Oversight Board criticizes Meta's handling of AI deepfakes
Meta's independent Oversight Board has ordered the company to remove two AI-generated videos from Facebook.
One video showed a Scottish Labour councillor falsely appearing to make inflammatory comments about refugees. The other mocked a Muslim campaign volunteer promoting menstrual health education, showing her exercising absurdly and eating junk food.
According to the board, both videos were AI-generated. In the politician's case, the board noted the audio was not fully synchronized with the facial movements. Videos targeting the volunteer reportedly received tens of millions of views online.
Meta had initially left the politician's video up, saying it had not been flagged by any of its trusted partner organizations and it was not labeled as AI-generated because the poster hadn't disclosed this.
The board described Meta's safeguards as "consistently and fundamentally inadequate" given the rapid rise of AI deepfakes. Board Co-Chair Pamela San Martin said such deepfakes are increasingly used to harass and silence women from public discourse.
The board recommended nine policy changes, including stronger labeling for high-risk AI content, reducing such content's visibility in feeds, and harsher penalties for repeat violations. According to the material, Meta has 60 days to respond to these recommendations.
WHY IT MATTERS
This case shows that even a major platform like Meta still lacks reliable rules to protect people from targeted AI-generated fake videos.
WHAT YOU CAN DO NOW
- Be skeptical of shocking video content If a video shows extreme or surprising statements, look closer. If the mouth movements and audio don't fully match, that can be a sign of a fake.
- Report suspicious videos to the platform Use the report function if you think a video is an AI fake. Briefly explain why you believe it's fake so it gets reviewed faster.
- Don't rely only on AI labels Labels marking AI content are often missing even when a video is clearly fake. Check if trusted news sources are reporting on the content too.
- Don't share unclear content right away Before forwarding a disturbing video, look for the original source or confirmation from known media outlets. This helps stop a fake from spreading further.