Meta ads apply an AI info label automatically
Background generation, image generation and add-animation in Ads Manager stamp AI info with no advertiser toggle. C2PA from third-party tools does too.
Organic "Made with AI" / "AI Info" and ads "AI info" are different surfaces. Meta's business help page is explicit: labelling for ads is not the same as labelling for organic content. Background generation, image generation, and add-animation inside Ads Manager stamp AI info on the ad with no advertiser-facing off switch. C2PA from Photoshop, Firefly, DALL-E and the rest of the third-party toolchain does the same when the file still carries it.
The FTC ad disclosure is a third layer. An AI info line in About this ad does not tell a viewer they are being sold to. UGC ads sold to a brand still need the paid-partnership sentence on the creative.
No advertiser toggle on Meta's own tools
Meta lists the generative features that auto-apply the label: Background Generation, Image Generation, and Add Animation. You find the result under the three-dot menu → About this ad → AI info. Sometimes AI info also sits next to the Ad / Sponsored mark on the face of the ad. Meta says minor changes — resize, colour correction — do not get the label. Generating a background, a scene, or an animated still does.
There is no documented "deliver this Advantage+ enhancement but do not disclose it" control. If the feature ran, the label is part of the delivery. For regulated categories that is a compliance fact, not a brand-safety preference: a background you did not approve can still ship with AI info attached.
Advantage+ Creative's other tweaks are not all generative. Brightness, a standard info badge, catalogue adaptations — those are not the same as Image Generation. Read the AI marker next to the enhancement. That marker is the one that predicts the label.
Video generation inside Advantage+ is in the same family. Meta's own note on that feature points back to the same AI-info behaviour. If the ad account is building clips from stills inside Ads Manager, plan for the label as a default, not as a surprise in reporting.
C2PA from the file you uploaded
Third-party generative tools are the other automatic path. Meta says it uses industry-standard detection, and it names C2PA, to identify ads created or edited with tools it does not own. If the export still has a Content Credentials manifest that asserts generation or a significant generative edit, the ad can pick up AI info at upload.
Two operational consequences:
- A clean Ads Manager setup does not save a dirty file. Turning off Background Generation does nothing if the hero still was made in Firefly and uploaded with its manifest intact.
- A stripped file is not a completed disclosure. Organic photoreal video can still need a self-disclosure. Ads that use Meta's own generative features will still be labelled at source. Stripping C2PA to "keep the ad clean" is how you lose the cooperative signal and keep every other duty.
Inspect the export you are about to upload, not the Figma frame. If the generator wrote Content Credentials, Meta can read them.
Organic AI Info is a different surface
Do not copy an organic workflow into Ads Manager and assume the same switches exist.
Organic labels fire from file signals, from the upload toggle, and sometimes from classifiers. You can at least choose to self-disclose on a Reel. The profile-level AI creator badge is a fourth, opt-in object, and it does not travel into ads at all.
Ads labels fire because you used a named generative feature, or because the file declared itself. The viewer path is About this ad, and sometimes the face of the unit next to Ad. There is no profile badge for an ad account. There is no "this advertiser is an AI creator" setting that substitutes for creative-level AI info.
If you produce UGC for a performance team, hand over the file and a note on whether Meta's tools or a third-party generator touched it. The media buyer cannot infer that from a Reel that ran on your personal account.
The FTC layer is still yours
AI info answers "was this made or significantly edited with a generative tool." It does not answer "did someone pay for this," "is this a real customer's experience," or "is this person a synthetic performer." Those are advertising-law questions. The FTC ad disclosure is a separate sentence. New York's synthetic-performer rule is a separate sentence again for ads that reach that state with a fully synthetic human.
Layer them on the creative, not only in About this ad. A muted Stories placement will not open the menu. A paid-partnership line burned into the first seconds still reads. AI info in a three-dot sheet does not.
Do not hide a paid integration behind Meta's automatic AI info and assume a regulator will treat the sparkle as a material-connection disclosure. It will not.
FAQ
Can I turn off AI info on an ad that used Background Generation?
Not by a documented advertiser toggle. Meta applies the label when those generative features create or significantly edit the asset. If you do not want the label from that path, do not run the feature.
Does cropping a generated still in Ads Manager remove the label?
Crop, resize, and colour correction are the examples Meta gives of edits that do not trigger AI info on their own. They also do not erase a C2PA assertion already in the file, and they do not undo Image Generation you already ran.
Will organic AI Info on my Reel carry into the ad?
No. Ads are a separate labelling path. Re-uploading the file to Ads Manager starts that path: Meta's generative features if you use them, C2PA if the file still has it. Your Instagram profile badge does not come along.
Is the AI info label the same as saying the ad is paid?
No. AI info is about generation. "Ad" / "Sponsored" is about the commercial nature of the unit. A material-connection disclosure for an influencer integration is a third line, on the creative, under the FTC Endorsement Guides. Keep all three if all three are true.