As Hurricane Melissa, a formidable Category 5 storm, approached Jamaica, social media platforms were inundated with AI-generated videos, diverting attention from essential safety information. AFP identified numerous fabricated clips, many bearing the watermark of OpenAI’s text-to-video model, Sora. These videos depicted exaggerated scenarios, from severe flooding and sharks in the water to human suffering and locals downplaying the storm’s threat, often reinforced by stereotypical Jamaican accents. Jamaican Information Minister Senator Dana Morris Dixon emphasized the importance of relying on official channels for accurate information, as fake content could lead to underestimating the storm’s severity. Experts, including University of Oklahoma meteorology professor Amy McGovern, warned that such misinformation could result in catastrophic consequences, including loss of life and property. The clips primarily spread on TikTok, with some lacking the required AI disclosure labels. TikTok removed over two dozen videos and related accounts after AFP flagged them, though similar content circulated on Facebook and Instagram. Cybersecurity expert Hany Farid highlighted how advanced text-to-video models have accelerated the spread of convincing fakes, contributing to public misinformation. Despite watermarks, many viewers appeared unaware the content was AI-generated, with some even engaging with the fabricated scenarios. Farid noted the paradox of the information age, where increased information leads to a less informed public.