Meta’s oversight board has instructed the company to remove two AI-generated deepfake videos from Facebook that depict a UK politician and a Muslim campaign volunteer, citing inadequate safeguards against the spread of manipulated media on its platforms.

The board, an independent entity established by Meta in 2020 to review content moderation decisions, ruled that a video falsely showing a Labour Party councillor from Scotland making inflammatory remarks about refugees violated Meta’s policies on hateful conduct. In the video, the councillor is shown saying, “Refugees are welcome here, even if they rape our women, because white people do that too,” a claim the board found to be fabricated and harmful. The video’s audio was not fully synchronized with the councillor’s facial movements, indicating it was AI-generated. Despite being notified, Meta initially decided the video did not breach its policies and refused to label it as AI content.

The oversight board disagreed, stating the video should have been removed because it portrayed refugees as a group engaging in criminal and predatory sexual behavior, which violates the platform’s hate speech guidelines. The board also recommended that the video be accompanied by a “high risk AI” label to inform users of its manipulated nature. The decision reflects growing concerns about the rapid emergence of AI deepfakes used to spread misinformation and target individuals.

In another ruling, the board ordered Meta to remove an AI-generated video of a young Muslim woman involved in a campaign to improve menstrual health education. The video, which depicted her performing absurd actions such as eating junk food while offering health advice, garnered tens of millions of views across online platforms, including Facebook. The board highlighted the role of AI deepfakes in harassing and silencing women, particularly those engaging in public discourse.

Pamela San Martin, co-chair of the oversight board, emphasized that existing safeguards had been “consistently and fundamentally inadequate” in tackling the proliferation of AI-generated deceptive content. The board’s binding rulings require Meta to comply with content removals and take further steps to address the challenges posed by AI-manipulated imagery.

Meta has been contacted for comment on the board’s decisions.