Short answer. Yes. None of the three platforms prohibits AI-generated ad creative, and all three have built disclosure machinery for it instead. What they regulate is disclosure, and they regulate it differently: TikTok requires a disclaimer on any ad whose image, video or audio is completely AI-generated or significantly modified by AI, while Meta and Google reserve mandatory advertiser self-disclosure for political, election and social-issue ads. The practical risk in 2026 is not a ban. It is skipping a toggle and letting the platform add the disclosure for you.
Key takeaways
- None of Meta, Google or TikTok bans AI-generated ad creative. All three ship their own generative tools inside their ad products, and the live question is disclosure, which each answers differently.
- TikTok has the broadest ad-side rule: a mandatory disclaimer on image, video or audio that is completely AI-generated or significantly modified by AI, with no sensitivity gate. Meta and Google gate mandatory self-disclosure to political, election and social-issue ads.
- Platform-side disclosure arrived in 2026: Google's "How this ad was made" panel on 9 July 2026, and Meta's statement, added on 1 June 2026, that it adds information to ads it detects were created with third-party AI tools.
- Content Credentials are not the ad-side mechanism people assume. Meta describes using C2PA to identify AI content outside of ads; Google embeds SynthID in its own generative outputs. No platform documents reading Content Credentials on ad upload.
- Realistic synthetic humans, altered real people, political and social-issue content, and depicted real events are the four creative choices that trigger the strict rules. Stylised and illustrated creative mostly sits outside them.
- Advertiser attitudes are badly calibrated to audience attitudes: 82% of ad executives think young consumers feel positive about AI-generated ads, against 45% who actually do.
Quick comparison
| Dimension | TikTok | Meta | Google / YouTube |
|---|---|---|---|
| AI ad creative prohibited? | No | No | No |
| Mandatory disclosure scope | Any ad with image, video or audio completely AI-generated or significantly modified by AI | Ads about social issues, elections or politics, in certain cases | Election ads with synthetic content inauthentically depicting real or realistic-looking people or events |
| Who applies it | Advertiser toggle in Ads Manager | Advertiser self-disclosure; Meta also adds information to ads it detects used third-party AI tools | Automatic for Google's own generative tools; an advertiser control for creative made elsewhere |
| How the viewer sees it | Textual disclosure at the bottom of the in-feed video for the ad's duration | Information added on the ad and in the Ad Library when an advertiser discloses | "How this ad was made" in My Ad Center; an on-ad label where local rules require |
| Provenance signals | Not documented for ads | C2PA used to identify AI content outside of ads | SynthID embedded in Google's own generative outputs |
| Political ads | Paid political advertising not allowed at all | Self-disclosure required, in certain cases | Disclosure required for verified election advertisers |
| Deception baseline | Prohibited regardless of AI | Prohibited regardless of AI | Prohibited regardless of AI |
Is AI-generated ad creative actually allowed?
Yes, and it is increasingly assumed. Meta, Google and TikTok all run generative creative tools inside their own ad products, and none of the three publishes a prohibition on AI-made ad creative. What arrived in 2025 and 2026 was a transparency layer, not a ban.
One honest caveat on how that answer is built. The platforms do not publish a sentence saying "AI-generated ads are permitted." They publish disclosure obligations, label mechanics and category restrictions — which is a stronger signal than a permission statement would be, because a platform does not build creative generation into its buying flow and a disclosure toggle into its ad setup while banning the output.
The policy foundation underneath is older than the AI question and has not moved. Announcing its ad transparency labels, Google restated that it continues to prohibit misleading and deceptive ads, whether created with AI or not. That is the baseline on every platform: a synthetic ad that misleads is a misleading ad, sanctionable under rules that predate the technology.
So the buyer question has quietly changed shape. "Can we run AI ads?" has a one-word answer. The real questions are which of your creative needs a disclosure, who applies it, which toggle lives where, and which narrow categories carry genuinely strict rules. Getting those right is administrative. Getting them wrong invites the one outcome worse than a label — the platform deciding your ad was hiding something.
What does each platform's rulebook require?
The three platforms run three different rulebooks. TikTok discloses by content type and has the broadest ad-side scope, Meta gates its mandatory rule to sensitive ad categories, and Google centralises disclosure in a transparency panel that is automatic for its own tools and manual for everyone else's.
Key terms
- AIGC is the industry shorthand for AI-generated content, used by TikTok in its labelling.
- C2PA Content Credentials are tamper-evident provenance metadata attached to a media file recording how it was made.
- SynthID is Google's imperceptible watermarking of outputs from its own generative models.
- Deception baseline is the long-standing rule on every platform that misleading advertising is prohibited regardless of whether AI made it.
- "How this ad was made" is Google's transparency section in My Ad Center that discloses when an ad's creative was made or altered with generative AI.
