Meta's AI Deepfake Policies Deemed Inadequate by Oversight Board

In two recent rulings, the Oversight Board concluded that Meta's guidelines for AI-generated content are fundamentally insufficient. One case involved a realistic deepfake of a Scottish politician making hateful statements about refugees and women; Meta kept it online because it was not flagged by trusted partners and lacked an AI label. The board said the video breached Meta's hateful conduct policy and required clear AI disclosure.

The second case featured a fabricated video of a volunteer promoting menstrual health, altered to mock her and spread widely; Meta initially closed the report without labeling it as AI-generated and only removed it after board intervention for bullying violations. The board warned that unlabeled deepfakes are increasingly used to harass, especially women, and urged Meta to prioritize likely AI content for review and improve labeling. Meta has 60 days to respond to the board's recommendations.