Officials said the updated posters have occasionally prompted new public reports and tips. - YouTube
SEOUL: Over 90 per cent of missing people in South Korea are found within a year, but 1,050 individuals who disappeared as children remain unaccounted for as at 2025.
Advances in artificial intelligence (AI) now allow authorities to generate realistic images of what these children may look like today, enabling wider distribution of posters featuring their presumed adult appearance.
The state-run National Center for the Rights of the Child (NCRC) has produced AI-generated posters for 60 long-term missing children, using technology developed by the Korea Advanced Institute of Science and Technology (KAIST).
According to KAIST, the programme analyses typical age-progression patterns and applies them to the child’s last known photo to produce an image of their likely adult face.
One example is Kim I-gon, born in 1972 and missing since 1985. Using a photo of him at age 13, the system generated an image of what he might look like at 52, with a squared jaw and visible wrinkles.
The NCRC’s project is jointly carried out with the National Police Agency and the Ministry of Health and Welfare, and officials say the updated posters have occasionally prompted new public reports and tips.
South Korea is not alone in deploying such technology.
Similar efforts are underway in Argentina, which recently used AI to re-create the adult faces of children who disappeared during the military dictatorship four decades ago.
KAIST developed South Korea’s age-progression system in 2015, but officials say it has taken a “major leap” since the addition of super-resolution imaging.
The NCRC has also used AI for other long-term missing-child campaigns, including its “Runway to Home” project announced in October.
The initiative re-creates video of the missing person walking on a virtual runway – both as a child and as an adult – to intuitively visualise how their appearance may have changed and spark renewed public interest. - The Korea Herald/ANN
