AI-Generated Ads Must Be Labeled Now: New York Disclosure Law (2026)
New York just became the first US state to require advertisers to disclose when they use AI-generated people in ads. The AI Transparency in Advertising Act (S.8420-A) took effect June 9, 2026, and it applies to any paid ad shown in New York featuring a "synthetic performer" - a digitally created humanlike figure designed to look like a real person. Penalties start at $1,000 per violation and jump to $5,000 for repeat offenses. If you've ever wondered whether the person smiling at you from an Instagram ad actually exists, New York now says you have a right to know.
What New York's AI Ad Law Actually Says
Governor Hochul signed S.8420-A into law on December 11, 2025, giving advertisers roughly six months to prepare before the June 9, 2026 effective date. The law is short and direct: if you run a paid advertisement featuring a synthetic performer, you must include a "conspicuous disclosure" that the person in the ad was generated by AI.
The law covers all paid advertising channels. That includes Meta and Instagram ads, Google display and YouTube pre-rolls, TikTok sponsored content, connected TV spots, programmatic display networks, and any other format where someone pays to put content in front of an audience in New York.
One quirk worth noting: the law requires "conspicuous disclosure" but doesn't define what that means or specify a format. There's no mandated label size, placement, or wording. Advertisers are left to figure out what "conspicuous" means in practice - which almost certainly guarantees inconsistency until enforcement actions or guidance establish a baseline.
What Counts as a Synthetic Performer
The law defines a "synthetic performer" as a digitally created humanlike figure made using AI to appear as a person performing in an advertisement. This is narrower than you might expect. It targets AI-generated people - not every use of AI in advertising.
A fully AI-generated person speaking in a video ad? That's a synthetic performer. An AI-generated model posing in a product photo? Also covered. But a real model whose photo was touched up with AI skin-smoothing filters? That's not a synthetic performer under this law. The distinction is between creating a fake person and editing a real one.
The law also carves out two specific exemptions. Audio-only ads don't count - so an AI-generated voice reading ad copy isn't covered. And language translations are exempt, meaning AI-dubbed versions of ads featuring real people don't trigger the disclosure requirement.
Who Has to Comply (And Who Doesn't)
The compliance burden falls on advertisers - the companies paying for the ads. If you're a brand running Instagram ads with AI-generated spokespeople, you need the disclosure. If you're a creative agency producing those ads, you're likely responsible too depending on your contractual setup with the brand.
Media platforms that merely host or distribute ads are explicitly exempt. So Meta, Google, TikTok, and YouTube don't have to police whether advertisers include the required disclosures. That's a significant carve-out - it means enforcement depends entirely on the state going after individual advertisers rather than requiring platforms to gatekeep compliance.
Individuals posting organic social media content aren't covered either. The law targets paid advertising specifically. If someone posts an AI-generated image to their personal Instagram feed without paying to promote it, this law doesn't apply.
Enforcement and Penalties
The New York Attorney General has exclusive enforcement authority. There's no private right of action, which means individual consumers can't sue advertisers directly for violations. If you spot an ad with an obvious AI person and no disclosure, your recourse is to report it to the AG's office and hope they pursue it.
Penalties are modest by advertising industry standards: $1,000 for a first violation and $5,000 for each subsequent offense. For a major brand spending millions on ad campaigns, these fines alone aren't much of a deterrent. The real pressure will come from reputational risk - nobody wants to be the first brand publicly called out by the Attorney General for using fake people in ads without telling consumers.
Is New York's AI advertising disclosure law effective? The law creates a legal obligation for advertisers to disclose AI-generated synthetic performers in paid ads, with the New York Attorney General as the sole enforcer. While the $1,000 to $5,000 penalty range is relatively small, the law establishes a precedent that other states are likely to follow with stricter requirements.
What This Means for You as a Consumer
The immediate practical impact is small. Most major advertisers using AI-generated people will add disclosures to avoid being the test case for a new law. But the disclosures will probably be easy to miss - a small text label, a brief flash at the end of a video, fine print in a carousel. Without specific format requirements, advertisers have every incentive to make disclosures technically compliant but functionally invisible.
The bigger shift is cultural. This law establishes the principle that you have a right to know when the person in an ad isn't real. That's new. Until June 9, 2026, no US law required this. Advertisers could use entirely AI-generated people - perfect skin, perfect lighting, expressions optimized by A/B testing - and never mention it.
This matters more than it might seem. AI-generated ad models don't have bad hair days, don't age, don't demand residuals, and can be iterated in hours instead of weeks. They're cheaper and more controllable than real people. Without disclosure requirements, the economic incentive is overwhelming: replace human talent with synthetic performers wherever possible. The law doesn't stop that - it just says you have to be honest about it.
