Meta's Oversight Board Calls for Deepfake Removal and Voting Interference Protections

Meta's Oversight Board has issued nine policy recommendations to address deepfake content on its platforms. The recommendations aim to remove deepfakes, protect against voting interference, and create new content moderation frameworks for generative AI.

The Board made the recommendations after two high-profile cases highlighted gaps in Meta's existing moderation playbook. One case involved a sexualized deepfake video targeting a non-public figure on Instagram, which was initially deemed acceptable but later removed by the Board.

The Board reclassified AI-generated impersonations as inherently non-consensual, requiring their removal without needing victims to come forward. This decision acknowledges the severe reputational and societal harms caused by synthetic impersonation content, particularly for women and children who are disproportionately targeted.

Another case involved deceptive AI content used during the Israel-Iran conflict in 2025, where synthetic media was fabricated to manipulate narratives. The Board called for a dedicated Community Standard specifically for AI-generated content, separate from the legacy misinformation policy.

The recommendations also include implementing 'High Risk' labels for potentially AI-generated content and embedding provenance metadata into outputs created by Meta's own AI tools. Furthermore, voting interference protections are to be extended globally with consistent standards applied everywhere Meta operates.