Can You Trademark Your Own Voice? Celebrities vs. AI Fakes
Celebrities such as Taylor Swift and Giorgia Meloni have filed their voices as sound marks to fend off AI clones. But how much protection does that really offer?
“Io sono Giorgia Meloni.” Those four words, spoken twice, make up the audio clip Italy’s Prime Minister filed as an EU trademark in early October 2026. Media reports were quick to supply the reason: Meloni, they said, wanted to protect her voice against AI fakes.
In doing so, she may have drawn inspiration from Taylor Swift. In April 2026, the U.S. pop star also filed applications for sound marks, among them a voice sample in which she says “Hey, it’s Taylor Swift.” Here too, the reported motive was protection against AI clones.
The wish to guard against AI fakes is easy to understand, especially for women in the public eye. But in both cases, the story as told in the media had a catch: under the current understanding of the law, a voice in its entirety – with its intonation and cadence – cannot be protected as a trademark.
A phrase is not a voice
What Meloni and Swift actually filed is a specific audio recording: one particular phrase, spoken at one particular moment. Once registered, protection would cover exactly that recording, along with signs that could be confused with it. However, it is not a blank check against every imitation of their voices.
What is a sound mark?
A trademark doesn’t have to be a word or a logo. Sounds can be protected too, such as jingles, signature tunes or short calls.
Requirements: The sound must be recorded in a clear and precise way, so that anyone can check exactly what is protected. Since the trademark law reform, an audio file is enough; musical notation is no longer required. And a sound mark, like any other mark, must be recognizable as a badge of origin. Generic or purely descriptive sounds and phrases cannot be registered.
Imagine an audio clip surfaces online shortly before an election, in which a politician appears to say something scandalous. A sound mark would be of little help here. It only covers the specific phrase that was registered – but only in commerce, only for the goods and services listed in the application, and only in the territory where it is registered.
What’s more, the registered recording never changes, but a person’s voice does – it evolves over a lifetime. The older the mark, the less it has in common with the voice an AI would imitate today. And finally, unlike personality rights, trademarks are tradeable assets: they can be sold, licensed or used as collateral.
Strong protection within narrow limits
A spoken call can still function as a trademark. This is shown by two examples from the U.S.: the rapper Pitbull has registered his trademark yell “EEEEEEEYOOOOOO” for sound recordings and live performances. The yell crops up in almost every one of his songs and anyone who hears it knows it’s a Pitbull track, just as the label on a bottle tells you who filled it. The same goes for Matthew McConaughey: his drawled “Alright, alright, alright” from the movie Dazed and Confused has long been his signature line, and in late 2025 he obtained a U.S. trademark registration for it.
That’s the crucial difference: a trademark has to show who a product or service comes from – just recognizing a voice is not enough. When Meloni says “I am Giorgia Meloni,” people think of the politician, not of a product. How the EUIPO will rule on her application, however, remains to be seen.
Sound Marks to listen to
Only 14 sound marks are currently in force at the Austrian Patent Office, including some very well-known ones such as the Almdudler call (“Aaaaalmdudler!”) and the Winzer Krems jingle (“Weinsame Klasse!,” a pun on “einsame Klasse,” roughly “in a class of its own” with a nod to wine).
All registered national sound marks can be played as audio files in see.ip, the Austrian Patent Office’s free trademark register.
Well-known international examples include Netflix’s “Tudum,” the roar of the MGM lion and the Looney Tunes theme. An overview of famous sound marks can be found on the website of the United States Patent and Trademark Office (USPTO).
How courts and lawmakers will deal with voices, faces and AI in the future is still an open question. The EUIPO, too, is currently examining the fundamental question of whether a human face, for example, can be a trademark. One thing is certain: the need to protect one’s own identity against AI misuse is only going to grow.