TikTok — disclosure by content type, broadest ad-side scope. TikTok Ads Manager lists AI-generated, synthetic or manipulated media among its mandatory disclaimers, covering images, video or audio that are completely AI-generated or real source material significantly modified by AI. There is no sensitivity gate: an AI presenter or a synthetic voice reading your script needs the disclaimer however commercial and harmless the message. The mechanics matter as much as the rule. The advertiser ticks a toggle reading This ad contains AI-generated content, the disclosure then shows at the bottom of the in-feed video for the duration of the ad, and the toggle cannot be unselected unless the media changes. Two traps follow: any new campaign creation, including duplication, resets the toggle even on a reused ad, and Spark ads are excluded from it entirely because they live organically and follow TikTok's Community Guidelines disclosure route instead. Note also that TikTok does not allow paid political advertising at all, so the election-disclosure question does not arise there the way it does on Meta and Google.
Meta — self-disclosure gated to sensitive categories, plus detection. Meta's mandatory rule is narrow and specific: advertisers are required to disclose when they use AI to create or alter ads about social issues, elections, or politics, in certain cases, and when an advertiser discloses, Meta adds information on the ad and in its Ad Library. Outside that category, Meta's documented position is about detection rather than obligation: in an update dated 1 June 2026 it says it is adding more information to ads it detects were created with third-party AI tools. The "AI info" label most marketers have seen is documented for content beyond political ads — organic content — rather than as an ad-side requirement, and Meta's use of C2PA is explicitly described as identifying AI content outside of ads on its platforms.
Google — centralised panel, automatic for its own tools. From 9 July 2026, Google's "How this ad was made" section in My Ad Center discloses when an ad's creative involved generative AI. Creative made with Google's own generative advertising tools gets that disclosure automatically with no advertiser step; for creative made elsewhere, the advertiser uses a control to indicate generative AI was involved. Based on local requirements a label may also appear directly on the ad. Google embeds imperceptible signals such as SynthID into outputs from its own generative tools — note that this is Google's own watermarking, not a claim that Google reads third-party Content Credentials on ads, and Google does not state that it independently verifies an advertiser's disclosure. The strict, enforced rule is the election one: verified election advertisers in regions where verification is required must prominently disclose synthetic content that inauthentically depicts real or realistic-looking people or events, across image, video and audio, with inconsequential alterations such as resizing, cropping and colour correction exempt.
Which ads trigger the strict rules?
Four creative choices do most of the triggering: realistic synthetic humans, altered real people, political or social-issue content, and depictions of real events. Regulated verticals add a fifth layer on top. Knowing them lets you design around review friction deliberately rather than discover it at upload.
Realistic synthetic humans. An AI presenter, a cloned or synthetic voice, a photoreal generated spokesperson. On TikTok this needs the ad disclaimer on its face, because the rule is about what the media is rather than what the ad is about. Stylised animation, motion graphics and obviously illustrated characters sit outside the realistic trigger, which is part of why so much performance creative has drifted that way.
Real people, altered. Depicting an identifiable real person doing or saying something they did not is the red line everywhere. In sensitive categories it is the definition of the mandatory-disclosure rule; outside them it collides with the deception baseline and with likeness rights. This includes your own executives: a cloned CEO voice needs documented consent as well as a disclosure, a chain-of-rights question we cover in Who Owns AI-Generated Video, and Whose Consent Do You Need?
Political, election and social-issue content. The category where Meta's and Google's self-disclosure rules carry enforcement weight, and where TikTok's answer is simply that paid political advertising is not allowed. If your brand runs issue-adjacent campaigns, treat any synthetic element as disclosable by default.
Depictions of real events. Generated "footage" of a news event, disaster or public moment trips the realistic-events language in the rulebooks, commercial framing notwithstanding.
And one trigger that is about product, not policy: regulated verticals. Finance, health and legal advertisers face platform ad-category rules and sector regulators on top of the AI layer, and the compliance-first production logic we set out in AIGC in Regulated Industries applies to the 30-second ad even more than to the long-form video.
What happens if you skip the disclosure?
It depends entirely on which rule you skipped. Outside the sensitive categories the likely outcome is that the platform adds the disclosure you did not, and you lose control of the wording and timing. Inside them you are in direct breach of an advertising policy.
At the mild end, nothing visible changes except authorship of the disclosure. Meta's statement that it adds information to ads it detects were created with third-party AI tools is exactly this mechanism: the transparency happens either way, and your only loss is that it reads as caught rather than offered.
In the middle sits review friction. TikTok's disclaimer feature carries ad-review consequences for incorrect use, and the toggle's reset-on-duplication behaviour is a reliable way to ship an undisclosed variant of a campaign you disclosed properly the first time. The operational cost is a paused campaign and a resubmission, not a penalty — but it lands mid-flight, which is when it is most expensive.
At the serious end are the sensitive categories, where undisclosed synthetic content in an election or social-issue ad is a policy violation rather than a transparency gap, and the deception baseline, where an ad that misleads is sanctionable whether AI made it or not.
