Meta’s Oversight Board has referred to as on the social media firm to strengthen its safety for strange individuals focused by sexualized deepfakes. The Board recommends the addition of AI-generated impersonations in Meta’s Grownup Sexual Exploitation coverage, arguing that these photos and movies are non-consensual by default. It additionally needs Meta to permit customers to designate “related accounts,” corresponding to trusted family and friends, who can report potential violations like non-consensual intimate imagery on their behalf.
Lastly, the Board recommends making AI-generated sexual impersonation a separate class from harassment and nudity within the firm’s content material reporting and enchantment types. For the time being, solely the residents of Texas and Florida have entry to a specialised kind that lists deepfake intimate imagery as a cause for the report. The Board needs all Meta customers to have entry to it, as a result of “AI-generated non-consensual intimate abuse, together with sexualized impersonations, is a world situation.”
The Meta Oversight Board got here up with these suggestions after investigating an incident whereby the corporate saved ignoring a person’s report in regards to the sexualized impersonation of a buddy on Instagram. It launched the investigation after it acquired an enchantment from the person who reported an AI-generated video on Instagram exhibiting a girl adjusting her costume, along with her underwear seen in a number of frames. In response to the Board’s report, the reporter stated they had been a buddy of the individual being impersonated within the video with out content material. The one that was depicted within the AI video had already closed her account.
Two customers initially reported the video to Meta, however the firm did not take away the deepfake. The person who appealed to the Board had submitted an enchantment to Meta first, however the firm nonetheless did not take away the video from Instagram. After the Board itself raised the difficulty with Meta, the corporate merely made the publish adults-only however concluded that it didn’t benefit removing beneath its group requirements.
Meta instructed the Board that on the time the publish was initially reported, it had no indications that the person within the AI deepfake was an actual individual. If the depicted particular person herself had reported the video, it might have violated its Grownup Sexual Exploitation coverage. The self-report would have served as a transparent signal of non-consent. Different credible indicators of non-consent in Meta’s eyes are studies from legislation enforcement, media or trusted companions. Captions or web page titles suggesting that photos or movies are shared in a “vengeful or sensationalist method” will work, as nicely.
The Board says Meta’s responses to its investigation signifies that the one viable approach for private figures to ascertain non-consent is to self-report. It would not be straightforward for strange individuals to get legislation enforcement or the media concerned, in any case. These avenues are largely accessible to public figures. Meta is required to answer these suggestions, but it surely’s not obligated to implement them. If it does select to undertake them, the Board will monitor its implementations. For the actual case that began this investigation, the Board has overturned Meta’s choice to depart the video up and has required the corporate to take away the publish.
“It’s clear that the dimensions, velocity and class of AI instruments have resulted in a proliferation of AI-generated sexualized non-consensual content material globally. The unfold of sexualized deepfake movies results in reputational and psychological hurt, which disproportionately impacts ladies and women, and has a chilling impact on participation in social and political life,” the Board wrote in its report of the investigation.
This is not the primary time the Board has been crucial of Meta in issues involving AI content material and moderation. In mid-2025, it referred to as the corporate’s incapacity to implement its guidelines persistently “incoherent and unjustifiable.” Again in March, the Board urged Meta to create a brand new rule for AI content material that is separate from its misinformation coverage. That advice got here from an investigation involving an AI-generated video that claimed to indicate broken buildings within the Israeli metropolis of Haifa. The video was posted by an account claiming to be a information outlet however was really ran by a person within the Philippines.
