Meta Ran Ads That Contained AI-Generated Child Sexual Abuse Imagery

Meta, the social media giant, has been found to have run numerous paid advertisements featuring artificial intelligence-generated child sexual abuse imagery over the past nine months. These ads, targeting individuals in the United States, United Kingdom, and numerous European countries, were approved and published by Meta without detection.

The ads, discovered by researchers at the Tech Transparency Project (TTP), included explicit images and videos depicting sexual acts involving children. Some ads linked to nudify or undressing apps, while others showcased minors in sexually suggestive situations. Despite the ads being removed by Meta for violating their policies, they had been accessible in Meta's ad library for months.

One video ad featured a thumbnail image of a child on the floor with text suggesting the use of imagination, while the video itself displayed adults engaging in sexual acts and then altered to include the child's face in a sexual scenario. Another ad depicted a young girl in a sexual position with text encouraging further explicit content.

Although many of the ads only reached a handful of accounts, at least one reached 2,563 accounts in Europe. Meta's ad library does not provide details on ad performance in the United States or all European countries, so the overall reach may be higher. Katie Paul, director of the TTP, emphasized that the ads did not attempt to conceal their content and were approved and published by Meta without interference.

Meta acknowledged the issue, stating that sexual exploitation is horrific and that they work aggressively to keep it off their platform. The company removed the ads and mentioned that only a minimal number had minimal reach, many of which were disabled before WIRED shared the information. Meta also noted that some ads predated recent AI technology launched to better detect and block violating ads at upload.

However, some of the offending ads managed to bypass Meta's new AI technology. The TTP initially discovered around two dozen ads displaying abusive content, but later found around 30 more. Many of these ads were published after WIRED's request for Meta's information, with multiple ads being live and shown to accounts at the time of discovery.

Paul described the discovered ads as some of the worst instances the researchers had seen. One video ad showed a thumbnail image of a girl putting on lip gloss, which morphed into the child performing a sex act when the video started playing. Meta's advertising policies strictly prohibit child sexual and exploitation materials, and the company makes reports to NCMEC when such content is identified.

The ads were removed from Meta's platform and labeled as violations of rules on child sexual exploitation, abuse, and nudity, as well as adult nudity and sexual activity.