China battles AI fake disaster videos as Typhoon Noul hits

Typhoon Noul made landfall in China over the weekend, and officials were forced to confront a second emergency online: AI-made disaster videos spreading fast across social media. The false clips included fabricated flood footage and images of bodies floating in floodwaters, adding confusion to a real weather disaster.

Stephen McDonell said the problem is especially dangerous when storms and floods are unfolding. “During natural disasters, like this week in #China, ‘AI’-generated misinformation is particularly dangerous and officials here are battling to control it,” McDonell wrote on X. His warning pointed to the same pressure emergency managers face in real time: separating authentic warnings from synthetic content before panic spreads.

The BBC also showed Chinese state media on CCTV airing a screenshot of AI content that had been marked as AI-generated, a sign that broadcasters and authorities are already trying to label synthetic material before it takes hold. That effort matters because the misleading videos are not just a social media nuisance. They can divert attention from evacuation orders, complicate emergency response, and erode confidence in real warnings when people are deciding whether to move, shelter or wait.

Other coverage described China as diving headlong into AI technology even as it confronts the downside of synthetic media during disasters. The same tools that can create convincing images and video at speed are now being used to manufacture flood scenes, dead bodies and other emotionally charged material that spreads rapidly when communities are already on edge. In a country facing more severe weather, that combination has become a new challenge for disaster response.

The issue reaches beyond China. As climate-driven disasters intensify, U.S. emergency agencies and social platforms are likely to face the same test: how to verify and amplify real warnings fast enough to keep fake clips from overwhelming them. The China case shows that the public-trust problem is no longer theoretical. When synthetic footage moves as quickly as a storm, the information battle becomes part of the disaster itself.