New York's AI Ad Disclosure Law Just Took Effect

For most of the generative AI boom, the rules about synthetic media in advertising lived in policy decks and voluntary frameworks. As of June 9, in one of the largest advertising markets in the country, they live in statute. New York's synthetic performer disclosure law is now in force, and it changes a creative team's obligations from a good-practice question into a compliance one.

What took effect on June 9

Governor Kathy Hochul signed S.8420-A into law in December 2025, amending the state's General Business Law. The operative date was June 9, 2026. The rule is short to state and broad in effect: any person who produces or creates an advertisement for commercial purposes must make a conspicuous disclosure when a synthetic performer appears in it, where that person has actual knowledge of the synthetic performer's use.

A "synthetic performer" is defined as a digitally created asset meant to create the impression of a human performer who is not recognizable as any identifiable natural person. Two details matter here. First, the definition is not limited to AI; content built with older software techniques is covered too. Second, the trigger is the appearance of a fake human, not the use of AI generally. Color grading, background cleanup, and copy written by a model do not pull an ad into scope. A digital person who looks real does.

Penalties are modest on paper and meaningful in aggregate. A first violation carries a $1,000 civil penalty, and each subsequent violation carries $5,000. For a brand running hundreds of creative variants, the per-violation structure is the part worth reading twice.

The word conspicuous is doing a lot of work

The statute requires the disclosure to be conspicuous and does not define the term. That gap is the whole ballgame. A disclosure that satisfies a static display banner may not satisfy a six-second vertical video where the synthetic performer is on screen for the entire spot. The law also exempts audio-only advertisements, AI used solely for language translation of a real performer, and expressive works such as films and games where the synthetic performance is consistent with its use in the work.

What the statute does not give you is a safe-harbor label. There is no approved badge, no standard placement, no minimum duration. Until the New York Attorney General's office or a court fills that in, every advertiser is making its own judgment about what conspicuous means across formats that did not exist when the bill was drafted. The Interactive Advertising Bureau's AI transparency framework, published in January, is the closest thing the industry has to a shared answer, and it is voluntary.

The reach is wider than the state line

The instinct of an out-of-state advertiser is to treat this as New York's problem. That instinct is wrong. The obligation attaches to advertisements directed at New York audiences, and most national campaigns are by definition directed at New York audiences. A wealth management firm in Toronto or Calgary running paid social into the United States is almost certainly serving impressions in New York. The law follows the ad, not the company's headquarters.

For programmatic and paid social campaigns, the targeting is rarely clean enough to carve New York out, and few advertisers would forfeit the market anyway. A state disclosure rule effectively sets a near-national floor for any campaign that uses synthetic humans.

This is one tile in a worsening patchwork

New York is not acting alone, and that is the deeper problem. California, Illinois, and several other states have moved on synthetic media, likeness, and AI disclosure with overlapping but non-identical rules. There is no federal preemption clearing the field. An advertiser running one creative concept across the country is now subject to a stack of state requirements that define their terms differently and trigger on different facts.

This is the regulated-industry compliance pattern arriving in mainstream advertising. Financial services, insurance, and healthcare marketers have lived with multi-jurisdiction disclosure rules for decades. What is new is that a beauty brand or a direct-to-consumer startup using a synthetic spokesperson now faces a version of the same problem, and the generative tools that made the synthetic spokesperson cheap did nothing to make the compliance cheap.

Disclosure has to live in the generation step

Here is the design lesson, and it is the one that matters for anyone building or buying marketing software. The expensive way to comply is to generate creative freely and then run a legal review pass that asks, for every asset, in every market, whether a disclosure is required and whether it is conspicuous enough. That review queue is exactly the bottleneck that kills velocity in regulated marketing.

The cheaper way is to make the system know, at the moment a synthetic human is generated, that a disclosure obligation has been created, and to carry that obligation through to the rendered asset for every targeted market. This is the thesis behind LeadLord, the AI marketing platform for wealth management and other regulated industries: compliance belongs inside the draft, not bolted on at the end. When the model that produces the creative is also the system that tracks which jurisdictions the asset will run in and what each requires, the disclosure stops being a manual review step and becomes a property of the asset. A team of three cannot staff a New York-versus-California disclosure review for hundreds of variants. Software can, but only if the rule is encoded where the asset is made, not in a checklist a human runs afterward.

What to watch before the first enforcement action

Three things will tell you how this plays out. The first is the New York Attorney General's enforcement posture: a single early action against a recognizable brand would do more to define "conspicuous" than any guidance document. The second is whether the IAB framework hardens into something courts treat as the reasonable baseline. The third is federal movement, because the more states pass their own versions, the louder the case for a single national rule becomes. Until then, the compliant move is to assume a synthetic human in any United States campaign needs a disclosure, and to build the workflow so that disclosure is automatic rather than remembered.