One of China’s largest and most pervasive surveillance networks got its start in a small county about seven hours north of Shanghai.
In 2013, the local government in Pingyi County began installing tens of thousands of security cameras across urban and rural areas — more than 28,500 in total by 2016. Even the smallest villages had at least six security cameras installed, according to state media.
Those cameras weren’t just monitored by police and automated facial recognition algorithms. Through special TV boxes installed in their homes, local residents could watch live security footage and press a button to summon police if they saw anything amiss. The security footage could also be viewed on smartphones.
In 2015 the Chinese government announced that a similar program would be rolled out across China, with a particular focus on remote and rural towns. It was called the “Xueliang Project,” or Sharp Eyes, a reference to a quote from communist China’s former revolutionary leader Mao Zedong who once wrote that “the people have sharp eyes” when looking out for neighbors not living up to communist values.
Sharp Eyes is one of a number of overlapping and intersecting technological surveillance projects built by the Chinese government over the last two decades. Projects like the Golden Shield Project, Safe Cities, SkyNet, Smart Cities, and now Sharp Eyes mean that there are more than 200 million public and private security cameras installed across China.
Every five years, the Chinese government releases a plan outlining what it looks to achieve in the next half-decade. China’s 2016 five-year plan set a goal for Sharp Eyes to achieve 100% coverage of China’s public spaces in 2020. Though publicly available reports don’t indicate whether the program has hit that goal — they suggest that the country has gotten very close.
China’s modern surveillance scheme started in 2003, according to Dahlia Peterson, research analyst at Georgetown University’s Center for Security and Emerging Technology, with the creation of the Golden Shield Project.
The Golden Shield Project, run by the Ministry of Public Security (MPS), is, in part, responsible for the country’s strict internet censorship. But the program also included physical surveillance. The MPS created databases that included 96% of China’s citizens, with one titled the National Basic Population Information Database. That database includes household registration information, called “hukou,” as well as information on past travels and criminal history, according to a report from the Immigration and Refugee Board of Canada.
Local population databases were also created, according to a paper published in the American Journal of Political Science. These local databases allowed for blacklists, which barred the use of public transportation. Police would be dispatched if someone who had been blacklisted tried to book a bus, train, or airline ticket.
Following Golden Shield, China launched two other surveillance projects focused on the installation of cameras. Safe Cities, launched in 2003, focused on disaster warnings, traffic management, and public security. SkyNet focused on installing cameras connected to facial recognition algorithms.
“Chinese state-run media has claimed Skynet can scan the entire Chinese population in one second with 99.8 percent accuracy, yet such claims ignore glaring technical limitations,” Peterson wrote.
Observers should take these figures with a grain of salt: Accurate and up-to-date information about China’s surveillance initiatives isn’t easily available, and what is publicly known is mainly generated by academics and journalists with some access to government officials or surveillance equipment manufacturers. It’s also unclear which cameras are exclusively viewed by village, city, and provincial governments, and which feed data back to the central government.
Just like Golden Shield, the SkyNet program still exists today, and benefits from 16 years of A.I. research, as well as the tech industry’s boom. According to the New York Times, SkyNet data is used at building complexes that use facial recognition to open security gates. The photos from those security gates are then shared with local police to build a database of the local population.
However, these surveillance schemes are mostly targeted at cities, where funding and population density makes centralized surveillance easier. Sharp Eyes, which is focused on rural areas, is meant to offload work from potentially understaffed police departments.
For instance, an article written by Chinese state media about the Sharp Eyes implementation in Pingyi notes that the county has a population of 1 million people, and only about 300 police officers.
What gets reported to police by the Sharp Eyes program isn’t just limited to crime. One Pingyi resident in the state media article spoke of reporting a collapsed manhole cover, while another mentioned that they had suspected a multilevel marketing scheme happening in a nearby building. The MLM organization was reported to the police, who allegedly broke it up with warnings and fines.
According to Peterson, the Sharp Eyes project is implemented differently depending on each city or town’s needs, but the general premise is the same: The city or town is divided into a grid, and each square of the grid acts as its own administrative unit. Citizens watch security footage from within their grid, giving a sense of ownership over their immediate surroundings. Municipal data can then be aggregated based on reports from each square on the grid.
Cities can also add new technology to the mix at their discretion. Though the system primarily relies on facial recognition and locally broadcast CCTV, the city of Harbin, for instance, published a notice that it was looking for predictive policing technology to sweep a person’s bank transaction data, location history, and social connections, as well as make a determination as to whether they were a terrorist or violent.