Meta’s independent Oversight Board has issued a sharp rebuke of the tech giant’s approach to artificial intelligence-generated content, describing its current policies as “consistently and fundamentally inadequate” to handle the surge of synthetic material on its platforms. The ruling follows the board’s release of two recent decisions addressing specific incidents involving AI-created videos.
In one case, a realistic synthetic video was circulated featuring a Scottish politician stating that “refugees are welcome here, even if they rape our women, because white people do that too.” The Oversight Board noted that the individual whose voice was replicated described the impact of the video as “quite traumatic.” Meta chose not to remove the footage when it was initially reported, arguing that it had not been flagged by trusted partner organizations and did not appear to interfere with electoral processes. Furthermore, the post lacked an AI label because the uploader did not voluntarily disclose its origin.
The board concluded that the video violated Meta’s prohibitions against hateful conduct and should have been prominently labeled as AI-generated. In a separate incident, an AI-generated video mocked a volunteer promoting menstrual health education after her interview footage was digitally altered and shared across social media. Meta automatically dismissed reports and appeals regarding the post, only removing it for bullying violations after the Oversight Board intervened.
Pamela San Martin, co-chair of the Oversight Board, emphasized that both cases illustrate a troubling pattern in which AI tools are weaponized against women. “From politicians to private citizens, AI-generated deepfakes are increasingly being used to harass and silence women from engaging in public discourse,” San Martin said. “These cases demonstrate a broader, troubling pattern in which women who engage publicly on issues are disproportionately subjected to harassment and misinformation.”
This criticism aligns with previous findings by the board, which has repeatedly urged Meta to adopt stricter guidelines for AI content. Despite these recommendations, Meta has made limited adjustments to its labeling practices, which currently rely on small banners reading “AI Info.” The company has also declined to implement similar labeling standards in other recent cases involving AI-generated content.
Meta did not immediately respond to requests for comment regarding the latest recommendations. The company now has a 60-day window to issue a formal response, during which it is expected to consider the board’s suggestion to prioritize likely AI content for review when it is reported by users.
I find it hard to believe they still rely on users to self-disclose AI content. How is that ever going to work in practice?
This is terrifying. Politicians and everyday women are being targeted with synthetic harassment, and Meta’s current policies are completely failing to stop it.