that AI has fundamentally altered the relationship between identity, technology, and the law — and that the legal infrastructure needed to govern that relationship is still very much under construction.