Meta ran AI-generated CSAM ads for months, investigation finds

More than 50 paid advertisements containing AI-generated child sexual abuse material circulated across Meta's platforms before the company pulled them down, according to reporting by WIRED published August 5, 2026. The ads appeared on Facebook, Instagram, Messenger, and Threads, some of them running since November 2025.

Researchers at the Tech Transparency Project uncovered the material by combing through Meta's public ad library. Several of the advertisements steered users toward AI "nudify" or undressing services, tools that digitally strip clothing from images of real people, often without their consent.

The campaigns weren't confined to a single region. Targeting data showed reach into the United States, the United Kingdom, and multiple European countries, including France, Germany, Ireland, Italy, the Netherlands, Spain, and Sweden. One ad alone reached 2,563 accounts across those European markets.

Katie Paul, who directs the Tech Transparency Project, was among those who flagged the findings publicly. Meta took the ads down only after WIRED's inquiries, with the company's own ad library later confirming the content violated its rules against child sexual exploitation, nudity, and sexual activity.

Not an isolated incident

This wasn't Meta's first brush with this problem in 2026. A BBC investigation published July 3 found Instagram serving paid ads in India using search terms like "rape video" and "child video," prompting scrutiny that has now expanded to Western markets.

Tech Transparency Project researchers also traced several of the advertiser accounts back to Chinese companies or profiles with negligible followings, a pattern that suggests coordinated, low-visibility ad placement designed to slip past moderation.

Meta has said it built newer automated detection tools specifically to catch and block policy-violating ads before they go live. The company acknowledged, however, that some of the ads in question slipped through those safeguards regardless, remaining in the ad library for months in certain cases.

Meta's own ad library later flagged the removed ads as violations of its rules on child sexual exploitation, nudity and sexual activity.

The recurrence of similar ad patterns across two continents and two separate investigations within weeks of each other raises pointed questions about whether Meta's pre-publication screening can keep pace with AI-generated abusive content, particularly as generative tools make such material cheaper and faster to produce at scale.