The Oversight Board, a global expert body that reviews Meta’s most difficult and significant content-related decisions on Facebook, Instagram and Threads, has selected two new cases for consideration. Both feature explicit AI-generated images of female public figures and sharing them on Meta’s popular social media platforms.
Today’s announcement It implies that the council will investigate whether the management of the content on these platforms meets the same level of standards worldwide or whether there are significant differences that require a policy change. Specifically, the Supervisory Board chose the following two cases for examination:
“The first case involves an AI-generated image of a naked woman posted on Instagram. The image was created using artificial intelligence (AI) to resemble a public figure from India. The account that posted this content only shares AI-generated images of Indian women. Most of the users who responded have accounts In India, where deep fakes are more and more of a problem.
“The second case concerns an image uploaded to a Facebook group for creating AI. It includes an AI-generated image of a naked woman with a man groping her breast. The image was created with AI to resemble an American public figure, who is also named in the caption. Most of the users who responded have US accounts.” .
Meta’s content management has consistently faced criticism for double standards – making additional mistakes or delays in reviewing content in countries outside the US market or in non-English-speaking regions.
Therefore, it is clear that the selection of these specific cases is not accidental. In the first case, an AI-generated nude image of an unspecified Indian public figure was reported by a user for pornography. However, the report was automatically rejected because it was not reviewed within 48 hours. The subsequent appeal was also automatically dismissed.
Meta only removed the post after the case was picked up by the supervisory board. The company admitted that the image did violate its community standards related to bullying and harassment.
The second case is the opposite of the Indian story. An AI image created by an American user was deleted, and the decision was upheld even after the user appealed. In fact, this wasn’t the first time this particular photo was deleted because another user had already posted it. After the initial removal, the image was added to the so-called Media Matching Service Bank, which helps automatically remove offending images upon reposting.
The council is now accepting public comments on the issue (until Tuesday, April 30) before it reaches its final decision.
The Supervisory Board is an independent body, however, with a fairly significant authority that entitles it to the label “High Court of Headquarters”, it can overrule the company’s decision in certain cases. On top of that, the council can issue policy recommendations to Meta. These recommendations are not binding, although Meta must respond to them formally within 60 days.