Protecting Celebrity Persona From AI

By Thomas P. Krzeminski and Andrew Fender 

The next major false endorsement dispute may not begin with a counterfeit handbag, a knockoff logo, or a misleading product label. It may begin with a synthetic voice, a cloned face, and a fifteen-second advertisement that millions of consumers believe is real. 

Generative artificial intelligence (“AI”) has made it possible to manufacture realistic celebrity endorsements, performances, social-media posts, and promotional content without the celebrity ever stepping into a studio or signing a license agreement. That development changes the practical stakes of trademark and brand-protection law. When AI can convincingly reproduce a person’s commercial identity, which legal strategy and tools are best positioned to prevent confusion and unauthorized monetization? 

A Non-Traditional Approach to Trademark 

The issue of generative AI’s reproduction of a likeness or image is different from an ordinary impersonation. Trademark law includes carveouts for uses where there is no likelihood of confusion, or where the use is parody, satire, criticism, or commentary, among others.¹ Generative AI presents a developing challenge: cheap, scalable, and believable AI impersonation requires that personal identity be treated as a part of a public figure’s protectable IP asset portfolio. 

A number of celebrities have already taken note of the growing threat that AI poses to the commercialization of their likeness and endorsement, and have taken proactive steps to protect their image in a non-traditional way. For example, Taylor Swift recently filed a number of trademark applications related to her voice and image specifically designed to protect her from the growing threat posed by artificial intelligence.² Swift is not alone in this strategic move. Other actors, such as Matthew McConaughey, have also registered trademarks for iconic sounds or gestures.³  

The protection that comes with these trademarks will likely allow for a clearer delineation of rights when generative artificial intelligence tools are used to reproduce these depictions iconically associated with Swift and McConaughey. If registered and used as source-identifying assets, these marks may provide an additional enforcement theory where AI-generated content creates likely confusion as to sponsorship, approval, affiliation, or endorsement.⁴ 

Trademarks related to a persona are relatively rare, but the new age of generative AI makes them all the more relevant. Musical artists, such as Swift, could in the past rely on copyright to protect the content of their music. AI’s adaptation of a voice or image in the production of a convincingly realistic depiction incorporating elements of a persona causes mere copyright protection to fall short. It is no longer only the lyrics—or even Swift’s voice singing those lyrics—that requires protection; the unique characteristics that the public now associates with musicians and others must also be protected. Without protection, many AI models can be used to quickly and easily generate a convincing portrayal of endorsement using a public figure’s unique characteristics or performance markers.  

Using Trademark to Protect Persona Assets 

Protecting voice or other recognizable gestures and images will become an increasingly important tool in preventing false endorsements or affiliations. It is not likely that every recognizable voice, face, or mannerism can automatically be registered as a trademark. Instead, the trademark protection to consider here involves protecting the persona of an individual, rather than merely identifying a person. The stronger trademark strategy is therefore to identify persona elements that consumers already understand as signaling source, sponsorship, approval, affiliation, or endorsement.⁵ 

Certain trademark assets are more likely to support that function than a generalized likeness claim. For example, a celebrity’s name or stage name may operate as a brand when used consistently across entertainment services, merchandise, endorsements, fan-facing commerce, or licensed products.⁶ Other public figures may consider protecting catchphrases, slogans, recurring taglines, or distinctive social media handles where their persona is associated with goods or services provided. 

The rapid pace of advancement in generative AI technology makes trademark protection increasingly important because a public figure’s persona can now be copied, scaled, and commercialized with unprecedented speed and realism. Trademark law is especially suited to address these harms where the copied persona elements function as source identifiers like those noted above. In that setting, trademark protection does more than protect personal identity; it protects consumers from confusion and safeguards the public figure’s brand from dilution, false association, and loss of licensing control. 

Rather than asking only whether an AI output resembles a public figure, counsel should consider which persona elements have been commercialized, whether they are used consistently as brand signals, whether they can be registered or documented as common law marks, and whether the challenged AI content is likely to make consumers believe that the celebrity sponsored, approved, endorsed, performed in, or is otherwise affiliated with the product or message. 

Considerations for Asset Protection Strategy 

We have yet to see how courts will approach the issue of AI replicating and commercializing persona elements. Swift and McConaughey are among the early movers to seek trademark protection for their unique mannerisms, style, and voice. These trend setters may therefore be at the forefront of the next major legal showdown that will define the rights associated with trademarks related to persona elements. In the meantime, below are practical pointers that will help practitioners evaluate whether to seek trademarks for their public figure clients.  

The first step is a persona audit. Counsel should identify the client’s protectable commercial identifiers such as names, stage names, signatures, logos, slogans, social-media handles, recurring visual branding, distinctive voice elements, and performance cues and determine which are consistently used in commerce. Not every aspect of fame or appearance will function as a trademark, so the strongest strategy is to isolate persona elements that operate as brand signals, not merely personal attributes. 

Those assets should then be registered where possible, strengthened through consistent commercial use, and documented as common law marks where registration is unavailable. Trademark filings and brand records should anticipate AI misuse by covering media contexts. Counsel should also preserve evidence of consumer recognition, licensing history, endorsement practices, and marketplace association. 

Trademark protection should be integrated into contracts and enforcement programs. Licenses, sponsorship agreements, and vendor contracts should address synthetic media, voice cloning, digital doubles, AI training, approval rights, and post-termination use. Monitoring should likewise extend beyond traditional counterfeits to AI-generated ads, deepfake endorsements, synthetic voiceovers, social-media impersonation, avatars, and chatbots. 

Don’t wait for a viral AI-generated endorsement to appear. A proactive trademark strategy gives counsel a clearer enforcement theory and will allow for anticipation of future market trends. 

Thomas P. Krzeminski - Tom Krzeminski is a first-chair trial attorney with over 20 years of experience litigating complex patent and trade secret cases in federal courts across the country. Tom has substantial expertise representing pharmaceutical clients in Hatch-Waxman litigation, often on the generic side. In addition to his Hatch-Waxman litigation practice, Tom has experience litigating patent and trade secret cases across many technologies, as well as litigating Section 337 actions before the U.S. International Trade Commission (ITC) in Washington DC. 

Andrew Fender - Andrew’s deep experience in both local and global business operations affords him strong insight into the pivotal role of intellectual property, trade secrets, and copyrights in driving and protecting business success. As a litigator, he is a relentless advocate for protecting these valuable business assets. Knowing that IP litigation is never one-size-fits-all, Andrew tailors his approach to each client’s unique challenges, pursuing the most effective path to safeguard their business interests. 

Maschoff Brennan - Maschoff Brennan provides legal counsel and representation to some of the world’s most innovative companies. With over 50 attorneys and offices in the technology-focused regions of Utah, California, and New York, our attorneys are known for having the breadth of experience and the forward-thinking insight needed to handle complex technological and business issues across all industries and geographic boundaries. In addition, we have extensive experience representing clients before the ITC, PTAB, TTAB, and other administrative agencies in Washington D.C.