Meta's self-appointed Oversight Board is ordering the company to delete an AI-generated sexualized video from Instagram and rewrite its rules to treat all AI impersonation as non-consensual by default — handing an unaccountable global tribunal more power to decide what Americans can post and see online.
The Board's ruling, issued after a user appealed Meta's refusal to remove a deepfake video showing a woman adjusting her dress, goes far beyond one takedown. The recommendations would add AI-generated impersonations to Meta's Adult Sexual Exploitation policy, create a new reporting category for AI sexual content, and let users designate "connected accounts" — friends and family — to report violations on their behalf. Currently, only users in Texas and Florida have access to a specialized reporting form for deepfake intimate imagery. The Board wants that expanded worldwide.
Here's how the case unfolded: Two users reported the AI-generated video to Meta. The company left it up. A friend of the woman depicted — who had already closed her Instagram account — appealed to Meta directly. Meta still refused to remove it, instead restricting the post to adults only. Only after the Oversight Board intervened did Meta face a binding order to take the post down.
Meta told the Board it had no indication the depicted person was real at the time of the original report. Under current policy, the company treats a self-report from the person depicted as a clear sign of non-consent. Other acceptable indicators include reports from law enforcement, media, or "trusted partners," or captions suggesting the content was shared in a "vengeful or sensationalist manner." The Board dismissed those standards as inadequate for ordinary people who can't easily get law enforcement or the press involved. According to Engadget, the Board concluded that self-reporting is effectively the only viable way for non-public figures to establish non-consent — a high bar when the person targeted may not even know the content exists.
That's a fair concern. But the Board's fix — declaring all AI-generated impersonation non-consensual by default — is a presumption that flips the burden of proof and expands censorship authority. Meta is required to respond to the recommendations but isn't obligated to adopt them. If it does, the Board says it will monitor implementation.
The Verge framed the story as the Board correcting Meta's failure to recognize non-consent. Engadget emphasized protections for "regular people" and the psychological harm of deepfakes. Neither outlet raised the structural question: why does a platform used by millions of Americans answer to a board it created and funds — one that operates with no democratic mandate?
Free speech has always meant tolerating the distasteful. The founders didn't convene to sort out which speech was polite enough to permit. The question isn't whether AI-generated smut is objectionable — it is. The question is who decides what disappears from the public square, and by what authority.
