Meta advisors are calling on the company to improve how it handles misleading videos created using artificial intelligence across its platforms.
The warning comes from the company’s Meta Oversight Board, which criticized Meta for releasing an AI-generated video that falsely showed extensive damage had occurred in the Israeli city of Haifa during the conflict with Iran. The video went viral on Facebook and received a lot of attention before being reviewed by moderators.
The board said Meta should update its policies to more aggressively label AI-generated content, especially during armed conflicts, when misinformation can spread quickly. Members warned that the proliferation of synthetic videos online is making it difficult for the public to distinguish between real footage and fabricated content.
Meta launched an oversight board in 2020 to review key content moderation decisions involving platforms such as Facebook, Instagram, and WhatsApp. Although the group provides recommendations, they often disagree with the company’s response to sensitive cases.
The board said Meta’s current system relies heavily on revealing when users generate posts with AI tools. If not, companies typically wait for complaints before deciding whether to add the label.
The board said this approach is not robust enough to manage the growing amount of media generated by AI, especially during a global crisis when misleading content spreads rapidly and attracts high engagement.
The incident began last year when a Facebook account calling itself a news source posted a fake video. Despite several user complaints, Meta initially determined that the clip did not pose a direct risk of physical harm and therefore did not need to be removed or labeled.
The watchdog disagreed, saying the video should have been given a high-risk AI label so viewers knew it wasn’t real. Meta said it would apply the labels to videos within seven days.
Related books:
