AI Advertising: Disclosure Obligation
The integration of Artificial Intelligence (AI) into advertising campaigns is increasingly raising compliance questions. Particularly in the USA, concrete legal regulations for the labeling of AI-generated content in advertising are developing. What was once considered creative freedom is now becoming a matter of transparency and consumer protection through laws and guidelines.
Source: Own Representation
AI Labeling in Advertising
In the USA, the obligation to label AI-generated content in advertising is becoming more concrete. New York signed a law in December 2025 that mandates labeling for advertising featuring AI "synthetic performers" ( governor.ny.gov). ). This law, S.8420-A/A.8887-B, was signed by Governor Kathy Hochul on December 11, 2025, and is known as "SIGNED CHAP.617" ( assembly.state.ny.us).
The core of the law is that advertising in New York that includes an "AI generated synthetic performer" must disclose this ( governor.ny.gov). ). A "synthetic performer" is defined as a digitally generated asset that uses generative AI or algorithms to create the impression of an audiovisual or visual performance by a human performer, without a real, identifiable person behind it ( assembly.state.ny.us).
This does not only apply to classic deepfakes where real people are manipulated, but also to completely fictional avatars that appear lifelike ( nyassembly.gov). ). Current advertising use cases such as virtual brand ambassadors, AI models in fashion, and synthetic creator faces for always-on ads are affected ( sagaftra.org).
Violations of this law can result in civil penalties: $1,000 for the first violation and $5,000 for subsequent violations ( nyassembly.gov, nyassembly.gov). This underscores the seriousness with which New York is handling transparency in AI-generated advertising.
For example: A DTC brand creates a performance creative for Instagram Reels with an AI avatar, lip-synced animation, and synthesized voice. Without labeling, this would be perceived as a genuine testimonial clip. New York requires that the use of synthetic performers must be recognizable to consumers ( governor.ny.gov).
Legal Aspects
Besides New York, there are other legal developments influencing the labeling of AI in advertising. The US consumer protection agency FTC (Federal Trade Commission) has for years pursued the principle that advertising must not be misleading. In September 2024, the FTC made it clear with "Operation AI Comply" that there is no special zone for AI: "Using AI tools to trick, mislead, or defraud people is illegal" ( ftc.gov). ). The FTC has already initiated several proceedings related to AI claims, fake reviews, and misleading promises.
Source: ciso.economictimes.indiatimes.com
For influencer ads, the FTC Endorsement Guides were updated in June 2023 ( ftc.gov). ). These state that claims that appear to be personal testimonials but are marketing require clear and conspicuous disclosure ( ftc.gov). ). The legal basis is found in 16 CFR Part 255 ( law.cornell.edu).
The AI effect amplifies this issue: an AI avatar can appear like an influencer without being one. A synthetic voice-over artist can sound like a real expert without being one. An AI-generated scene can look like documentary material without it ever having happened. This means that the question of AI labeling in the USA is no longer just an ethical debate, but an operational necessity, especially where laws like New York's Chapter Law apply ( assembly.state.ny.us).
Another important aspect is the topic of "consent", especially concerning voices and real people. Tennessee, with the ELVIS Act (Ensuring Likeness, Voice, and Image Security Act), significantly strengthened the protection of voice and likeness in March 2024 ( publications.tnsosfiles.com). ). Tennessee is thus the first US state to specifically introduce protection for musicians against AI imitation ( apnews.com). ). Reuters classifies this as a reaction to generative AI cloning models ( reuters.com).
New York is drawing a post-mortem line in parallel: the law S.8391/A.8882, also signed on December 11, 2025, requires consent from heirs or estate administrators if the name, image, or likeness of a deceased person is to be used commercially ( governor.ny.gov). ). This is relevant for retro campaigns and "virtual revivals" that have become technically cost-effective ( copyright.gov).
At the federal level, there have been drafts since 2024/2025 such as the NO FAKES Act (S.4875) ( congress.gov) ) or the No AI FRAUD Act (H.R. 6943) ( congress.gov), ), which aim to address "digital replicas" and imitations as a separate right or tort. This shows a shift in expectations from "it can be done" to "one must prove one is allowed to do it".
Practical Implications for Companies
The clear line is not between "AI" and "non-AI", but between "transparent" and "misleading". New York is forcing the industry to translate this line into advertising materials so that it is understood even at a rapid scrolling pace ( nyassembly.gov).
Source: brandma.in
In practice, this means that as soon as a creative simulates a human performance (avatar, synthetic person, AI actor, digital presenter), companies must establish a fixed decision point internally: Is this considered a "synthetic performer" and how will it be labeled in the respective channel? ( governor.ny.gov). ). For influencer collaborations, a second decision point is added: Is it advertising (material connection), and is the disclosure clear and conspicuous, regardless of whether the face is real or synthetic? ( ftc.gov).
Source: insideglobaltech.com
Cases where AI creates an "impression of evidence" are particularly sensitive: before/after videos, supposed user reactions, "street interviews", or testimonials with a genuine appearance. If fake reviews are created from this or tools for it are sold, one quickly enters the area that the FTC explicitly addresses in its AI enforcement ( ftc.gov).
Regarding voice, consent becomes a standard contract clause. Those who work with "sound-alike" or "voice clone" must document chains of rights. Otherwise, reputational damage and, depending on the state, clear liability are threatened ( publications.tnsosfiles.com).
The USA is moving from "best practice" to "regulatory text" in AI labeling in advertising. New York is turning the question "Do I have to disclose AI in ads?" into a concrete obligation concerning synthetic performers ( governor.ny.gov). ). At the same time, the FTC is exerting pressure in the same direction through deception protection and endorsement transparency ( ftc.gov).
Companies that operate cleanly today are building an advantage: consistent disclosure language, clear asset documentation, and consent proofs for voice, image, and likeness. This ensures that AI creativity does not become a legal risk once a campaign scales or is rolled out to new states ( ftc.gov).