Deepfakes, Disclosures, and “Alright, Alright, Alright”: New York’s AI Ad Law and the Rise of Persona Trademarks

I. Introduction

Generative artificial intelligence (AI) can now convincingly replicate voices, faces, mannerisms, and other personal attributes, enabling the creation of unauthorized content that can imitate humans generally and individuals specifically. Critically, AI need not copy an existing recording or image; it can generate new content that convincingly imitates a generic human or a specific individual’s voice or likeness—with the former such generation likely falling outside traditional copyright and former right of publicity protections. Two notable responses to this gap have emerged. First, New York¹ has enacted a Synthetic Performer Disclosure Law that requires advertisers to conspicuously disclose their use of AI-generated “synthetic performers,” and pending legislation would extend that requirement to certain digital replicas of real individuals. Second, celebrities and, increasingly, athletes are turning to federal trademark law, including sound and motion mark filings, to claim rights in their signature voices, gestures, and catchphrases, all to protect them against AI replication.

II. New York’s Synthetic Performer Disclosure Law

Overview of the Statute

New York’s Synthetic Performer Disclosure Law, enacted June 9, 2026, requires anyone producing advertising content to conspicuously disclose the use of AI-generated “synthetic performers” when the advertiser knows such performers are being used.² Under the statute, a “synthetic performer” is a digital asset created, reproduced, or modified by computer through generative AI or a software algorithm and intended to give the impression of a human performer engaging in a visual or audiovisual performance, where that performer is not recognizable as any identifiable real person.³ Three features of the statute merit particular attention:

1. The statute regulates “assets,” not the full person. Because the definition turns on a digitally created “asset,” it may reach isolated human features, such as a synthetic voice, used in an audiovisual or visual advertisement.
2. The statute covers a broad set of actors. The law applies to any person engaged in the business of “dealing in any property or service” that “produces or creates” an advertisement, in any medium, for a commercial purpose.⁴ This includes brands, agencies, and production studios.
3. The statute reaches only fictional performers. Because the statute covers only synthetic performers who are not recognizable as real people, it does not reach deepfakes of actual individuals. But that may soon change. Assembly Bill A11683 would eliminate the unrecognizability requirement, extending the disclosure obligation to digital replicas of real people, unless authorized by the rights holder or permitted by law.⁵ The bill would also require audiovisual disclosures to remain visible whenever a synthetic performer appears.

Violations carry civil penalties of $1,000 for a first offense and $5,000 for each subsequent offense.⁶

Exemptions and Open Questions

The law exempts (i) ads for expressive works, such as films, television, streaming documentaries, and video games, where the use of synthetic performers is consistent with the work; (ii) audio-only ads; (iii) AI used solely to translate a human performer’s language; and (iv) media outlets that publish or distribute third-party ads. Further, because the law covers only synthetic performers, other common AI uses in advertising, such as content enhancement or nonhuman AI-generated content, do not trigger disclosure.

Several questions surrounding application and enforcement remain open:

1. What counts as “conspicuous.” The statute does not prescribe disclosure language or placement. Commentators suggest wording such as “AI-generated performer” and caution that vague phrases like “digitally enhanced” may fall short.⁷
2. Whose knowledge counts. The duty to disclose applies only when a business has “actual knowledge” of synthetic performer use. The original bill reached knowledge held by the person “or agent or employee thereof,” but the enacted version requires that “such person has actual knowledge.”⁸
3. Territorial reach. The statute is not limited to New York-based companies, and commentators generally read it to cover advertisements shown to New York audiences. National digital campaigns will likely need to comply nationwide rather than attempt to carve out New York.

New York in the National Landscape

New York’s law joins a growing wave of state legislation addressing AI-generated likenesses. For example, Tennessee’s ELVIS Act extends publicity protection to AI-simulated voices,⁹ and California now expressly covers unauthorized AI-generated digital replicas of deceased individuals.¹⁰ Where those laws target unauthorized AI-generated digital replicas, New York’s statute focuses on transparency in AI-generated commercial content.

III. Celebrities and Athletes Turning to Sound and Motion Trademarks

As AI-generated content increasingly exploits celebrities’ and athletes’ voices and images, some celebrities and athletes are filing sound and motion mark applications to secure federal trademark rights in elements of their persona, such as voices and catchphrases. Because a trademark (any word, name, symbol, or device used to identify and distinguish goods or services) is used to indicate commercial source rather than personal identity, in securing a trademark registration for an element of persona, the applicant and application must tie such element to specific commercial activities.

Filings to Date

Matthew McConaughey’s eight federal registrations, including for his catchphrase “Alright, alright, alright” and motion marks tied to his gestures, offer a model for the registration of sound and motion marks.¹¹ By linking his mannerisms and delivery to Class 9 goods (downloadable audio-visual media content in the field of entertainment) and Class 41 (providing entertainment services), these registrations convert elements of his persona into federally recognized source identifiers. Each registration is prima facie evidence that the mark is valid, that McConaughey owns it, and that he has the exclusive right to use it for the listed goods and services. That shifts the burden to any challenger to prove otherwise. Once the registrations become incontestable after five years of continuous use, a challenger will no longer be able to argue that McConaughey’s catchphrase or gestures are mere elements of his performance rather than indicators of source, which is the most common objection to persona-based marks. In practice, this means McConaughey can assert his trademark rights against AI-generated imitation of his registered voice or gestures without first having to prove that the public sees them as his brand. Others are following his lead. Taylor Swift has three pending applications, including for the phrase “Hey, it’s Taylor” and an image of her performing in a signature outfit from her world tour,¹² and Jimmy Kimmel has applied to register a sound mark and on-air images closely associated with his identity.¹³ How the USPTO handles these applications will signal how examiners view persona-based non-traditional marks.

