Meta's Ad System Approved and Ran Dozens of AI-Generated Child Abuse Ads

Meta’s Ad Approval System Failed to Block AI-Generated Child Abuse Content

Between November 2025 and August 2026, more than 50 paid advertisements containing AI-generated child sexual abuse material ran on Meta’s platforms with the company’s explicit approval, according to research by the Tech Transparency Project (TTP), an independent watchdog group. The ads, discovered in Meta’s ad library and confirmed by Wired AI, targeted users across the United States, United Kingdom, and over a dozen European nations, reaching thousands of accounts before removal. This marks the second documented case in recent weeks of child sexual abuse material distributed through Meta’s paid advertising system.

How AI-Generated CSAM Bypassed Meta’s Review Process

The ads combined AI-generated imagery of minors with sexually explicit content and links to “undressing” or “nudify” applications designed to digitally remove clothing from images. According to Wired AI’s reporting, one video ad used a child’s face in a thumbnail and, when clicked, superimposed that same image onto footage of adults engaged in sexual acts. Another ad displayed an image of a young girl in a sexual position accompanied by text reading “I can show you more.”

Katie Paul, director of TTP, told Wired AI that the ads “made no effort to mask the images or hide what they were promoting,” emphasizing that these were not user-posted content but paid advertisements “reviewed, approved, and allowed to run by Meta” while the company collected ad revenue. The ads remained discoverable in Meta’s transparency library for months after posting, with most undetected during active distribution.

Geographic Reach and Audience Scale

Meta’s ad library data, reviewed by TTP and reported by Wired AI, shows that while the majority of ads reached only dozens of accounts, at least one advertisement was distributed to 2,563 accounts across eight European countries: France, Germany, Ireland, Italy, the Netherlands, Spain, Sweden, and the United Kingdom. Targeting filters in Meta’s system suggest several ads were specifically aimed at male users.

Why This Matters

This incident exposes a critical vulnerability in Meta’s automated content moderation infrastructure: the company’s systems failed to detect AI-generated child sexual abuse material at the point of ad approval, a stage that should function as a final safeguard before monetized distribution. The fact that ads remained catalogued and accessible in Meta’s public ad library for months after posting suggests neither automated scanning nor human review caught the violations during the campaign lifecycle.

For regulators scrutinizing Meta’s compliance with child protection laws—particularly the Digital Services Act in Europe and potential US legislation—this incident demonstrates that paid ad systems may require separate, more stringent review workflows than organic content. For advertisers and platforms relying on Meta’s systems, the discovery raises questions about the reliability of safety certifications on ad placements. The persistence of similar violations weeks apart suggests systemic rather than isolated failures in content classification.