Typhoon Dolphin made landfall in China on Sunday, held together better than meteorologists were expecting, and pushed inland, bringing heavy rains and flooding to central China and raising severe weather alerts in the national capital of Beijing.
As the storm advanced, Chinese Communist Party censors worked furiously to delete “flood-related rumors” that were alleged generated with artificial intelligence (A.I.).
Typhoon Dolphin is the strongest tropical cyclone to strike China this year, with rains and wind well above the standards for a Category 1 hurricane. Heavy rain has been a particular concern, as the storm made landfall in the eastern Zhejiang province and worked its way into flood-prone central China.
The initial reaction from Zhejiang residents to the storm’s arrival was muted, as they felt it was not especially large or strong by the standards of the region.
“To us Guangdong people, this typhoon is as ordinary as having a meal and feeling the wind. It’s just part of everyday life,” a tourist stranded at Shanghai’s airport by the storm told Reuters on Sunday.
Opinions of the storm shifted on Tuesday as Typhoon Dolphin demonstrated unusual endurance after landfall, holding together as a massive storm system and dumping heavy rain across the central province of Hubei.
Over 500 miles away from the edge of the storm, four of the outlying districts of Beijing activated emergency flood responses and warned of incoming torrential rain on Tuesday. The heavily populated and industrialized Hubei province sounded a “yellow alert” for landslides and flash flooding. Highways and drainage systems struggled to cope with rising water levels from a projected ten inches of rain in a single day.
On Tuesday morning, provincial officials evacuated almost ten thousand people because the water levels in dozens of reservoirs had surged past flood control limits.
Henan, the province to Hubei’s northeast, issued a red alert for flash floods as Typhoon Dolphin began to compress against elevated terrain, causing its rainfall to intensify even further. Fortunately, the typhoon finally began to fall apart as the last heavy rains were squeezed from its edges, and it was downgraded to a tropical depression on Tuesday afternoon.
Hubei province was hit by another typhoon called Maysak last month and that one spawned a pair of rare tornadoes, causing at least 11 fatalities.
China’s state-run Global Times reported on Monday that the Ministry of Public Security filed 15 cases of “netizens” allegedly “fabricating misinformation using A.I. to garner attention.”
“For instance, a 37-year-old male netizen based in Shenzhen, South China’s Guangdong Province, doctored local official typhoon safety notices using AI, posting the manipulated versions on social media to garner views. The false posts drew widespread public discussion and impacted disaster relief operations,” the report said.
Another example was a man who allegedly created false notices for school and business closures and then spread them on WeChat. Some of the suspects allegedly used “deceptive editing techniques” to create fake news stories about people struggling to survive high winds and heavy floods.
“These netizens have been issued administrative penalties by local public security authorities according to law,” the Global Times somberly declared.
The BBC reported last month that China’s online pranksters have grown highly adept at using AI to produce “dangerously misleading videos” during natural disasters, especially floods.
Some of these videos are exaggerations of real news stories, like the prankster who used AI to add crocodiles to a swarm of snakes escaping from a flooded snake farm, while others are purely fabricated images of disturbing spectacles like corpses floating in flood waters.
“Many of these content creators use disaster news to try to gain followers, knowing that more sensational or exaggerated content will garner more likes and more views. They are deliberately fabricating information to grab people’s attention,” an irate online public security officer from China said.
The BBC noted that Chinese officials and state media have a habit of “downplaying disasters,” sometimes with dangerous consequences, so the Chinese Internet exists in a “trust vacuum” where users equally consume photos and videos that seem to reveal truths the government would rather suppress.
In early July, the government deployed propaganda to make a breach in a major dam appear less alarming — and at least 26 people were killed when the dam burst. This incident appears to have made the public even more likely to believe the Chinese state is lying to them about flooding.
The Cyberspace Administration of China (CAC) launched a nationwide crackdown in late July, with an emphasis on fake news about floods. CAC said it was troubled about AI technology making it so easy for pranksters to create realistic-looking videos and strongly hinted that platform providers would be held responsible for failing to police their users.