Meta’s deepfake moderation isn’t good enough, says Oversight Board

The semi-independent Oversight Board for Meta has criticized the tech giant's current strategies for detecting and labeling AI-generated media, describing them as inadequate for managing the rapid spread of digital misinformation during global conflicts.

This critique follows a detailed inquiry into a manipulated video depicting property damage in Israel that circulated on Meta's networks last year. The advisory body emphasized that the platform's current reliance on users self-reporting synthetic media, combined with manual escalation processes, fails to address the challenges of today's fast-paced digital ecosystem, where deceptive materials frequently cross over from rival platforms like TikTok and X before spreading widely.

To address these security loopholes, the board recommended that Meta implement a dedicated community policy specifically for synthetic content, enhance its automated detection capabilities, and provide much clearer transparency regarding the penalties for violating these rules. Additionally, the panel urged Meta to adopt broader and more consistent implementation of the C2PA metadata standard to ensure that altered files are clearly flagged for everyday users, pointing out that even Meta's own proprietary AI-generated tools currently fail to apply these labels uniformly. Although the board's recommendations are non-binding, they mirror internal discussions among Meta executives regarding the urgent necessity of establishing reliable verification systems for authentic digital media across all social networks.