New York’s AI advertising law: What it means for your brand or agency right now
TL;DR (For Brands & Agencies Running Paid Media)
New York is the first US state to require advertisers to disclose when an ad uses an AI-generated human (the law calls this a "synthetic performer"). The disclosure law takes effect June 9, 2026, and it applies to every paid advertising channel an ad runs on — Meta, Google, YouTube, TikTok, CTV, display, etc. — not just paid social. If an ad shown in New York state features an AI-generated person in an image or video, brands must conspicuously label it as AI, or face civil penalties of $1,000 for a first violation and $5,000 for each one after. AI-generated product-only or background/wide images that do not contain any humans in the visual are not required to carry the same distinction. The fix isn't complicated (you just have to add text within the creative asset to indicate it is AI-generated), but it does require brands to inventory their ad creative immediately and fix any issues, or get fined, that is, if you plan to target any advertising in the state of New York.
Let's break down what the law actually says, who it affects, and what brands should do about it.
What the law actually says
On December 11, 2025, New York Governor Hochul signed two pieces of legislation. The one most of us need to care about is S.8420-A / A.8887-B, which amends New York's general business law to require advertisers to disclose the use of "synthetic performers" in commercial advertisements. It's the first law of its kind in the United States, and New York is openly positioning it as a template other states (California is the obvious next one) will copy.
A "synthetic performer" is defined as a "digital asset created or modified by a computer using generative AI or a software algorithm that's meant to give the impression it's a real human in a visual or audiovisual performance." In plain terms, it's the AI-created person that isn't a real, living human, in your ad creative. The law covers both images and video.
The requirement is that the disclosure be "conspicuous." Notably, the statute does not dictate exact wording, size, or placement. That sounds like freedom, but law enforcers and regulators will most likely judge it against the standards we already use for paid influencer disclosures, which have to be prominent and basically unavoidable for a normal viewer. A tiny gray watermark purposely camouflaged in a corner is asking for trouble… and a fine.
Penalties are $1,000 for a first violation and $5,000 for each subsequent one. The disclosure rule does not apply to audio-only ads, to AI used solely for language translation, or to ads for "expressive works" like films, TV, and games where the synthetic performer also appears in the underlying work.
What counts - and what doesn’t
This is where the practical line matters, because not every AI-touched asset is in scope. The rule is about AI-generated humans. So:
If your ad shows an AI-generated person (a model holding your product, a synthetic spokesperson, a generated "customer" in a lifestyle shot) in an image or a video, it needs an AI disclosure. If an ad is a product-only shot, a background, or a wide environmental scene (AI generated or a real photo) with no AI-generated person in view, then it's outside the requirement. The important thing to note is that AI models are still completely usable in ad creative, brands just have to indicate they're AI.
The distinction that AI humans need labels, but AI backgrounds and products don't is the single most useful thing to teach your creative team this week.
Why this news is bigger than just Meta
Most of what we write at Foxwell is Meta-first, because that's where so much of the spend and the daily problem-solving lives. But this law is channel-agnostic. It applies to the advertisement "in any media in which that advertisement appears." That means the same AI-model creative running across Meta, Google, YouTube, TikTok, Axon, programmatic display, and connected TV needs the disclosure everywhere it shows up in New York, not just in your Ads Manager.
This means it's a cross-channel audit of all live assets, not a one-platform fix. The brands that handle this cleanly will be the ones with a single disclosure standard applied consistently across every placement, rather than a scramble per channel.
The questions Founders members are asking (and honest answers)
The minute this hit the Foxwell Founders community, the real questions started flying. Here's where we've landed.
"Is this just New York adding friction?" Maybe. But it's now a state law, not a city ordinance, and the part that trips people up is this: it reaches any ad shown to people in New York, regardless of where your brand or agency sits. You don't get to opt out by being based in Austin or London. If New Yorkers can see the ad (or if you're targeting the entire US and someone physically in the state of New York could be targeted with an ad), you're in scope.
"Who actually enforces this — is Meta the AI police now?" No. The obligation sits on whoever produces or creates the advertisement (the brand and its agency) not the platform. Enforcement runs through civil penalties at the state level, and the statute doesn't hand consumers a private right of action to sue you directly. So don't wait for Meta to build a magic disclosure toggle or to tell you what to do. Meta is generating enormous volumes of AI imagery that will absolutely surface in New York feeds, but the compliance buck stops with the advertiser, not the ad network.
Meta is already preparing for AI transparency
New York's law puts the onus on advertisers to disclose AI use, and to help with this, Meta has rolled out new AI transparency tools in Ads Manager. Advertisers can now indicate if their ad has "media created or edited with AI," adding an AI label to their content.
