Taylor Swift's Trademark Filings and Who the Law Actually Protects
Earlier this year, Taylor Swift's company filed trademark applications for two short audio clips — "Hey, it's Taylor Swift" and "Hey, it's Taylor," spoken in her own voice, along with a visual mark covering one of her stage looks. Matthew McConaughey did something similar earlier this year, filing trademarks covering his voice and image specifically to get ahead of AI-generated deepfakes.
Trademark attorneys are calling this a new legal frontier. That's accurate, but it’s worth being precise about what's actually new. Sound marks themselves aren't novel (the NBC chimes and the MGM lion's roar have been registered for decades). Right of publicity laws, which exist in some form in most states, already address unauthorized commercial use of someone's name, image, or voice. What's untested is using trademark's source-identification doctrine — built to prevent consumer confusion about who's behind a product — to protect a celebrity's spoken voice as if it were a brand. It's a known legal category being stretched toward a new application, and the answer says less about Swift and McConaughey than it does about a recurring feature of American law: rights that exist in theory for everyone often become meaningful only for people who can afford to use them.
Trademark law was never designed to give someone ownership over a voice or a face. Its purpose is narrower and, historically, more public-facing: preventing consumer confusion about who stands behind a product or service. The Coca-Cola script, the Nike swoosh, the NBC chimes — these exist so that when you encounter them, you know what you're getting and who's accountable for it. That's the deal. Trademark isn't supposed to hand someone a monopoly on a sound or an image. It's supposed to protect the public from being misled.
Patents work on a similar logic, even though people rarely think of it this way. The constitutional basis for patent protection is "to promote the progress of science and useful arts" — the temporary monopoly is the incentive, not the point. The point is that the invention eventually benefits everyone. Intellectual property, at its foundation, is supposed to serve the public. The rights holder's benefit is the mechanism, not the goal.
What Swift and McConaughey's attorneys are doing is clever precisely because it stretches a public-protection doctrine to do something closer to a personal-property claim. Registering a voice as a sound mark has never been tested in court. Nobody knows yet whether it will hold up. But the move itself is a signal: when existing legal categories don't quite fit a new problem, people with enough resources will reshape the categories rather than wait for the law to catch up.
My guess is that if it's litigated, the protection will end up narrow, contingent on a high burden of evidentiary proof of actual consumer confusion, and full of carve-outs for parody, satire, and other protected uses, the same way most expansive trademark claims eventually get cut back to size. What's worth noticing instead is who gets to try this in the first place.
Most people don't realize that trademark rights are earned through use, not registration. You don't need to file anything with the USPTO to have certain protections, the same way copyright exists the moment you create an original work, no registration required. But the substantial benefits of registration — the presumption of validity, the ability to sue in federal court, the leverage to actually enforce the right against a well-funded opponent — cost money: filing fees, attorneys, ongoing global enforcement. Taylor Swift can absorb those costs without noticing, but most people whose voice or likeness gets deepfaked cannot.
If the only meaningful protection against AI-generated exploitation is a trademark filing that requires resources most people don't have, the law isn't protecting the public from a new harm. It's protecting whoever can afford to lawyer up first. The gap between formal rights and usable rights is widening, and every new technological disruption reveals who can afford to bridge it.
I didn't come to trademark law through law school. I came to it by spending nearly a decade inside a case that reshaped it. I learned the way most people do who end up living inside a legal system rather than studying it from outside: by discovering that rights on paper and rights in practice are not the same thing.
In Matal v. Tam, the disparagement provision of the Lanham Act required anyone seeking to register a name connected to their own identity to prove, to the government's satisfaction, that the name wasn't disparaging — even when the name belonged to the community in question and was being used to reclaim it. The burden of that proof was insurmountable for most people, not just financially but in terms of time, energy, and dignity. Despite having one of the largest appeals in U.S. history, with survey evidence, community leaders, dictionary experts, and more, I was not able to meet that burden. It took multiple appeals, years of legal costs, and the constant possibility that the government could simply decide an identity I was reclaiming wasn't acceptable. What that actually cost is its own story. The short version: the cost of defending your own identity against a government determination was a tax that fell disproportionately on the communities with the least power to pay it. Trademark-as-AI-defense risks becoming the same thing in a different shape: a tool that exists in theory for everyone, but that functions in practice as a privilege for whoever can afford to wield it.
This is where I think the conversation needs to go further than "should celebrities be allowed to trademark their voices." The deeper problem is that we're trying to solve a 21st-century harm with 20th-century property law, and property law was never going to be sufficient on its own.
I'll admit there's a version of this argument that should make anyone uneasy. Calling for new regulatory structures around AI risks handing more discretionary power to the same kinds of institutions that misused the disparagement provision in the first place. And simply asking IP law to expand further, without confronting the extraction at the root of the problem, risks treating a structural harm as something to be managed rather than challenged. Those are fair concerns, and I don't think the answer is more regulation for its own sake. I think it's regulation built with enough public participation and independent oversight that it doesn't just relocate the same concentration of power into a new form.
If trademark and copyright are the only tools available to address AI-generated exploitation, the law has already failed the public — because it means we've decided the solution is to let people purchase protection rather than to prevent the harm structurally. There are other places this could go instead. Defamation and fraud law already have frameworks for addressing knowingly false representations that cause harm; they could be extended rather than leaving everything to IP. Consent law has long recognized that agreements made under duress, or without full understanding of the consequences, aren't enforceable — that principle seems relevant to how AI companies currently obtain (or simply assume) rights to train on someone's voice, image, or creative work. Shifting the default from opt-in to opt-out, requiring genuinely transparent terms instead of buried legal language, and creating public oversight boards — especially for companies profiting from training on public and copyrighted material — would all do more than another decade of IP litigation.
Sam Altman has suggested that access to advanced AI should eventually be treated something like a utility. I'd push that further in a different direction: the knowledge these systems are built from, much of it created by ordinary people who never consented to have their work become training data, shouldn't be treated like a utility that a company can meter and monetize. It should be treated like a public resource — air, water, the commons — protected because it benefits everyone, not owned because it profits a few. The question isn't whether society benefits from shared knowledge; it always has. The question is whether companies should be able to convert that shared inheritance into private advantage without meaningful transparency, consent, or accountability.
The last few decades of IP law have generally favored people and institutions with the resources to defend themselves, sometimes at real cost to the public the law was originally designed to serve. Taylor Swift and Matthew McConaughey testing the limits of trademark doctrine is a clever and possibly effective short-term defense for two people who can afford excellent attorneys. It is not a solution.
Laws don't adapt to new technologies on their own. They change because people challenge them — through courts, legislatures, public comment periods, journalism, organizing, and voting. The question isn't whether AI will force us to rethink old rules. It already is. The question is whether ordinary people will have a meaningful voice in deciding what comes next.
I speak on First Amendment law, intellectual property, and the gap between legal doctrine and lived experience at law schools, bar associations, and civic organizations. If these questions connect to work you're doing, simontam.org/ip-law-keynote-speaker is the place to start.
