Meta’s Ad Library Exposed AI-Generated CSAM Ads That Passed Review, Report Says

Researchers at the Tech Transparency Project (TTP), whose findings were reported by Wired on Aug. 5, searched Meta’s public Ad Library, a transparency database that lets anyone review advertisements running across the company’s platforms.

There they identified more than 50 image and video advertisements containing AI-generated child sexual abuse material (CSAM) that had appeared across Facebook, Instagram, Messenger and Threads.

The investigation, reported by Wired based on research from the Tech Transparency Project (TTP), identified more than 50 AI-generated image and video advertisements containing child sexual abuse material (CSAM) in Meta’s public Ad Library.

“These ads made no effort to mask the images or hide what they were promoting. It’s important to point out that this isn’t content posted by third parties on Facebook or Instagram, these are ads that were reviewed, approved, and allowed to run by Meta, never encountering interference while the company collected the ad dollars,” Katie Paul, director of the Tech Transparency Project, told Wired.

This matters beyond Meta

Unlike ordinary posts uploaded by users, paid advertisements are expected to pass Meta’s review system before reaching audiences. According to Meta, advertising is primarily reviewed using automated systems, with human reviewers involved when additional scrutiny is needed.

What makes the findings particularly significant is that these were paid advertisements, not ordinary user posts. User-generated content is often moderated after publication through automated detection and user reports. Advertisements, by contrast, are expected to pass Meta’s review process before appearing to users.

Meta says ads are primarily reviewed using automated systems, with human reviewers stepping in when additional scrutiny is needed. The investigation suggests that safeguard failed in cases that should have been intercepted before publication.

Part of a broader pattern

The findings also fit into a series of platform-safety investigations during 2026 rather than standing alone.

Meta said many of the advertisements identified by TTP predated newer AI-powered detection systems introduced to improve ad screening. The company also reiterated that sexual exploitation is prohibited and that it works aggressively to remove violating content.

“Sexual exploitation is horrific, and we work aggressively to keep it off our platform. The majority of these ads had minimal reach and many were disabled before WIRED shared them. Many of these ads also predate new AI technology we launched recently to better detect and block violating ads at upload and we’re constantly improving these systems,” a Meta spokesperson said in a statement to Wired after the publication sought comment.

Why AI makes this harder

Generative AI systems can rapidly produce new variations of illegal material, making it harder for classifiers trained on previous examples to recognize every version. As image-generation models improve, detection systems must continuously adapt to keep pace.

“From removing over 36 million pieces of child sexual exploitation content last year to taking legal action against nudify app developers, we will continue to relentlessly fight this abuse,” the Meta spokesperson said in the company’s response to Wired.

That technical challenge is real across the industry. But it also raises expectations for platforms that approve advertisements before publication. If automated systems are responsible for reviewing billions of ads, the highest-risk categories require particularly reliable safeguards.

What comes next

The investigation arrives as regulators increasingly scrutinize AI-generated illegal content. Spain has already directed prosecutors to investigate Meta, X and TikTok over synthetic CSAM, reflecting a broader push to examine whether platforms’ moderation systems can keep pace with rapidly advancing generative AI.

For Meta, the incident also highlights an irony: the company’s transparency initiative ultimately worked as intended. Independent researchers were able to identify problematic advertisements because the Ad Library made them publicly searchable.

“This raises significant questions about how seriously Meta is taking child exploitation in paid advertisements if its own systems can identify identical ads but it does not remove those identical ads for CSAM material,” Katie Paul told Wired

As generative AI becomes more capable, platforms will increasingly be judged not only by the sophistication of their AI tools, but by whether their safety systems can prevent harmful content from being approved, monetized and distributed in the first place.