AI Ad Disclosure Labels in 2026 — Google, Meta, TikTok & State Law Rules

If you're running AI-generated or AI-edited video ads in 2026, disclosure is no longer optional on any major platform — and the rules differ enough between Google, Meta, and TikTok that following one platform's policy does not clear you on the others. This guide is the reference we wished existed when we started shipping AI UGC ads at volume: what each platform actually requires, who has to act versus who gets auto-detected, and where state law adds a second layer on top of platform policy.

None of this is about whether AI-generated ads are allowed — they are, everywhere covered here. It's about whether the ad has to say so, how visible that disclosure has to be, and what happens if you skip it.

The short version

• Google: added a "How this ad was made" panel to My Ad Center (rolled out broadly through 2026), accessible via the three-dot menu or info icon on Search, YouTube, and Discover ads. Disclosure is automatic when you use Google's own generative ad tools; for ads made with third-party AI tools, advertisers must self-flag using a control in Google Ads.
• Meta: labels ads created or edited with generative AI in the "About this ad" panel (three-dot menu). Meta auto-detects use of its own AI features (Background Generation, Image Generation, Add Animation) and applies the "AI Info" label automatically — no advertiser action needed. For third-party AI tools, Meta relies on detected content-provenance metadata, and as of June 1, 2026 layered in automated detection specifically for political/social-issue ads using third-party generative AI, which still requires a "conspicuous" self-disclosure from the advertiser on top of Meta's own detection.
• TikTok: the strictest of the three. Advertisers must manually enable an "AI Disclosure" tag in TikTok Ads Manager before submission if the ad contains AI-generated elements (synthetic faces, cloned voices, AI backgrounds, photorealistic AI product renders). This triggers a visible on-ad "Contains AI-generated content" label — not a hidden panel. Undisclosed AI content that's later detected faces ad rejection and account review; TikTok Shop sellers face an added requirement to label AI-rendered product images consistently across both ad creative and product listings.
• New York State: since June 9, 2026, any advertisement reaching New York audiences that includes a "synthetic performer" (an AI-generated digital asset intended to look like a human performing) must carry a conspicuous disclosure, regardless of platform. Civil penalties run $1,000 for a first violation and $5,000 for each subsequent one. About 30 U.S. states separately regulate AI disclosure specifically in political advertising.

Google: "How this ad was made"

Google's transparency panel sits inside My Ad Center and is reachable from any Search, YouTube, or Discover ad via the three-dot menu or info icon. The mechanics split cleanly in two:

• Google's own generative AI ad tools (Product Studio, Vids, image/asset generation inside Google Ads): disclosure triggers automatically. No advertiser step is required.
• Third-party generative AI tools: if you build a video or image ad using an outside AI platform — including Prizmad, Creatify, Arcads, HeyGen, or any AI UGC generator — and then run it through Google Ads, you're expected to use the disclosure control Google Ads now surfaces at the ad level to flag AI involvement. This is advertiser-declared, not auto-detected, which puts the compliance burden on you rather than the platform.

Google's ad policies already prohibited misleading or deceptive ads before this update; the disclosure panel adds a transparency layer on top rather than a new content restriction. Practically: nothing about an AI-generated UGC ad gets rejected for being AI-made on Google, but skipping the disclosure control when required is a policy violation in its own right.

Meta: the "AI Info" label

Meta's system, accessible through "About this ad" on any promoted Facebook or Instagram post, has two tiers depending on the source of the AI content:

Meta's own generative tools. Using Background Generation, Image Generation, or Add Animation to create or significantly edit an image or video triggers an automatic "AI Info" label. You cannot opt out of this, and you don't need to do anything to trigger it correctly — Meta's system handles it.

Third-party generative AI tools. This is the tier that matters for anyone using AI UGC or avatar platforms. Meta detects embedded content-provenance metadata (the same C2PA-style signals TikTok and Microsoft also read) from tools like Photoshop, DALL-E, and — where present — other generative platforms, and labels the ad accordingly when detected. The label appears next to "Sponsored" rather than buried, in regions where self-disclosure applies.

Political and social-issue ads get a stricter layer. Since June 1, 2026, Meta added automated detection specifically for social-issue, election, or political ads using third-party generative AI to depict a real person doing/saying something they didn't, a realistic person or event that doesn't exist, or altered footage of a real event. Advertisers running these categories must self-disclose regardless of whether Meta's detection catches it; failure to disclose gets the ad rejected outright, and repeat failures carry account-level penalties. Non-political commercial ads don't carry this self-disclosure obligation — but the auto-detected "AI Info" label still applies if Meta's systems flag the content.

