Do I have to disclose AI-generated creative in Facebook and Instagram ads?

For ordinary commercial ads, Meta does not require you to self-disclose. Meta labels the ad itself: it automatically applies an 'AI info' label when an ad is created or edited with Meta's own generative AI tools, and since the June 2026 product update it also automatically detects ads created or edited with third-party generative AI tools using industry-standard signals such as C2PA metadata. Self-disclosure is mandatory only for ads about social issues, elections or politics.

Which AI edits trigger a label in Meta Ads Manager?

For Meta's in-house tools, Meta says a label is applied when features such as Background Generation, Image Generation or Add Animation significantly edit an image or video. Meta states that if its in-house tools do not result in significant edits and do not include a photorealistic human, no AI label is applied. For third-party tools, labelling depends on Meta detecting industry-standard provenance metadata in the file rather than on a written list of edit types.

What are the rules for AI in political ads on Meta?

Advertisers must disclose when an ad about social issues, elections or politics contains a photorealistic image, video or realistic-sounding audio created or edited with AI to depict a real person saying or doing something they did not, to depict a realistic-looking person who does not exist or an event that did not happen, or to depict a realistic event that is not a true recording. Failure to disclose means the ad is rejected or removed, and repeated failure may result in account penalties. Edits that are inconsequential or immaterial to the claim the ad makes do not need disclosure, and Meta names image size adjusting, cropping, colour correction and image sharpening as examples.

Do I need to disclose AI in an organic Instagram or Facebook post I plan to boost?

Yes. Meta's Misinformation Community Standard requires people to disclose, using Meta's AI disclosure tool, whenever they post organic content with photorealistic video or realistic-sounding audio that was digitally created or altered, and Meta may apply penalties for failing to do so. Meta states that AI-generated organic content from brands follows the organic labelling approach, and that organic labels may remain present when the post is boosted into an ad.

Does the AI info label hurt ad performance on Meta?

Meta has not published any performance data on the AI info label. Its public material frames the label as a transparency measure, not a ranking signal, and the label is usually not visible on the ad surface at all since it sits behind the three-dot menu. Anyone quoting a specific reach or CTR penalty from the label is not citing Meta.

Meta's AI-Generated Creative Ad Policy (2026)

Quick answer: if you run ordinary ecommerce ads on Facebook and Instagram, Meta does notask you to tick an AI box. Meta labels the ad for you, applying an "AI info" label behind the three-dot menu when it detects generative AI, including third-party tools, as of the June 2026 update. Mandatory self-disclosure exists in exactly one place: ads about social issues, elections or politics. The real obligation for everyone else sits in the organic rules and in the general ban on deceptive claims, which AI creative makes easier to break.

Meta's Advertising Standards barely mention AI, and that is the point

Read Meta's Introduction to the Advertising Standards end to end (roughly 30,000 characters covering everything from discriminatory practices to branded content) and you will not find a general AI disclosure clause. Search it for "artificial intelligence", "AI-generated", "synthetic" or "manipulated media" and you get nothing.

Meta has routed AI obligations through three other places instead: the Misinformation Community Standard (which ads must comply with), the separate standard for ads about social issues, elections or politics, and product behaviour, meaning automatic labelling rather than an advertiser duty. Working out which of the three applies to a given asset is the whole job.

Which rule applies to your asset?

Asset	Must you self-disclose?	What Meta does	Penalty for skipping
Commercial ad, Meta's own AI tools	No	Auto "AI info" label when the edit is significant	Not applicable
Commercial ad, third-party AI tools	No (optional in some regions)	Auto-detects via industry-standard signals, then labels	Not applicable
Organic post with photorealistic video or realistic audio	Yes, via Meta's AI disclosure tool	May also label; may add context labels on high-risk content	Meta "may apply penalties"
Boosted organic post	Yes, as organic	Organic labels may carry into the ad	As organic
Ad about social issues, elections or politics	Yes, mandatory toggle in Ads Manager	Labels near the "Paid for by" disclaimer, notes it in Ad Library	Ad rejected or removed; repeated failure risks account penalties

The row most ecommerce advertisers misread is the second one. "No self-disclosure required" is not the same as "nobody will know". Meta detects and labels it anyway.

How the automatic labelling actually works

Meta's newsroom post on GenAI ads transparency, originally published February 2025 and updated 1 June 2026, describes the current state. "About this ad" is now a unified destination in the three-dot menu on every ad, sitting alongside "Why am I seeing this ad?". The AI info label lives inside it, together with payer and beneficiary details and "Paid for by" disclosures.

For Meta's in-house creative features, the logic is spelled out. If tools such as Background Generation, Image Generation or Add Animation do not significantly edit the asset and do not include a photorealistic human, Meta applies no label. If they do significantly edit it, a label appears behind the three-dot menu. If they produce an AI-generated photorealistic human, Meta says the label appears next to the Sponsored label rather than being tucked away.

For third-party tools, and this covers every AI UGC platform, Meta says it uses industry-standard detection methods such as C2PA to identify when ad content has been created or edited with them, and labels accordingly when it finds that metadata. Two practical consequences follow. First, no advertiser action is required. Second, detection is only as good as the metadata, and re-encoding or editing a clip routinely strips provenance data, so labelling across a set of near-identical variants can be inconsistent through no fault of yours.

One small but visible change worth knowing: Meta has updated ad labelling so that ads previously marked "Sponsored" now display an "Ad" label. Confusingly, the same help article still describes some AI labels as appearing "next to the Sponsored label", and adds that the label name may differ by region. Read "Sponsored label" in Meta's documentation as "whatever the ad marker is called in that market".

