Meta approved and monetized 332 ads with AI child abuse imagery

This was not content that slipped past a feed filter. It was paid inventory that went through a review process and produced revenue.

What the investigation found

The Tech Transparency Project published its findings on September 8, 2026. Researchers identified 332 advertisements containing child sexual abuse imagery that cleared Meta’s advertising review and ran on Facebook and Instagram.

262 of them, close to 80 percent, ran in the United States. 298 of the 332 promoted AI image and video generation applications, and 182 pointed specifically to nudify tools capable of producing synthetic sexual imagery. TTP says most of the material consisted of AI-manipulated photographs of children, and that some ads used images of real children.

Meta’s response

Meta had already removed 183 of the ads by September 1. After TTP shared its list on September 2, the company took down the remaining 149.

Meta said it does not tolerate nudify apps or child exploitation whether real or AI-generated, and that most of the ads had fewer than 200 impressions with total spend under 5,000 dollars.

The supply chain, not the platform

The reach figures matter less than the mechanism. An ad that runs is an ad that was reviewed, approved and billed, which places this in a different category from moderation failure on user-generated content.

It also points at a chain rather than a single company. The nudify applications these ads promoted were reachable through the App Store and Google Play, which means the review failure spans more than one gatekeeper.