Taylor Swift, trademarks, and brand control in the age of AI

Taylor Swift has never been shy about taking control of her intellectual property. From her well‑documented battle over master recordings to an extensive global trademark portfolio covering everything from album titles to tour names and fan engagement, Swift has consistently treated IP as a strategic business asset rather than a legal afterthought. Her latest move has been to file US trademark applications for two vocal “sound marks” and one image mark; a further indication that she remains focused on brand control and is prepared to adapt her strategy in the age of generative AI.

In April 2026, Swift’s company, TAS Rights Management, filed three applications with the US Patent and Trademark Office. Two relate to short spoken phrases in her own voice: “Hey, it’s Taylor Swift” and “Hey, it’s Taylor.” The third is a distinctive image showing Swift on stage holding a pink guitar during her Eras Tour. The filings have been widely reported as a response to the growing risk of AI‑generated deepfakes and unauthorized digital replicas: they are aimed at providing an additional layer of protection against misuse of Swift’s identity.

Trademarks have traditionally been used to protect names, logos, and slogans. They are indicators of commercial origin that help consumers understand where their goods and services come from. Sound marks exist, but they are relatively rare and often cover jingles or short audio cues, such as the Netflix ‘Ta-Dum.’ Registering the sound of a person speaking is an unusual application of the system, but it is not without precedent: in January, actor Matthew McConaughey secured a sound trademark in the US for his iconic catchphrase “Alright, alright, alright.”

What value do these sound trademarks really have? Can we expect to see a slew of applications for trademarks covering varying elements of personality rights, or is Taylor Swift’s move a public announcement of intent to protect her image rather than a sound legal strategy?

The answer, in my opinion, is that sound marks achieve both of those things. If an AI model generates new audio that sounds like Taylor Swift without copying an existing recording, a copyright infringement claim would be difficult to sustain. Similarly, image rights (where they exist) are fragmented and limited. In the US, rights of publicity vary from state to state, and protection often focuses narrowly on commercial endorsement rather than broader reputational harm. In the UK, there is no image right, meaning that celebrities such as Rihanna have sought to rely on passing off to enforce rights in their image.

Trademark law offers something different: by registering specific vocal phrases and a recognizable performance image as trademarks, Swift may be able to argue that an AI‑generated voice or image that is confusingly similar to her registered trademarks, or that takes unfair advantage of her reputation, infringes her trademark rights. Registered rights make social media takedowns much faster, so they have a practical as well as a legal value.

That said, trademark protection is not a panacea. Swift’s applications are currently limited to the US, leaving significant gaps in territorial coverage. Trademarks are national (or regional) rights, and a US registration does little to address AI‑generated misuse taking place elsewhere, particularly in jurisdictions with weaker enforcement frameworks. For an artist with a genuinely global fanbase and digital footprint, this creates an obvious mismatch between risk and protection.

As generative AI makes it easier to create convincing imitations without copying existing works, we are likely to see trademarks increasingly being used creatively to police misuse of voice, image, and identity where copyright or image rights fall short. While we are yet to see alleged infringement of these types of trademarks tested in court, Swift and McConaughey’s applications reflect a pragmatic response to a rapidly evolving threat landscape.

← Back to News