Advertisement
Science
ChinaScience

China’s internet police losing man-versus-machine duel on social media

  • Hordes of bot accounts using clever dodging tactics are causing burnout among human censors, police investigative paper finds
  • Authorities may respond by raising a counter-army of automated accounts or even an AI-driven public opinion leader

Reading Time:3 minutes
Why you can trust SCMP
14
The Chinese government and internet companies have long employed an army of human censors to monitor social media content. Photo: Shutterstock
Stephen Chen
Automated social media accounts engaging in political discussions are stretching China’s internet police to the limit, a new study has found.

These social media bots are difficult to identify because they use artificial intelligence technology to mimic human language and online behaviour. Often working in groups, they are able to generate and spread a huge amount of information within a short time.

And these accounts are overpowering state censors, because “humans get tired easily, [and] cannot endure a sustained fight,” according to Shao Lei, associate professor of digital investigation with the Sichuan Police College.
Advertisement

“As countries around the world use information technology as a means of power integration and social control, the scale of false information dissemination and propaganda is larger than ever,” lead investigator Shao and his colleagues said in a paper published in the domestic peer-reviewed Journal of Intelligence last week.

Their suggested solutions? Raising a counter-army of bot accounts, or even AI-driven public opinion leaders.

03:19

China’s delivery drivers pushed to the brink by food app algorithms

China’s delivery drivers pushed to the brink by food app algorithms
This comes as, starting early last year, China’s internet police began to detect an unusual increase in posts with negative views on the government on the country’s largest social media platforms, WeChat and Weibo.
Advertisement
Advertisement
Select Voice
Choose your listening speed
Get through articles 2x faster
1.25x
250 WPM
Slow
Average
Fast
1.25x