Athletes are turning to trademark law as well. With the rise of name, image, and likeness (NIL) opportunities, some athletes are filing for trademark protection for names, nicknames, and other brand identifiers earlier in their careers. Ahead of the NFL Draft, for example, quarterback Fernando Mendoza filed 12 applications covering his name and nickname, including “MENDOZA FERNANDO,” “FLIPPIN’ HE15MENDOZA,” and “MENDOZA15.”¹⁴ Athletes are also claiming marks tied to signature footwear and apparel, an area manufacturers have traditionally controlled.

The next step may be motion and sound marks for touchdown dances, goal celebrations, and similar gestures. Cole Palmer, a midfielder for Premier League club Chelsea, is an early test case: after registering his signature “shiver” celebration—where Palmer crosses his arms and rubs his hands up and down his biceps as if he is shivering—in the United Kingdom, he applied to register it in the United States.¹⁵ Like McConaughey’s motion marks, his application rests on the theory that a gesture closely associated with an individual can identify source, and its outcome may signal how readily athletes can protect such gestures.

Limitations of Trademark Protection

Still, trademark law offers only limited protection against AI replication. For instance, a motion mark that is not inherently distinctive requires proof of acquired distinctiveness, or secondary meaning, through use in commerce, and the USPTO requires that nontraditional marks be nonfunctional and promoted long enough for consumers to view them as source identifiers. That burden is heavy for common sounds and motions. Palmer, for example, may struggle to show that consumers associate his celebration exclusively with him when other athletes perform similar ones. Further, registration alone does not create rights in the United States: the mark must be used in commerce as a source identifier, and because trademark law targets consumer confusion, it may not reach noncommercial AI content, such as memes and parody, which the First Amendment or fair use may shield. Accordingly, for most corporate brands, the added value of a sound or motion mark may be modest, unless the company has invested in audio or motion branding and can point to consumer recognition.

IV. Practical Steps

The foregoing trends carry practical governance and diligence implications for advertisers, talent, and the companies that work with them.

Advertisers, Agencies, and Media Companies

• Audit advertising practices. Sort current and planned creative work into three categories (fictional synthetic performers, identifiable living persons, and deceased New York performers) to determine which disclosure requirements apply. A real-person deepfake is not safer merely because it falls outside the disclosure requirement; it shifts the risk to privacy and publicity claims and, if A11683 passes, could trigger disclosure obligations as well.
• Adopt a standard disclosure and apply it nationally. Use plain-language wording such as “AI-generated performer,” rather than vague phrases like “digitally enhanced.”¹⁶ Because the statute likely reaches ads viewed in New York, a uniform standard across national campaigns is usually the more practical approach.
• Update agency, production, and AI vendor contracts. Require counterparties to represent whether generative AI was used in any deliverable and to identify each synthetic performer or digital replica, so that the advertiser has the knowledge needed to comply. Contracts should also require that disclosures be delivered intact through editing and distribution, assign responsibility for disclosure compliance, and include indemnities for publicity, false endorsement, and consent claims.

Consumer Brands and In-House Counsel

• Expand clearance and identify protectable assets. Monitor USPTO sound and motion mark filings by celebrities, athletes, and competitors, update clearance procedures for audio and visual branding, and assess whether existing sonic logos, product sounds, or motion sequences have enough consumer recognition to support registration.
• Review imitative AI content and agency compliance. Because AI content evoking a specific public figure can create infringement exposure without copying any specific images or audio, route such campaigns through legal review before launch.

Talent, Sports Clients, and Their Representatives

• File early and strategically. Target persona elements used in commerce and tied to specific goods and services and gather evidence of consumer recognition before filing trademark applications. For celebrations or gestures, first assess whether they are distinctive enough to acquire secondary meaning.
• Layer protections. Supplement trademark registrations through licensing, endorsement, and NIL agreements that expressly define the protected persona elements (including NIL, signature phrases, and celebrations), grant or reserve publicity rights, restrict unauthorized imitations and look-alike or sound-alike uses, and require approval rights over any use of the persona.
• Address AI in NIL and endorsement agreements. When drafting or negotiating these contracts, define permitted digital replica uses by media, duration, and territory, require separate consent and compensation for each new use, and allocate disclosure responsibility under the Synthetic Performer Disclosure Law and, if enacted, A11683.

Deal Counsel and Investors

• Diligence persona and AI rights. In media, consumer, and sports transactions, review sound and motion mark portfolios, the chain of title for talent-related marks, AI use in content creation and traditional advertising, and synthetic performer disclosure practices.

V. Conclusion

Generative AI has outpaced the legal frameworks built to protect identity, and the gap will not stay open for long. New York’s Synthetic Performer Disclosure Law shows that legislatures are moving from debate to enforcement, and pending legislation signals that disclosure obligations may soon reach digital replicas of real people. At the same time, the McConaughey registrations and the wave of celebrity and athlete filings that followed show that voices, catchphrases, and gestures can be claimed as protectable brand assets, and that the companies using them without permission may face claims they did not anticipate. The rules are being written in real time, but the direction is clear: transparency will be expected, consent will be required, and persona will be treated as property. Organizations that treat these developments as a present obligation rather than a future concern (i.e., by inventorying their AI-generated assets, updating their contracts, and building disclosure and clearance into the creative process) will be ready for the next wave of disclosure mandates and persona-based claims, rather than reacting to them after the fact.