Meta is making strides to improve transparency for AI-generated content. However, it's still uncertain how their labels will align with New York's legal requirements, especially regarding how noticeable they are and if they’ll stay consistent across different edits and formats. As with many of Meta’s new features, we’ll have to wait and see how it all plays out in real life.
"It's a grey area, right?" On the format of the disclosure, yes — "conspicuous" isn't defined. Whether you need one at all for an AI human is not gray: you do. Hands and other appendages are NOT included in the NY AI-generated humans law for indicating they are AI
"Should we just stop using AI entirely?" That's the overcorrection we'd push back on. The law doesn't ban AI creative altogether. It asks brands to be honest about it with their viewers and customers. Killing your whole AI creative pipeline over a labeling requirement is throwing out a productivity engine to avoid adding four words to a corner of the frame.
One practical read making the rounds: counsel for a large haircare brand told their team that AI-generated hands (and similar non-identifiable body parts holding or using a product) are fine without a performer disclosure — because a hand isn't a "performer" appearing as a person. That tracks with the statute, but treat it as informed interpretation, not settled law, and get your own counsel's sign-off before you bank on it.
What to do before June 9, 2026
1. Inventory all live ad creative on all digital advertising platforms. Pull every active and queued ad and flag the ones with an AI-generated human in an image or video. Turn them off while you decide next steps.
2. Decide: relabel or replace. For each flagged asset, choose to add a clear AI disclosure or swap it for real talent or UGC. This is also a good moment to ask whether the AI model was earning its keep in the first place, or if it's time to let that one die out.
3. Standardize one disclosure treatment. Build a single, legible label — something like "AI-generated" in a consistent spot (a bottom-corner overlay is common) that stays readable across every placement and aspect ratio. Don't let it become per-designer guesswork.
4. Lock down who's responsible. Update your agency, production, and vendor agreements to spell out who applies the disclosure, and confirm whether any partner is quietly using synthetic performers in assets they hand you. Get indemnities where it makes sense.
5. Update brand guidelines and your review process. Add an AI-human checkpoint to your pre-launch QA so nothing ships into New York without being cleared. Make sure the whole team knows the AI-human-vs-product distinction.
Our honest take
This law lands almost exactly where we already stand at Foxwell Digital: be transparent about AI, and tell people when you're using it. In a feed and a world where we're surrounded by AI, being upfront is quickly becoming a trust advantage. The brands that treat "this is AI" as a normal part of the asset, not a confession, are going to look a lot better than the ones who got caught hiding it.
The hard part isn't the label, it's keeping up with the rules, with the platforms, and with what's actually working in accounts week to week. That's the whole reason the Foxwell Founders membership exists. If you've ever thought "I need to find a community of the best paid social operators and marketers in the world" or "where can I get real-time, trusted advice on what's working in Meta or Google Ads right now?" — this is it. It's where agency teams and freelancers go to learn how to scale brands profitably without guessing, where creative strategists trade notes on UGC sourcing and iterating Meta ad creatives like a pro, and where you can genuinely learn how to operate your Meta ads agency more effectively and help grow the brands you work with. When a law like this drops at 11am, our members have a vetted answer by lunch.
New York is first, not last. Build the disclosure habit now and the next ten states won't faze you.
UPDATE (07-14-2026): It's already spreading, and it's not staying in New York's lane
When we first published this, we called California the "obvious next state." Turns out we undersold it. There are now roughly 16 other states with AI-disclosure bills in some stage of movement, and several of them are written far more broadly than New York's synthetic-performer rule. If you only build your compliance process around "label the AI humans," you're going to be caught flat-footed elsewhere. Three worth knowing right now:
Georgia HB 478 doesn't limit itself to human figures at all. It targets AI-generated content used in commerce and trade broadly, which means the "product shots and backgrounds are safe" carve-out that works in New York may not hold up here.
Massachusetts H.81 goes even further on scope: it covers any content "substantially created or modified" by generative AI, and it's written to include all content types. Not just images and video, but blogs, articles, and other text. If this passes as written, your AI-assisted written content is potentially in scope too, not just your ad creative.
California's SB 942 (the AI Transparency Act) is a different animal. It puts the disclosure obligation on platforms rather than brands, requiring them to label AI-generated content at the point of distribution. That makes it less of a direct compliance headache for advertisers and more of a coming headache for Meta, Google, TikTok, and the rest.
The practical read: don't build a one-state playbook. A disclosure standard designed only around "no AI humans in view = no label needed" won't survive contact with Georgia or Massachusetts. Treat New York as the floor, not the ceiling, and expect your inventory-and-label process to expand to cover text content and non-human AI assets sooner rather than later. We'll keep tracking this as more of these bills move.
Note: This blog was written by a human but aided by AI to draft the outline and synthesize ideas.