The practical takeaway for DTC advertisers: run an AI UGC ad through Meta and, if the platform's detection reads AI-generation metadata in the file, it gets labeled — automatically, without extra steps on your end, and the label sits one tap away rather than on the video itself.

TikTok: the strictest and most visible

TikTok's policy is the one advertisers most often get wrong, because it requires action rather than relying purely on detection, and the resulting label is visible on the ad itself rather than tucked into a menu.

• The AI Disclosure tag is mandatory, not automatic. Before submitting a campaign in TikTok Ads Manager, advertisers must manually enable the AI Disclosure tag if the creative contains AI-generated content — synthetic faces, cloned or AI-generated voices, AI-generated backgrounds, or photorealistic AI-rendered products. Enabling it triggers a visible "Contains AI-generated content" label directly on the ad, not a hidden panel.
• Enforcement is retroactive and harsh. Ads submitted without the tag that TikTok later detects as AI-generated face immediate rejection and account review — a materially bigger consequence than a missing label on Google or Meta.
• TikTok Shop adds a consistency requirement. AI-generated product images must be labeled identically in both the ad creative and the product listing; a mismatch (labeled in one place, not the other) triggers listing removal, which cascades into rejecting every ad campaign tied to that listing. TikTok's crawler also checks that landing-page content matches ad-level disclosures.
• Spark Ads shift liability onto the brand. When a brand boosts a creator's organic video as a Spark Ad, the brand becomes co-responsible for that creator's AI disclosure compliance — if the original post has undisclosed AI elements, both parties face enforcement.

For AI UGC advertisers specifically: if your ad uses an AI avatar or AI voice, the honest move on TikTok is enabling the AI Disclosure tag every time, since the platform's stated posture is "when in doubt, disclose," and the downside of getting caught without it (rejection, account review) outweighs any perceived delivery cost of the visible label.

New York's synthetic-performer law — the state-level layer

Separate from platform policy, New York's Synthetic Performer Disclosure Law took effect June 9, 2026, and applies to any advertisement reaching New York audiences regardless of where the advertiser is located. It requires a "conspicuous" disclosure whenever an ad contains a "synthetic performer" — a digitally created asset intended to look like a human performance but not recognizable as any identifiable real performer. That definition covers AI UGC avatars squarely.

Key carve-outs: the law exempts advertisements for expressive works (films, streaming, video games) where the synthetic performer's use matches its use in the underlying work, audio-only ads, and cases where AI is used solely for translating a human performer's speech. Civil penalties are $1,000 for a first violation and $5,000 for each subsequent one. It's the first state law of its kind, and legal trackers expect similar bills in other states — treat it as a preview of where platform-level disclosure is heading, not an isolated rule.

A practical compliance checklist for AI UGC advertisers

1. Assume every AI avatar or AI voice ad needs disclosure somewhere. The safe default across all three platforms plus New York law is to disclose, not to guess whether your specific tool's output will get auto-detected.
2. On TikTok, always manually enable the AI Disclosure tag for avatar-based or voice-cloned creative before submission — don't rely on TikTok's detection to catch it for you, since the enforcement path for undetected-then-caught content is harsher than for tagged content.
3. On Meta, check "About this ad" after your first few campaigns to confirm the AI Info label is appearing as expected — this tells you whether your creative pipeline's metadata is being read correctly by Meta's detection.
4. On Google, use the AI-content disclosure control in Ads Manager when the creative came from a third-party tool; don't assume Google's own-tool auto-labeling extends to outside platforms.
5. If any of your spend touches New York, treat the conspicuous-disclosure requirement as a floor, not a nice-to-have — the penalty structure is per-violation, and "reaches New York audiences" is broad by design.
6. For political or social-issue ads on Meta, self-disclose regardless of platform detection — this is the one category where skipping self-disclosure gets the ad rejected outright, detection or not.

None of this changes whether AI-generated ads perform — disclosure requirements govern transparency, not creative quality or eligibility to run. The tools that generate the ad (Prizmad included) don't file the disclosure for you; that step lives in each platform's ads manager and is the advertiser's responsibility to enable correctly, every campaign.

---

Building AI UGC ads and want the disclosure step to be a checkbox, not a scramble? Prizmad generates the ad — script, avatar, voiceover, captions, music — from a product URL in about 5 minutes; enabling the right disclosure tag on each platform's ads manager takes seconds once you know which one applies. See the pricing or paste a URL into the generator to see the workflow.