How to Spot AI-Generated People in Ads
Even with disclosure requirements, many ads (especially from non-compliant or out-of-state advertisers) won't carry labels. Here's what to look for when something feels off:
- Hands and fingers: AI still struggles with hands. Look for extra fingers, fused digits, or unnaturally smooth skin texture on palms
- Hair boundaries: the line where hair meets skin or background is often too clean or has subtle artifacts where strands merge into the background
- Teeth uniformity: AI-generated smiles tend to produce teeth that are too uniform in size, spacing, and color - real teeth have slight variations
- Jewelry and accessories: earrings, necklaces, and glasses frequently have asymmetric rendering or physically impossible geometries
- Background consistency: check where the person meets the background - warping, repeating patterns, or blurred transitions are common tells
- Too perfect: if someone in an ad looks flawless in a way that feels uncanny, trust that instinct - real humans have pores, asymmetry, and imperfections
These visual tells are getting harder to spot as the models improve. The long-term answer isn't training your eye - it's provenance technology like C2PA watermarking that embeds origin data directly into image files. But right now, a critical eye is your best tool.
The Bigger Picture: More States Will Follow
New York is first, but it won't be alone for long. The EU AI Act's transparency requirements for AI-generated content take effect in August 2026 - just two months after New York's law. California has already passed AI transparency legislation targeting training data disclosure. Multiple states have active AI regulation bills moving through their legislatures.
The C2PA (Coalition for Content Provenance and Authenticity) now has over 6,000 member organizations working on technical standards for content authenticity. Companies like Adobe, Google, Microsoft, and Sony are embedding provenance data into their creation tools. The infrastructure for tracking whether content is real or synthetic is being built right now.
The pattern is familiar: one state passes a law, others follow with their own versions, and eventually the patchwork either forces federal action or becomes the de facto standard. California's privacy law (CCPA) followed exactly this trajectory. AI transparency in advertising is on the same path.
What to do regardless of where the law stands
Laws change, get amended, and vary by state. But the underlying question - "is the content I'm seeing real?" - isn't going away. A few things hold true regardless of which specific regulations apply in your state:
- Default to skepticism with polished ad imagery. If a person in an ad looks impossibly perfect, consider whether they might be entirely AI-generated
- Check for disclosures, but don't assume their absence means the content is real - compliance will be uneven, especially across state lines and international advertisers
- For photos you share yourself, use platforms that preserve original metadata and support authenticity verification - provenance data only works if it survives the sharing process
- Keep personal photos private. The less publicly available imagery of you exists, the harder it is for anyone to create convincing synthetic versions of you for any purpose, including advertising
This is part of why we built Viallo the way we did. When you share photos through Viallo, they're stored at full resolution with all original metadata intact. Albums aren't publicly indexed and can be password-protected. It's a different model from social platforms where your photos become public training data. In a world where AI can generate convincing fake people for ads, keeping your real photos private and verifiable matters more than ever.
Frequently Asked Questions
What is the best way to tell if a person in an ad is AI-generated?
Look for visual artifacts: unnatural hands, too-perfect teeth, asymmetric jewelry, and hair that blends oddly into the background. In New York, advertisers are now legally required to disclose synthetic performers, so check for small-print labels or end-of-video text. Viallo preserves all embedded metadata in shared photos, so if an image contains C2PA provenance data proving its origin, that information survives when shared through the platform. Google Photos and Instagram both strip this metadata during upload.
How do I report an ad using AI-generated people without disclosure?
In New York, file a complaint with the state Attorney General's office through their consumer complaint portal. The AG has exclusive enforcement authority under the AI Transparency in Advertising Act. There's no private right of action, so you can't sue the advertiser directly. For ads on Meta or Google platforms, you can also flag the ad through each platform's ad reporting tools, though platforms themselves aren't legally required to enforce compliance.
Is it legal to use AI-generated people in advertising?
Yes, it's legal. New York's law doesn't ban synthetic performers in ads - it only requires disclosure. Advertisers can use fully AI-generated people in any paid advertising channel as long as they include a conspicuous notice. Other states don't yet have similar requirements, though the EU AI Act will mandate comparable transparency starting August 2026. The penalties in New York are $1,000 for a first offense and $5,000 for subsequent violations.
What is the difference between AI-generated ads and deepfakes?
AI-generated ads use synthetic performers - entirely fabricated people who don't exist - in paid advertising. Deepfakes use AI to manipulate imagery of real, identifiable people, often without their consent. New York's ad disclosure law covers synthetic performers specifically, while deepfakes fall under separate legislation like the federal TAKE IT DOWN Act. Viallo's private sharing model helps protect against both: photos shared through Viallo aren't publicly crawlable, making them harder for AI tools to use as source material for either synthetic performers or deepfakes. TikTok and Instagram, by contrast, make posted photos readily accessible for scraping.
Will other states follow New York's AI ad disclosure law?
Almost certainly. The pattern mirrors what happened with data privacy after California passed the CCPA - multiple states drafted their own versions within 18 months. Several states already have active AI regulation bills, and the EU AI Act's August 2026 deadline will create additional pressure for US alignment. Viallo's approach of preserving original photo metadata and keeping content private positions it well for whatever transparency standards emerge, since provenance data embedded by cameras survives the sharing process intact.