Connecting decision makers to a dynamic network of information, people and ideas, Bloomberg quickly and accurately delivers business and financial information, news and insight around the world
Americas+1 212 318 2000
EMEA+44 20 7330 7500
Asia Pacific+65 6212 1000
Connecting decision makers to a dynamic network of information, people and ideas, Bloomberg quickly and accurately delivers business and financial information, news and insight around the world
Americas+1 212 318 2000
EMEA+44 20 7330 7500
Asia Pacific+65 6212 1000
Taylor Swift, one of many celebrities victimized by artificial intelligence, is fighting back against AI-generated sound clones that can now replicate voices to a remarkable level of accuracy.
TAS Rights Management filed trademark applications for Swift’s spoken phrases “Hey, it’s Taylor” and “Hey, it’s Taylor Swift” to leverage trademark law against AI voice cloning. It’s an area that the Lanham Act—the 1946 federal trademark statute—wasn’t designed to cover.
Several states have introduced deepfake-specific legislation to address the threat, but the lack of a federal deepfake statute and varying state right-of-publicity rights have prompted some brand owners and talent representatives to turn to the Lanham Act for a solution.
Doctrinal hurdles and limitations on enforcement will make registering sensory marks a challenge. In the context of these constraints, the sound mark registration strategy appears to be a stopgap providing some protection against AI deepfakes while the legislative landscape catches up.
A trademark can be any word, phrase, symbol, design, or a combination of these things that identifies the source of particular goods or services and has three main requirements.
Distinctiveness: All trademarks can be inherently distinctive or can acquire distinctiveness through evidence of secondary meaning. This can include longstanding use, advertising, media coverage, and consumer recognition.
The opening chimes and tones that accompany streaming platforms function as source-identifying trademarks in the form of sound marks. Much like a visual logo, a sequence of consistently used musical notes can create a distinctive auditory signature that consumers immediately associate with a particular service, reinforcing brand recognition and recall. These sounds operate as mnemonic devices that signal the origin of content before any words appear.
Under trademark law, such sounds are protectable when they’re sufficiently distinctive. If a sound is unique and unlikely to be perceived as commonplace, it may qualify for registration. But if it consists of ordinary or familiar tones, the applicant must demonstrate that consumers have come to recognize the sound as identifying a single source, typically through evidence of extensive and exclusive use, under principles akin to acquired distinctiveness.
A spoken phrase can function as a trademark only if consumers perceive it as identifying a single source, not just everyday speech. The more unique and consistently used the phrase, the more likely it is to qualify as distinctive.
Non-functionality: Section 2(e)(5) of the Lanham Act prohibits registering marks that are essential to a product’s use or purpose or whose exclusive use would place competitors at a significant non-reputation-related disadvantage.
The critical distinction is whether a sound is tied to the product’s utility or serves only to identify its source. For example, the tagline of a film is likely protectable as a trademark, as that tagline is used to promote the film to customers. Dialogue within the film itself that isn’t used in promotion is less likely to be protectable, as it is a functional part of the product.
Specimen requirements: Acceptable specimens for sound service marks may include advertisements, website content, or other materials showing “a direct association between the mark and the services.”
The Trademark Trial and Appeal Board’s 2023 decision in In re Duracell U.S. Operations, Inc. adopted a more flexible “aural equivalent” framework, holding that a sound mark played in-store can constitute an acceptable display associated with goods, even without physical affixation to packaging.
Taylor Swift’s pending applications illustrate these requirements in practice.
The first application covers the phrase “Hey, it’s Taylor,” supported by a specimen recording in which Swift opens with that phrase and continues with promotional language about a new album. The second covers “Hey, it’s Taylor Swift,” similarly supported by a recording that opens with that phrase followed by additional promotional messaging directed at a streaming platform.
The US Patent and Trademark Office will assess whether these opening phrases are inherently distinctive for entertainment services or require secondary meaning evidence. They also must demonstrate a sufficient direct association between the mark and the identified services. The more realistic vulnerability lies not in prosecution but in post-registration challenges.
Matthew McConaughey filed applications for his signature phrase, “alright,alright alright,” reflecting a greater concern within the entertainment industry about the lack of protection against deepfakes and other AI-powered digital cloning technologies.
When registered, a sound mark protects only against confusingly similar uses tied to related goods or services. This creates a key limitation in the deepfake context: an AI-generated voice that mimics a distinctive voice, but which doesn’t use the exact trademarked phrase, usually will fall outside the mark’s scope. For example, a deepfake of Taylor Swift endorsing a product may not necessarily infringe a mark limited to “Hey, it’s Taylor,” unless consumers hear substantially the same phrase used in a similar context.
The US has no federal law directly addressing AI voice clones, despite proposals such as the NO FAKES Act and NO AI FRAUD Act being introduced in Congress. Right of publicity laws, which appear to be the most natural fit for addressing AI cloning, traditionally are the purview of state legislators, creating an inconsistent patchwork.
As a result, sound marks aren’t effective standalone tools against deepfakes. Their protection is narrower than right-of-publicity claims, which cover the voice itself rather than a specific phrase.
While not a standalone solution for brand owners and talent representatives, sound marks can serve as part of a broader, layered enforcement strategy that complements state right-of-publicity claims.
Successful registration creates nationwide rights with availability of statutory damages and Customs recordation that generally are unavailable under state law. Sound marks should be combined with copyright protections, voice likeness restrictions, and contractual safeguards to address issues such as unauthorized AI training.
Applicants should focus on registering short, distinctive vocal phrases that are used consistently across promotions, building the evidence needed to show inherent or acquired distinctiveness. At the same time, they should maintain a broader portfolio of legal protections capable of addressing voice-cloning harms that fall outside the relatively narrow scope of sound mark rights.
This article does not necessarily reflect the opinion of Bloomberg Industry Group, Inc., the publisher of Bloomberg Law, Bloomberg Tax, and Bloomberg Government, or its owners.
Mitesh Patel and Charlie Rieder are associates in the emerging technologies and intellectual property groups at Reed Smith.
Jason Garcia is partner in the intellectual property group at Reed Smith.
Interested in writing? Review our author guidelines, and submit pitches to Insights@bloombergindustry.com.
Bloomberg Law provides trusted coverage of current events enhanced with legal analysis.
Log in to keep reading or access research tools and resources.
Leave a Reply