South Korea plans to launch a test program to track those infected by Covid-19 and everyone they may have come into contact with. The initiative will use AI facial recognition and the nation’s network of CCTV cameras.
Critics have raised concerns over privacy.
The government will launch the program in Bucheon, one of South Korea’s most densely populated urban areas. According to Reuters, it will go online in January 2022.
According to a business plan from the city, the program uses an AI algorithm to track the infected, those they come into contact with, and whether or not they’re wearing a mask. The program will only use the city’s 10,000+ CCTV cameras for now.
South Korea isn’t alone in its bid to use facial recognition to combat Covid-19. Multiple US states have experimented with the tech, along with countries like Russia, China, and India.
Australia rolled out a facial recognition quarantine app too. However, that uses more limited tracking methods.
What is different is that this move follows criticism thrown at South Korea for its collaboration with AI services. Reporters discovered that it has been giving millions of photos to facial recognition companies.
Recognizing a Pattern
This plan has just increased criticism of the government attempting to institute “Orwellian” measures. While this system is useful for combating pandemics, critics fear the country could use it for authoritarian purposes.
Governments could use it to track political dissidents, activists, opposition politicians, and journalists. It would be an all-in-one combination of people search sites and background checks services, but with unlimited scale.
It isn’t only South Korea that could do this. The program is creating a blueprint oppressive states can draw on. This is compounded by companies like Clearview AI that are developing stronger tech and selling it to governments.
For now, it appears the plan will go ahead, as it’s been cleared by all the relevant authorities. Depending on its performance, the government will decide whether to expand it.