The regions where you can, and probably should, self-disclose

Meta notes that where local laws or guidelines carry specific AI transparency requirements, advertisers may be given the option to self-disclose that generative AI was used, and that if they do, the label appears on the ad surface next to the Sponsored label instead of being hidden in the menu. Meta lists these regions as the European region, California, New York, India and Taiwan. It separately notes that in Taiwan, AI-generated photorealistic humans in creative made with Meta's own tools may be auto-labelled.

If you sell into any of those markets, take that option deliberately rather than by accident. It changes where the label appears, and therefore what your audience sees.

The organic rule that catches AI UGC brands

Meta's Misinformation Community Standard is blunt: Meta requires people to disclose, using its AI disclosure tool, whenever they post organic content with photorealistic video or realistic-sounding audio that was digitally created or altered, and it may apply penalties for failing to do so.

Note what the sentence covers and what it does not. It names photorealistic video and realistic-sounding audio. Still images are not in that clause, even though Meta labels AI images elsewhere. An AI-generated video presenter is plainly inside it.

This matters for the common workflow where a brand posts AI creative to its Instagram feed and then boosts the winners. Meta states that AI-generated organic content from brands is subject to the organic labelling approach, that labelled organic content can be boosted provided it complies with the Advertising Standards, and that organic labels may remain present when a post is boosted directly. So the disclosure decision happens at post time, not at boost time, and it is a requirement rather than a courtesy.

Political ads: the one place Meta will reject you for it

The disclosure covers a photorealistic image, video or realistic-sounding audio created or edited with AI to depict a real person saying or doing something they did not, to depict a realistic-looking person who does not exist or an event that did not happen, or to depict a realistic event that allegedly occurred but is not a true recording. Meta says undisclosed ads may be removed and that repeated failure may result in account penalties.

The mechanics are two clicks: select Social issues, elections or politics in the Special Ad Category section of Ads Manager, then select "Includes media created or edited with AI" in the Ad creative section. The result is an AI info label near the "Paid for by" disclaimer, plus a line in the Ad Library reading that media in the ad was created or edited with AI. Meta exempts edits that are inconsequential or immaterial to the claim the ad makes, naming image size adjusting, cropping, colour correction and image sharpening. Since 1 June 2026 Meta also runs automated detection on these ads independently of what you declare.

What Meta still leaves unclear

"Significant edit" has no definition for third-party tools. Meta defines it operationally for its own features and leaves the third-party case to metadata detection. There is no published edit-type list you can check your pipeline against.
Detection coverage is unstated. Meta names C2PA as an example of the industry-standard methods it uses and does not publish the full set, or what happens when a file carries no provenance metadata at all.
"May not be available in your region" appears on both the ads AI info article and the political-media article, without a list of where the features are live. Availability is not the same as exemption.
Organic images. The Misinformation standard names photorealistic video and realistic-sounding audio. Whether an AI still image triggers the same disclosure duty is not stated in that clause.
Platform rules are not the law.Meta not requiring a commercial AI disclosure says nothing about the FTC's position on endorsements and testimonials, or about EU and state-level transparency rules. An AI presenter making a first-person product claim is a legal question before it is a Meta question.

The rule that will actually get your AI ad rejected

In practice, ecommerce advertisers running AI creative get rejected for the same reasons everyone else does, not for the AI. Meta's Advertising Standards require ads to comply with the Misinformation Community Standard, prohibit ads with content debunked by third-party fact-checkers, and require that ad components are relevant to the product and that what is promoted in the ad matches the landing page. Ads promoting branded content must tag the featured business partner using the branded content tool.

AI raises the risk on the first and third of those because it makes it trivially easy to generate a confident before-and-after, an invented statistic, or a synthetic person delivering a claim your product cannot support. The label is the easy part. The claim is the part that gets accounts restricted.

Practical checklist before you launch

Ordinary commercial ad, third-party AI tool: no self-disclosure needed, but expect an AI info label to appear behind the three-dot menu.
Watch for Meta's pre-publish message. Meta says you will see a notice during ad creation when an AI info label will appear on your ad.
Posting organically first? Use Meta's AI disclosure tool at post time for any photorealistic AI video or realistic-sounding AI audio, then boost.
Targeting the European region, California, New York, India or Taiwan? Decide consciously whether to self-disclose, since it moves the label onto the ad surface.
Anything touching a social issue, election or politics: set the Special Ad Category, then tick "Includes media created or edited with AI". This one is mandatory.
Never generate a real person's likeness or voice for an ad without permission.
Audit the claim, not just the pixels. Match the ad to the landing page, drop unverifiable before-and-afters, and keep substantiation for anything your synthetic presenter says.
Using a creator or partner? Tag them with the branded content tool.

Sources

Meta Transparency Center, "Introduction to the Advertising Standards"
Meta Transparency Center, Community Standards, "Misinformation"
Meta Transparency Center, Advertising Standards, "Ads about social issues, elections or politics"
Meta Business Help Centre, "About AI info on ads created or edited with generative AI tools"
Meta Business Help Centre, "About media created or edited with AI"
Meta Newsroom, "Expanding GenAI Transparency for Meta's Ads Products" (published 3 February 2025, updated 1 June 2026)

All six read directly in August 2026. Meta ships policy changes through product updates as often as through policy pages, so check the newsroom post's update stamp before trusting anything written about this, including this page.

TikTok runs a different regime entirely, with a genuine advertiser-side labelling duty: TikTok's AI content disclosure rules covers it. For the creative side of Meta buying, see our ecom creative testing framework. UGC Vids AI makes the ads themselves, from $49/month, free for 3 days.