Meta ordered take down of AI-generated videos
Meta Oversight Board mandates removal of AI-generated deepfakes targeting a Scottish councillor and a Muslim activist while calling for stricter platform.
- Incident date
- Sep 2026
- Target
- Scottish Labour councillor and an unnamed Muslim activist
Meta’s Oversight Board has ordered the removal of two AI-generated videos from Facebook, highlighting critical deficiencies in the company’s current content moderation and deepfake safeguards. The board’s intervention follows incidents where AI was used to malign public figures and activists, prompting a call for more robust policies to address deceptive synthetic media.
What happened
The Oversight Board reviewed two distinct cases involving AI-generated content. In the first instance, a video falsely depicted a Scottish Labour councillor making inflammatory comments that linked refugees to sexual violence. The board determined the clip was AI-generated, evidenced by poor synchronization between the audio and the subject's facial movements. Meta had initially concluded the video did not breach its policies, but the board overruled this, citing violations of hateful conduct rules and arguing the content should have carried a high-risk AI label.
In the second case, an AI-generated video mocked a young Muslim woman involved in menstrual health education. The video depicted her performing absurd exercises and eating junk food while offering health advice. The board identified this as a violation of Meta’s bullying and harassment policy, noting that such manipulated imagery targeting private individuals contributes to the harassment of women in public discourse.
To address these issues, the Oversight Board issued nine policy recommendations. These include expanding the application of high-risk AI labels, implementing warning screens that require user interaction, and instructing algorithms to demote content flagged as high-risk. The board further suggested that Meta revise its definition of unwanted manipulated imagery to specifically include deepfakes that portray private individuals performing actions or speaking words they did not actually commit to. The board emphasized that current safeguards are inadequate to mitigate the growing spread and impact of synthetic media on social platforms.