September 24, 2026 7 min read By yan-dev

Why There Is "AI info" on My Ads, and What It Means

Meta detects third-party AI tools in ad media and adds an AI info label with no action from you. Where it shows, what triggers it, and when you may not touch it.

You opened the three-dot menu on your own ad, tapped About this ad, and found a line that says AI info. You did not tick anything in Ads Manager. Nobody reviewed the creative and made a call.

That label is applied automatically, and since 1 June 2026 it is applied to ads built with tools that have nothing to do with Meta. Here is what triggers it, where it shows up, and the one case where you must leave it exactly where it is.

Two placements, and the difference matters

This is the part worth understanding before anything else, because the two are not the same problem.

Behind the three-dot menu. The label sits inside About this ad, which almost nobody opens. It is a disclosure record, not something in the scroll.
Next to the Sponsored label. Meta moves it into the ad itself when the creative contains an AI-generated photorealistic human. That one is visible to every person the ad reaches, right beside your brand name.

Meta’s announcement on GenAI transparency in ads sets out both placements. If your label is behind the menu, you have a compliance record. If it is next to Sponsored, you have a creative decision to make, because a synthetic human is the thing Meta has decided the audience should be told about up front.

The two ways the label gets applied

1. You used Meta’s own generative features. Advantage+ creative, image expansion, background generation and the rest. Meta labels these when the tool made a significant edit, and its position is stated in the negative: if the in-house features “do not result in significant edits to the image or video and do not include a photorealistic human, then we will not apply any AI labels.” Small enhancements pass. Generated pixels do not.

2. Automated detection of third-party tools. This is the new one, live since 1 June 2026. Meta detects ad media created or edited with outside AI tools through industry-standard signals, then applies the label with no advertiser action required.

“Industry-standard signals” is the phrase that matters, and it is not vague. It means C2PA Content Credentials and the IPTC fields that mark a file as algorithmically generated: the same signed records Meta has read on organic uploads since 2024. The ads pipeline now reads them too.

So the label is not a judgment about your ad. It is a read of your file.

Which tools in an ad pipeline write that signal

The gap most advertisers fall into: you never generated anything, and the creative still arrives marked. That is because the signal is written by ordinary production steps, not by generation.

Canva, when a Magic feature touched the design: BG Remover, Magic Edit, Magic Expand, Magic Media
Photoshop and Lightroom, on generative fill, generative expand and AI-branded retouching
Firefly, on anything it produces
AI upscalers and background removers in the retouching step
A shared template or brand asset that picked up a credential once and passes it into every creative cut from it

That last one is why the label often appears across a whole flight at once rather than on a single ad. It is the same mechanism that marks a product photo when you only removed the background, and the same reason to audit the template rather than the exports.

⚠️ Where you do not touch this

Ads about social issues, elections or politics. Meta’s SIEP policy makes disclosure mandatory when an ad contains a photorealistic image, video or realistic audio created or edited with third-party generative AI. The policy is explicit about the consequence: “we will reject the ad, and repeated failure to disclose may result in penalties against the advertiser.”

If you run political or social issue ads, the label is not a problem to solve. It is a requirement to meet, and the automated detection is a backstop, not the rule. Disclose in Ads Manager and leave it alone.

The same holds anywhere else disclosure is genuinely owed. If your creative is AI-generated, the honest move is to be labelled as AI-generated.

What is actually actionable

What is left is the false positive: a creative that carries a credential from a production tool without being an AI ad in any meaningful sense. A background cut out of a real product photo. A real founder headshot that went through an AI retoucher. A template someone built in Canva two years ago.

There is no switch for this. Meta says the label needs no advertiser action, and the flip side of that is you cannot turn it off from Ads Manager. The only lever is the file you upload.

Fix it upstream. Find the step that writes the credential and change it, or rebuild the shared asset from a source you can vouch for. This is the durable fix, because it stops the whole catalogue from inheriting the mark.

Or clean the file before it goes up. A genuine re-encode rebuilds the media so the old credential and metadata are not carried into the new copy. On iPhone, CleanAi does this through Apple’s native CoreImage and AVFoundation pipelines, on-device, at full resolution. It works on files you own, and it changes what the detector reads, not what your audience sees.

Before you do either, answer the compliance question first: was AI genuinely used in a way that needs declaring? If yes, declare it. If no, you are correcting an inaccurate signal on your own file, which is a different thing entirely.

Where this is going

The direction is one way. Meta read provenance on organic uploads in 2024 and on ads from 2026, TikTok joined the C2PA Steering Committee in July 2026, and more tools write credentials every quarter. Building a creative pipeline where you know which steps write provenance is worth more than a per-campaign fix.

AI info on Meta ads FAQ

Can I turn off the AI info label on my ad?

No. Meta applies it automatically and states that no advertiser action is required, which also means there is no control to remove it. The only input you govern is the media you upload.

Why did my whole ad set get labelled at once?

Almost always a shared asset. A credential can survive re-editing, so one marked element in a brand template or a reused backdrop passes its record into every creative built on it. Audit the template rather than the individual exports.

Does the AI info label hurt ad performance?

Meta has not described it as a delivery or ranking signal, and it is a disclosure surface rather than a quality score. Where it can matter is placement: next to the Sponsored label it is visible to everyone, and that is a brand perception question rather than an algorithmic one.

My creative has no AI in it. Why is it marked?

A production tool wrote the signal. Background removal, generative fill, AI retouching and AI upscaling all write a Content Credential even when the subject is a real photograph. The platform reads that record rather than examining the picture.

Does this apply to political or social issue ads?

Those are held to a stricter, mandatory standard. Meta requires disclosure for photorealistic AI content in ads about social issues, elections or politics, rejects undisclosed ads and can penalise repeat failures. Disclose there; do not treat the label as something to clear.

Is this the same label as Made with AI on organic posts?

It is the same family and the same underlying signals, but a different surface. Organic posts show AI info on the post itself; ads show it in About this ad, or next to Sponsored when a photorealistic AI human is involved.