In brief
Recent wild storms and imagined catastrophes have taken the internet by storm in China. The BBC’s investigative team looks at how AI‑generated videos can be mistaken for reality, leading to panic and policy responses. By uncovering the signs of deep‑fakes, they show that even sharp‑eyed viewers can spot fakes with the right tools. The piece highlights government attempts to curb misinformation and gives readers a quick guide to verify online video claims before sharing. Finally, it invites young readers to test their skills in a fun, interactive fact‑check challenge.
Zoom‑in clips of continents flooding, planes crashing, or earthquakes ending up on Chinese social media have shocked users in the past weeks. Most shiver‑worthy footage turns out to be real, but a growing number of videos are trick videos—all created by AI and passed off as factual. The surge in fake videos has even confused emergency services and unintentionally led to panic across the nation.
The BBC’s China correspondent, Stephen McDonell, dives into the science behind these mis‑claims. He explains that generative neural‑net models can now stitch together shaky video footage, animate weather events, and add realistic effects that fools most viewers. Even old‑flicks of crashes are being regenerated to look brand‑new and, paradoxically, even “original.”
When government agencies notice a sudden spike in such videos, they launch captcha tests and public advisories. China’s Ministry of Public Security announced a data‑sharing task force last month, demanding that distributors submit metadata to help trace the source of fake video launchpads. Yet McDonell points out that the technology is advancing so fast that data‑protection gaps keep widening.
How can we spot a bot’s bluff? The experts suggest four quick checks: 1) look for odd background colors, 2) watch for lighting glitches, 3) verify timestamps via metadata, and 4) cross‑check with trusted news outlets or the official meteorology portal. “Short videos played on TikTok often lack the breadcrumb trail you see on the BBC or ChinaGov cloud,” says McDonell.
Below is a quick challenge to test whether you can distinguish real from AI: scan the clip for a hidden watermark, search for the original HD source, and cross‑check the headline from a trusted news site. Share your verdict in the comments and let’s keep the conversation real.
