Meta's new AI disclosure rules - what buyers need to know
Meta has ramped up enforcement on AI-generated ad content. If AI was used to create or significantly modify an image, video, or audio in your ad, it now has to carry an "AI info" label.
The tag appears automatically when a creative is made or edited via Advantage+ Creative, Photoshop, DALL-E, or other AI tools. Usually it sits quietly in the "About this ad" section, but for photorealistic AI-generated people, it can show up right next to the ad itself - in plain sight, not buried in a menu.
What this means in practice:

Undisclosed AI content is already the 3rd biggest rejection reason for ads - not a small category anymore

Detection goes beyond fully generated images - background replacement, synthetic voiceovers, and "animated" faces all get flagged too

If undisclosed AI is detected, the ad gets rejected

For photorealistic AI people specifically, the visible label can hit CTR and creative trust - people notice it more than a hidden disclosure

If Meta considers the AI content "deceptive" (e.g. a realistic video depicting something that didn't happen), reach can get throttled quietly - no ban, no notice, the campaign just stops delivering the way it used to
What buyers should do:
AI is fine for ideation and drafts, but keep a human in the loop on the final creative assembly. Check your source files for C2PA metadata from tools like Midjourney, DALL-E, or Stable Diffusion - that's exactly what Meta's detection is picking up on.
We're seeing this affect campaigns across multiple GEOs, so it's worth reviewing your current creative pipeline now rather than after a rejection wave.
Questions on traffic, GEOs, or creative strategy - reach out to our managers:
Dima, Jane, or Vova
OmniCPA