What happened
Following deadly tornadoes in Southwest Michigan on March 6, AI-generated images and videos depicting the disaster rapidly went viral online, often without disclosure. One Facebook page posting such AI visuals gained over 1 million followers, offering a $1 subscription. On March 10, Meta's Oversight Board called for new rules for AI-generated content depicting real-world events, citing the Israel-Iran conflict, and timely responses to deceptive AI output. Meta currently requires users to employ an AI disclosure tool for photorealistic video or realistic audio.
Why it matters
Undisclosed AI-generated content during a crisis undermines public access to accurate information, creating confusion and panic. This mechanism allows bad actors to monetise misinformation, as one account gained over 1 million followers through deceptive posts. For content moderation teams and platform architects, this necessitates immediate implementation of effective detection and mandatory disclosure systems for AI-generated media, particularly concerning real-world events. The Oversight Board's intervention, following its March 10 rebuke of Meta's AI handling, highlights the need for platforms to enforce clear policies and tools.
