ADventures in Law

Home | ADventures in Law

Hello from your friendly human lawyers.&nbsp;If your marketing team is using generative&nbsp;artificial intelligence (AI)&nbsp;in advertising, the issue is no longer whether&nbsp;the&nbsp;rules apply. It’s which rules apply, when they are triggered, and how&nbsp;best to&nbsp;navigate&nbsp;them.&nbsp;

core/paragraph

This&nbsp;analysis becomes more concrete on&nbsp;June 9, 2026,&nbsp;when New York’s synthetic-performer disclosure law takes effect.&nbsp;At the same time, contract law, labor rules, and social media platform policies are all shaping how brands create, track, and distribute AI-generated (or enhanced)&nbsp;creative content.&nbsp;And we’re here to help make sense of it all.&nbsp;

core/heading

Before we dive in, let’s take a moment to first define some key terms:  

Synthetic Performer: A digitally created asset that (1) is intended to and does create the clear impression that the asset is a human performer (one that is not recognizable as an actual identifiable human performer), (2) is not voiced by a human performer, and (3) is not a digital replica (see SAG-AFTRA’s definition <a href="https://www.sagaftra.org/sites/default/files/2025-05/2025%20Commercials%20Contract%20MOA.pdf" target="_blank" rel="noreferrer noopener">here</a>).

core/list-item

Digital Replica: A digital simulation of a human individual’s voice, image and/or likeness, depicted in a manner that so closely resembles the individual’s real voice, image, or likeness, that the average person could not distinguish the two.

Generative AI: A subfield of AI that uses generative models of machine learning to create content, including audio, code, images, text, simulations, and videos based on patterns in data. Some examples include OpenAI’s ChatGPT, Anthropic’s Claude, Microsoft’s Copilot, Google’s Veo, and ByteDance’s Seedance.

core/list

State laws in New York, California, and Tennessee may impact the creation and dissemination of ads that include synthetic performers, digital replicas, and other assets produced by generative AI. 

Synthetic Performer Disclosure Law (N.Y. Gen. Bus. L. § 396-b) 

On June 9, 2026, New York’s&nbsp;<a href="https://www.nysenate.gov/legislation/laws/GBS/396-B" target="_blank" rel="noreferrer noopener">Synthetic Performer Disclosure Law</a>&nbsp;takes effect. Hailed as the “first in the nation,” the law requires advertisements&nbsp;in New York to&nbsp;include&nbsp;a&nbsp;conspicuous disclosure&nbsp;when&nbsp;they&nbsp;feature&nbsp;digitally&nbsp;created images of a human who is not recognizable as any identifiable natural performer. In other words, when an ad contains a synthetic performer, it must be accompanied by a clear and conspicuous disclosure.&nbsp;

Importantly, the disclosure requirement is not triggered simply&nbsp;by the use of&nbsp;AI&nbsp;or even generative AI on its own.&nbsp;Instead&nbsp;the trigger is the use of “any software algorithm”&nbsp;–&nbsp;including generative AI&nbsp;–&nbsp;to create a synthetic performer that appears in the ad.&nbsp;By way of example, if an ad includes a digitally created background or a digitally created car, the disclosure requirement would not apply. However, if that same ad also included the image of a man driving the car,&nbsp;which image had been wholly created by a generative AI tool,&nbsp;the disclosure requirement would be triggered.&nbsp;&nbsp;

While the law does not dictate where and how such disclosure should appear, the use of the term “conspicuous” logically invokes FTC guidance. Clear and conspicuous disclosure should be difficult to miss and&nbsp;in close proximity to&nbsp;the ad.&nbsp;Failure to comply with the disclosure requirements may result in civil penalties of $1,000 for a first violation and $5,000 for each subsequent violation.&nbsp;

New York State Fashion Workers Act (N.Y. Labor Law, Article 36)

In June 2025, the New York <a href="https://www.nysenate.gov/legislation/laws/LAB/A36" target="_blank" rel="noreferrer noopener">Fashion Workers Act</a> (NYFWA) went into effect (we wrote about it previously <a href="https://www.bakerlaw.com/insights/supermodel-werk-new-yorks-fashion-workers-act-takes-effect/" target="_blank" rel="noreferrer noopener">here</a>), creating workplace and wage protections for models who provide services in New York, which includes any individual who appears in a photo shoot, runway show, or other live or recorded performance (including on social media). For digital replicas, advertisers must:

Obtain clear and conspicuous written consent directly from the model (and not the agency) in a separate signed writing, which must include scope, purpose, pay rate, and length of time for which the digital replica will be used.

Obtain separate written consent for each use or creation of a digital replica.

Violations of the NYFWA are subject to a civil penalty of $3,000 for the first violation and $5,000 for each subsequent violation. Additionally, the NYFWA creates a private right of action that may allow models to seek actual damages, attorney’s fees, and liquidated damages. 

Contracts for the Creation and Use of Digital Replicas (N.Y. Gen. Oblig. Law § 5-302 and Cal. Labor Code § 927(a)) 

On Jan. 1, 2025, digital replica laws in <a href="https://www.nysenate.gov/legislation/laws/GOB/5-302" target="_blank" rel="noreferrer noopener">New York</a> and <a href="https://legiscan.com/CA/text/AB2602/id/3021235/California-2023-AB2602-Chaptered.html" target="_blank" rel="noreferrer noopener">California</a> went into effect. These laws aim to prevent individuals from contracting away their rights without (i) adequate representation and/or (ii) receiving clear disclosures about how a digital replica will be used. They further render void and unenforceable any contractual provision for the use or creation of a digital replica if:

The provision does not include a reasonably specific description of the intended use of the digital replica; and

The contracted individual was not represented either (a) by legal counsel where the licensing terms are clearly and conspicuously stated in a separately signed writing or (b) by a labor organization (e.g., SAG-AFTRA) where the terms for the use of digital replicas are expressly addressed in a collective bargaining agreement. 

Importantly, as these are labor laws, they apply to any individual (not just union talent) hired in New York or California.  

ELVIS Act (Tenn. Code § 47-25-1101 et seq.) 

On July 1, 2024, Tennessee’s Ensuring Likeness, Voice, and Image Security Act (<a href="https://publications.tnsosfiles.com/acts/113/pub/pc0588.pdf" target="_blank" rel="noreferrer noopener">ELVIS Act</a>) took effect. It extends right of publicity protections generally reserved for name, image, and likeness (NIL) to an artist’s vocal performance and makes “voice” a protected attribute under the state’s right of publicity law. Specifically, the ELVIS Act prohibits the use of generative AI to create a vocal digital replica of an artist without the artist’s explicit written consent. Violations of the ELVIS Act can result not only in civil lawsuits but also in criminal prosecution as a Class A misdemeanor, which could carry fines of up to $2,500 and potential jail time.

The 2025 <a href="https://www.sagaftra.org/sites/default/files/2025-05/2025%20Commercials%20Contract%20MOA.pdf" target="_blank" rel="noreferrer noopener">SAG-AFTRA Commercials Contract</a> (operative through March 2028) establishes a framework for the creation and use of a performer’s digital replica and provides guidelines for the use of a synthetic performer. It’s worth noting that these requirements do not apply to the use of generative AI solely in connection with traditional post-production edits.

To create a digital replica of a performer, an advertiser (or its third-party SAG-AFTRA signatory) must follow some protocols, including: 

Providing at least 48 hours notice to performer

Obtaining clear consent in a separate signed writing (a rider will suffice) from the performer to create or use a digital replica. The writing must include a reasonably specific description of both the generated performance and the anticipated use of the digital replica.

Obtaining additional written consent for any use of the digital replica that (a) falls outside the initial description, (b) materially alters the performer’s physical characteristics (e.g., hair color, voice) or (c) depicts the performer in the nude (full or partial).

While featuring a synthetic performer does not require advance consent, it may require additional payments to SAG-AFTRA for any use, depending on whether the advertiser is a signatory. The amount of payment will vary depending on whether another human appears in the commercial.  

Even where state law does not require disclosure, social media platforms often do. For example, <a href="https://support.tiktok.com/en/using-tiktok/creating-videos/ai-generated-content" target="_blank" rel="noreferrer noopener">TikTok</a> requires labeling for realistic AI-generated content, <a href="https://support.google.com/youtube/answer/14328491" target="_blank" rel="noreferrer noopener">YouTube</a> requires disclosure for altered or synthetic content that appears real, and <a href="https://www.facebook.com/business/help/1010479435004531" target="_blank" rel="noreferrer noopener">Meta</a> automatically applies labels and expects disclosures where AI materially alters or generates content.

AI Product Imagery: Disclosure is not enough

And keep in mind that even clearly labeled AI-generated visuals can create risk if they imply something untrue about a product. For example, AI-generated depictions that exaggerate texture, performance, or scale of a product may be considered misleading, regardless of any “AI-Generated” disclosure that may accompany it. Be mindful – if your AI-generated product demonstrations look too good to be true, you face risk. 

To help tie this all together, we’ve outlined some scenarios that could apply to your business. 

Q: We have historic print ads featuring models who granted us perpetual rights. Can we reuse them and animate then as digital replicas via our generative AI tools?

A: Not necessarily. If the original agreements did not specifically contemplate AI-generated reuse or new forms of performance, those rights may not carry over. You would need to get renewed, specific consent from the model. 

Q: If we disclose “GenAI Ad,” are we protected if the product depiction is not accurate?

A: No. Disclosure does not cure a misleading impression. If the ad creates a false impression about the product, liability may still arise under consumer protection law. 

Q: If we use a fictional AI spokesperson, do we need a disclosure? We did not use any real person as inspiration.

A: Yes. If the character appears realistic and humanlike, disclosure is required for use of a synthetic performer in New York as well as on most social media platforms. 

A: Many agencies are on top of generative AI developments, including disclosures. That said, the advertiser will be targeted in the event of a misstep. Expectations should be clear in your agency agreements, and you will want to have mutually agreed guidelines in place.

Q: What if we want to use an asset but don’t know if it was created using generative AI?

A: Ideally, you won’t be in this situation. It’s critical to adopt commonsense protocols and technology to label and track your assets internally now so you can determine what rights and disclosures are required. Share these requirements with your agencies so you use the same approach with raw assets and final deliverables.

While these regulations do not necessarily impose onerous obligations, they do require diligence. So keep them in mind when contracting with on-camera and voice-over talent as well as when disseminating advertisements that contain digital replicas and synthetic performers.  

Contracts, Consent, and Clones: Navigating Requirements for AI-Generated Advertising