What the policy actually requires
Meta's AI disclosure rule applies to any ad creative that was generated or significantly altered by artificial intelligence. This includes:
- Images created entirely by AI tools (Midjourney, DALL-E, Adobe Firefly, Canva AI, etc.)
- AI-generated voiceovers or voice clones — even if the voice sounds like a real person
- Product images where the background, lighting, or composition was AI-generated
- Videos with AI-generated faces, backgrounds, or animations
- Images where AI was used to significantly alter a real photo (replacing a background, changing a person's appearance, generating a setting that didn't exist)
Minor AI-assisted edits — removing an unwanted object from the background, adjusting color, improving sharpness — are generally not covered by the disclosure requirement. The rule targets content where the AI is doing creative generation, not cleanup.
How Meta detects undisclosed AI content
Meta's ad review system — which they refer to internally as the Andromeda system — reviews not just your ad copy but your images, video, audio, and even the landing page your ad links to. AI detection is baked into this review process.
Modern AI detection tools can identify common patterns in generated images: pixel-level regularities, artifact signatures, and patterns in how lighting behaves that differ from real photographs. Meta's system flags these during the automated review that happens before your ad goes live.
How to add the disclosure
When creating or editing an ad in Meta Ads Manager, look for the AI content disclosure toggle in the ad creative section. It appears as a labeled checkbox or toggle depending on your campaign type and placement.
Navigate to the specific ad (not ad set or campaign) where you're using the AI content.
Scroll down in the creative editing panel. Meta has added a disclosure field in the ad creation flow. Toggle it on and select the type of AI content present.
Disclosure needs to be applied to existing active ads, not just new ones. Go through your active campaigns and add the label to any creative that uses AI-generated content.
What about AI tools inside Canva or other platforms?
If you used Canva's AI image generator, Adobe Firefly, or any other third-party AI design tool to create your ad image, and then uploaded that image to Ads Manager — the disclosure rule still applies. The source of the AI generation doesn't matter; what matters is that the creative output is AI-generated.
With Canva Grow 2.0 now allowing direct publishing to Meta, this is especially relevant. If you're generating AI images in Canva and pushing them directly to Meta as ads, you need the disclosure label on those placements.
The bottom line
AI creative is fine. Meta isn't banning AI-generated ads — they're requiring you to be transparent about them. The label doesn't hurt your performance. The missing label does.
Take 10 minutes to audit your active campaigns. If any of your current ads include AI-generated images, voiceovers, or significantly altered visuals, add the disclosure now. It's a minor step that protects your account health from an avoidable policy strike.