Researchers say Meta ran 350+ ads with CSAM in 2026 so far; some featured morphed images of real children, and many linked to “nudification” apps from China
Early last month, Meta deleted around 50 ads posted on Facebook, Instagram, and Threads that directly included child sexual abuse material …
Context & Ripple Effects
Meta had already pursued one alleged source of this advertising, suing the maker of CrushAI in 2025 to block ads across its apps. The new findings follow an August report that a nudify-app ad ran on Meta services, showing that enforcement against a single advertiser had not closed the paid-distribution route for this category. Meta's earlier action against the CrushAI maker is the relevant backdrop.
The issue also expands Meta's child-safety exposure beyond recommendation systems: 2023 testing found its products promoting sexualized content involving minors, while the reported ads point to failures in the separate ad-review pipeline. Earlier testing of Meta's recommendation systems had already put the company’s safety controls under scrutiny.
First-order effects
- Meta’s removal of roughly 50 identified ads cuts off paid acquisition for the nudification apps promoted through Facebook, Instagram, and Threads, while leaving Meta to review why hundreds of reported ads passed its approval systems.
- People whose images were allegedly morphed into abuse material face a platform-removal process after the ads have already been distributed, making rapid detection and takedown central to the immediate harm response.
Second-order effects
- Meta’s youth-safety defense faces sharper scrutiny in the state-AG litigation context: the reported material concerns paid placements, and public reaction highlighted the gap between Meta’s stated ad-review process and the ads researchers found. State AGs' youth-safety allegations against Meta remain contested by the company.
- Nudification-app operators become more dependent on channels outside Meta’s ad products when platform enforcement targets both the abusive creative and links to their services; Meta had already targeted an alleged advertiser behind CrushAI. Meta's lawsuit against the CrushAI maker
Third-order effects
- If repeated cases span recommendation and paid-ad systems, child-safety controls will be judged less as isolated moderation tools and more as an end-to-end governance requirement covering advertiser onboarding, creative review, link destinations, and post-publication detection.
- Synthetic-image abuse is pushing platforms toward a dedicated control layer for provenance, detection, and distribution restrictions, rather than treating manipulated imagery as ordinary policy-violating content.
The trend: AI-enabled image manipulation is turning advertising review into a frontline safety system, with platforms pressured to police both harmful media and the services promoted alongside it.