AINEWS
Typhoon Noul came ashore with fake disaster videos riding alongside it. A year after China's 23 July 2025 crackdown on fabricated disaster content, the case shows what emergency communication looks like when generative tools let anyone fake a flood.
Typhoon Noul came ashore in southern China over the weekend. The fake disaster videos arrived with it. Police were logging fake clips of power outages that had never happened, fabricated flood scenes in counties that were still dry, and a synthetic cut-and-paste of crocodiles being released into a river, riding on a real snake escape from a breeding farm in Hengzhou, Guangxi, earlier in July.
A real climate-driven emergency now lands on top of an evidence layer that anyone with a phone and a prompt can contaminate. The state response has reached first for arrests, detention pending criminal prosecution, and fines.
False power-outage clips triggered panic buying of emergency supplies. Fabricated flood scenes in unaffected districts confused evacuation messaging. Zhejiang Police Deputy Chief Huang Zhihua told the BBC the motive is engagement, not politics.
Prof Chen Bin, writing in the Central Party School newspaper, calls it a one-prompt problem. Text, image, audio, and video can now be made at scale by non-state actors with no production background. Prof Xu Xiaoke of Beijing Normal University puts the technical arms race plainly: detection is always behind generation. It is a cat-and-mouse game, and the mouse gets faster every quarter.
On 23 July 2025, China's Cyberspace Administration announced a nationwide crackdown on fake internet content tied to floods and disasters, ordering platforms to strengthen review and singling out fabricated data, malicious editing, staged scenes, misattributed historical events, and officials impersonation. Penalties already on the books range from fines to detention, with prison sentences of up to seven years where the misinformation fallout is serious enough.
The crackdown is not a clean solution. The harder problem is the trust vacuum it is trying to fill. China's official information layer has its own credibility gaps. The Liulan Dam near Nanning, where at least 26 people died downstream, was described in official statements as "opening" or having a "gap" rather than breached. The 2021 Henan floods concealed 139 deaths for weeks. Prof Gao Fuping of East China University of Political Science and Law argues for public education and visible enforcement, with real cases publicized so people learn to verify before they forward. Shanghai and Washington-based lawyer Peter Pang points the same direction at the platforms, calling them the enablers and arguing that legal accountability should sit with them, not only with creators.
In November 2025, Nikkei Asia reported China tightening rules after a viral AI-fabricated photo of a baby who "survived" an earthquake. The OECD AI Incident Monitor logs Gansu and Henan cases with police warnings, content deletions, and administrative penalties. Prof Xu's forward-risk list reads like a forecast: political elections, public health emergencies, armed conflicts, natural disasters, livelihood issues.
China is the first country-scale test of what happens when cheap generative tools meet a population that already lives through climate-driven disasters every season. The next typhoon will land somewhere else. The fakes will land first.