Meta ran over 300 ads containing AI-generated child sexual abuse material (CSAM) this year, many linked to Chinese developers, according to a Tech Transparency Project investigation and reporting by WIRED. The ads used real children’s images, including a European royal family member, and were promoted through platforms like Facebook and Instagram before being partially removed.
The discovery of hundreds of ads containing AI-generated child sexual abuse material (CSAM) on Meta’s platforms has exposed significant gaps in the company’s content moderation systems, according to a report by the Tech Transparency Project (TTP) and coverage by WIRED. The ads, which ran this year, included explicit imagery of real children, some of whom were identified through social media posts, and were linked to apps developed by Chinese entities, including state-controlled companies.
The Scale of the Problem
TTP’s investigation found more than 300 Meta ads containing CSAM, many of which used AI to alter real children’s images into explicit content. The ads often featured a snippet of adult pornography followed by a manipulated photo of a child, making it appear as if the child was engaged in sexual activity. One ad included a photo of a member of a European royal family, which had been shared on social media, according to TTP. The ad was promoted through a Chinese AI app and targeted adult male users.
WIRED reported that the ads pointed users to nearly 50 apps across Google’s Play Store and Apple’s App Store, with 300 of the 350 ads identified being linked to Chinese developers. Some apps had already been removed by the time the reporting began. Apple said 17 of the 20 apps included in the research had been taken down, while Google claimed it removed all the apps cited in the report. However, TTP found that Meta continued to run dozens of such ads, including the one featuring the European royal, in the days following its findings.
Chinese Developers and Ad Networks
The majority of the CSAM ads were linked to Chinese developers, with several involving “top-tier” ad agency partners in China, according to TTP. These partners, known as “resellers,” channel billions in Chinese advertising to Meta’s platforms despite the company’s ban in China. Reuters has reported that Meta often exempts these resellers from penalties for policy violations, allowing their ads to bypass automated moderation systems.
Meta’s Response and Ongoing Issues
Meta said it has “incorporated” the language used in the ads into its moderation systems and that continually takes proactive steps to detect and remove apps with harmful content.
However, TTP found that Meta’s Ad Library had previously omitted child-related violations in ads, suggesting the company may undercount instances of CSAM. After TTP shared its findings, Meta removed about 150 ads and revised records for over 100 others, marking them as violating policies on child exploitation.
Apple and Google both claimed to have taken action against the apps. Apple stated that developers who initially submitted compliant apps later added “nudify” capabilities, which violate its guidelines. The company said it had suspended 17 of 20 apps in the report and taken action against the remaining three. Google said it had removed all the apps cited in the research, adding that it restricts search terms like “nudify” on its store.
Legal and Ethical Implications
The ads raise significant legal questions, as they involve content that is explicitly illegal under U.S. law. Section 230 of the Communications Decency Act protects platforms from liability for user-generated content, but it does not shield them from criminal charges related to child exploitation. The FBI has warned that AI-generated CSAM is also illegal, and the National Center for Missing and Exploited Children (NCMEC) has received data from TTP to help identify and block such material.
The revelations have intensified scrutiny of Meta’s content moderation practices, particularly as the company positions itself as a leader in child safety. TTP’s findings contradict Meta’s public efforts to push rivals like YouTube and TikTok to adopt similar teen safety measures. The investigation also underscores the growing risks posed by AI-generated content, which is increasingly being used to create and distribute CSAM at scale.
This is not something that it should take a small nonprofit doing research to stumble across and then have to investigate and report,
said a TTP researcher.
