Meta ran hundreds of ads for apps that turn real teens' photos into fake sex videos

Researchers at the nonprofit Tech Transparency Project (TTP) found more than 250 additional ads containing child sexual abuse material (CSAM) running on Facebook, Instagram, and Threads since the start of August, months after Meta had already deleted around 50 similar ads following earlier reporting by Wired. Some of the new ads were identical to ones Meta had supposedly removed. Overall, TTP says Meta has published more than 350 of these abusive video ads since the end of last year, and The Straits Times separately reported the group's tally as more than 300 ads reaching over 29,000 people.

Unlike the first batch, researchers say they were able to identify real children in the new ads, including a European royal family member whose officially released photograph was turned into a video described as a "graphic sex act." TTP also identified three American minors: two teenagers with public social media influencer accounts, and a stock photo model whose image had been labeled as "pre-teen" on stock photo websites. TTP is not naming the royal family member to protect the victim and says it contacted officials in that country.

Meta's ad library data shows the ads collectively reached more than 29,000 accounts in the European Union and roughly 7,000 in the United Kingdom, with more than 250 appearances in the US, 120 in Australia, and 80 in India, where Meta does not publish detailed ad performance figures. About 80 percent of the flagged ads ran in the US, according to TTP. Meta spokesperson Tracy Clayton said the company does not tolerate nudify apps or child exploitation, "whether real or AI-generated," and said many of the ads had already been removed.

The ads followed a consistent pattern: an ordinary photo of a preteen or teenage girl, overlaid with text claiming the image was "not just a photo" and that there were "no restriction on what can be done" with it. When played, the seconds-long videos morphed the child's face into graphic sexual content, often showing simulated sex acts. Clicking the ad directed users to download face-swapping or video-generation apps from Apple's App Store or Google Play, with slogans like "This is the AI that men actually use."

Meta says every ad on its platforms is reviewed before publication, but researchers found that some ads evaded detection because the clips of minors were brief or blurred, and some content was scaled to feature what Meta described as adults alongside the abusive material. In one case flagged by TTP, it took Meta about a week to review a live ad reported through its own tools, during which it accumulated hundreds more views. TTP also said some removed ads initially lacked disclosures indicating they had violated Meta's child exploitation policies, and those details only appeared after Wired asked Meta about the gap.

A former Meta employee who investigated nonconsensual imagery on the platform, David Pham, said people uploading the ads constantly adapted to avoid new detection systems, sometimes building workarounds within days of Meta closing a loophole. Many of the ads linked to Chinese developers and were placed through Chinese advertising resellers, since Meta's platforms are banned within China itself. Of the roughly 350 ads reviewed, about 300 pointed to AI image or video generators, most tied to China-based developers.

A separate BBC World Service investigation found Instagram running paid ads in India that promoted child sexual abuse material directly, using terms like "rape video" and linking users to Telegram channels selling the material for as little as 99 rupees, about 80 pence. When the BBC reported one such ad showing a crying young girl, Instagram responded 24 hours later saying its review team found the ad did not violate community standards; Meta later disabled several accounts and additional ads after the BBC's follow-up. Meta acknowledged that "no system is perfect, and our review process may not detect all policy violations."

In one particularly striking case cited by Ars Technica, a Facebook page called "Divine Productions," which claimed to represent the Church of Jesus Christ of Latter-day Saints, ran a CSAM ad that Meta only disabled after TTP flagged it directly. A Facebook page called "Divine Productions," which claimed to represent the Church of Jesus Christ of Latter-day Saints, ran a CSAM ad that escaped Meta's automated detection entirely. Katie Paul, TTP's director, said Meta profits from ad placements even as harmful content circulates, though Meta has said the flagged ads averaged fewer than 200 impressions and under $5,000 in total ad spend across the flagged batch.