AI-generated misinformation during emergencies

Generative artificial intelligence is increasingly being used to create hyper-realistic synthetic media, such as images and videos, that spread rapidly on social media platforms. This misinformation often surfaces during extreme weather events and natural disasters, spreading faster than official meteorological or emergency updates and causing public panic.

In specific instances in India, authorities have addressed various AI-generated hoaxes, including a fabricated video of a tiger in Odisha and false visuals of helicopter rescue operations during floods in Assam. These incidents have prompted legal warnings and increased efforts by fact-checking units to use forensic analysis to identify fabricated content.

The challenge has expanded to North America, where Ontario is experiencing a surge of AI-generated and manual misinformation during a severe wildfire season. This includes fabricated maps of non-existent communities, false air quality data, incorrect evacuation alerts, and graphics directing users to non-existent government websites. The proliferation of ‘AI slop’—low-quality, AI-generated content—alongside conspiracy theories regarding arson and ‘15-minute cities’ has complicated public response efforts.

Recent developments in Ontario show that severe wildfire activity has consumed over 728,000 hectares, displacing thousands of people, including members of the Namaygoosisagagun and Whitewater Lake First Nations. The resulting smoke plumes created a transboundary health issue, affecting approximately 106 million people across North America, including major U.S. cities. Simultaneously, the rise of generative AI has fueled a surge in misinformation, with malicious actors and automated bots using hyper-realistic fake images, cloned audio warnings, and fabricated maps to spread panic or conspiracy theories. This digital distortion complicates emergency management and erodes trust in official communications.

This trend reflects broader difficulties in information literacy; Statistics Canada reported in May that 43% of Canadians find it more difficult to differentiate between facts and misinformation online compared to three years ago, a shift coinciding with the increasing ubiquity of AI-generated content.