There is also a cost no policy page lists: being caught undisclosed is a trust event with your audience, and the audience research says the ground is already soft. IAB's study of 505 US Gen Z and Millennial consumers alongside 104 industry executives found 82% of executives believe those consumers feel positive about AI-generated ads, while only 45% of the consumers do — a 37-point perception gap, with Gen Z nearly twice as likely as Millennials to feel negatively. A disclosure you applied reads as transparency. One the platform applied reads as a catch.
How do the laws behind the platforms change the picture?
The platform toggles are the surface of a legal layer converging on one requirement: AI-generated content must be disclosed. Google's own rollout timing makes the point — the ad transparency labels landed on 9 July 2026, weeks before the EU's transparency obligations began to bite.
Article 50 of the EU AI Act has required machine-readable marking and deepfake disclosure for specified AI-generated content since 2 August 2026. China's labelling rules for synthetic content and the patchwork of US state deepfake and election statutes complete the triangle; the jurisdiction-by-jurisdiction map is in our guide to AI content labelling law in the EU, China and the US. This is why Google's treatment varies by market: the panel is global, but whether a label also appears on the ad face depends on local requirements.
The connective tissue is provenance metadata, and here the popular summary is wrong in a way worth correcting. It is widely repeated that all three platforms read C2PA Content Credentials on ad upload. The platforms' own pages do not say that. Meta describes C2PA as helping identify AI content outside of ads; Google describes embedding SynthID in its own generative outputs; TikTok's ad-side disclosure runs through the advertiser's toggle. Provenance metadata is therefore worth preserving because it is where the ecosystem is heading and because it governs the organic surfaces today — not because it will quietly handle your ad compliance. How credentials and watermarks survive, and fail, through editing pipelines is its own subject, covered in Content Provenance: C2PA, SynthID and What Survives.
For planning purposes the direction of travel matters more than any single rule: disclosure scope has only widened, platform by platform and year by year. Building campaigns that assume disclosure, and testing whether it costs you performance rather than assuming it does, is the posture that will not need rework in 2027.
How does Lifewood produce ad creative that clears review?
Lifewood runs ad and campaign creative through the same human-in-the-loop AIGC pipeline as long-form work, with a disclosure pass built into delivery. The disclosure decision is made once, in production, rather than rediscovered ad by ad at upload.
Lifewood's AIGC video production ships every asset with a record of what was generated versus filmed, which disclosures it will trigger under each of the three rulebooks above, and the per-platform toggle checklist for the team doing the trafficking. Where campaigns localise across markets the review layer carries the jurisdictional differences too, since an asset that needs no visible label in one region may need one in another. The multi-variant mechanics — one approved hero film becoming cut-downs, stills and local-language versions — are covered in our guide to AI video campaigns for brands, and buyers comparing providers can start with the best AI video production companies for advertising and commercials. The broader managed offer sits under AIGC services.
The toolchain matters as much as the producer. Platform-native generation carries the advantage of automatic, correctly-scoped disclosure inside its own ecosystem, which is the single easiest compliance win available. Everything made outside a platform's tools is disclosed by you, by hand, on every platform — so the asset register that records which is which is not administrative overhead, it is the control that makes the toggles reliable.
What should you do before your next AI-made campaign goes live?
Classify every asset, apply the realistic test, switch the disclosure toggles yourself, and screen for the red lines before generation rather than after. Then re-check the platform policies each campaign cycle.
- Classify every asset once: fully generated, significantly AI-edited, AI-assisted (copy help, cropping, colour), or conventional. The classification drives every downstream decision.
- Apply the realistic test. Photoreal humans, voices or events anywhere in the cut means plan for disclosure. On TikTok, a completely AI-generated or significantly AI-modified image, video or audio needs the disclaimer regardless of how realistic it looks.
- Switch the toggles yourself: TikTok's This ad contains AI-generated content, Meta's disclosure for social-issue, election and political ads, and Google's control for creative made outside its own tools.
- Re-check the toggle on every duplicated campaign. TikTok resets it on new campaign creation, including duplication, even for an ad you disclosed before. This is the most common way a compliant campaign ships an undisclosed variant.
- Route Spark ads separately. They are excluded from the Ads Manager toggle and follow the Community Guidelines disclosure route, so a paid push behind an unlabelled organic post inherits the original's problem.
- Screen for the red lines before generation, not after: no real people altered without documented consent, nothing issue-adjacent without a disclosure plan, nothing that fails the plain deception test.
- A/B the disclosure. Run disclosed creative against your conventional baseline and measure. Given a 37-point gap between what executives assume about audience sentiment and what audiences report, assumption is the expensive option in both directions.
- Keep one owner for the platform-policy watch. Quarterly change is now the norm, and every page cited here carries a revision date worth checking.