Tech Transparency Project (TTP) reported that Meta had failed to detect 332 ads this year containing child sexual abuse material (CSAM). The vast majority of the ads promoted AI applications developed in China, while a significant portion promoted “nudify” applications that digitally undress photos of children. TTP matched many of the ads with real photos of children previously published online. These included a photo of a young European royal being turned into a video depicting sexual assault, and an image of a 14-year-old Instagram influencer being turned into another sexually explicit video. Another ad used a photo of a child wearing pink sportswear. The ads were said to mostly depict preteen girls.
The ads used phrases such as “I can show you more” and “This is the AI men actually use” to encourage users to click. TTP had previously reported 53 ads to Meta in August together with Wired. Although Meta said at the time that it was developing new systems to better detect child abuse, TTP found that hundreds of new CSAM ads had been published following that statement. Wired also identified hundreds of ads containing similar wording, as well as other “nudification” ads using explicit images of adults.
Meta removed the ads reported by TTP and Wired; however, TTP said that the company was slow to delete some content and applied its policies inconsistently. After TTP reported more than 300 ads, Meta removed approximately half of them from its archive within a few hours. However, many ads were not flagged as violating the child abuse policy until TTP pointed this out. Reviewing some reports took at least a week, and often no action was taken. Meta was also criticized for removing one ad while leaving up a similar ad that used the same image and text. One ad was deleted days after it was reported, despite increasing from four impressions to 330 impressions.
TTP reported that the ads reached more than 29,000 people in the EU and 6,800 people in the UK; approximately 80% were shown in the US. It was reported that the China-based advertising partners GIMC, Meetsocial and BlueFocus had published the same CSAM content without detection. While the attorneys general of Michigan and Florida launched an investigation, Senator Mark Warner (D.-Va.) called on Meta to stop the ads. Meta said it had no tolerance for “nudify” apps or real or AI-generated child abuse. Australia’s eSafety internet regulator also requested information from the company.
Why it matters
The findings raise questions not only about Meta’s capacity to remove content in ad moderation, but also about its ability to identify violations at the initial stage and apply the same criteria to similar ads. This distinction shows that the growing visibility of abusive AI applications in the advertising ecosystem needs to be examined not only in terms of user safety, but also with regard to the platform’s policy enforcement processes. The fact that the ads reached different regions demonstrates that the problem was not limited to a single piece of content or a narrow group of users. The actions taken after reports by TTP and Wired leave open the question of whether company statements can be considered sufficient without external oversight. Investigations by attorneys general and the information requested from Meta have brought the process of determining how responsibility will be shared between advertising partners and the platform to the fore.
Background
Meta is not a new name in the FikirPilot archive: we have published 6 articles mentioning this name in the past 90 days; the latest was dated 4 September 2026.