China confronts fake AI disaster videos

Chinese authorities are confronting a surge of fake AI-generated disaster videos as Typhoon Noul and recent floods strain emergency response systems. Misleading clips have shown bodies in floodwater, severe flooding in unaffected areas, false power outages that prompted panic buying, and fabricated scenes tied to real incidents, including escaped snakes in Hengzhou city and fake crocodile releases.

Police and regulators have responded with arrests, fines and detention, while China’s Cyberspace Administration has announced a crackdown on fabricated flood and disaster content. Punishment could include a fine or up to seven years in prison if misinformation causes serious consequences. Experts warn that AI tools have made convincing false text, images, audio and video easy to create at scale.

China’s AI push adds complexity to the problem. In 2017, Beijing described AI as the main driving force of national progress, and Xinhua says the country has over 6,000 AI companies. Specialists say labels and detection tools are not enough because generation technology is outpacing verification.

Public trust remains fragile after cases in which official sources downplayed disasters, including a Liulan Dam breach that led to at least 26 deaths downstream and 2021 allegations that Henan officials concealed or underreported 139 deaths. Police are urging residents to rely on accredited sources, while legal experts debate whether platforms and AI app makers should share liability for harmful fake content.