The government will begin full-scale use of cutting-edge artificial intelligence (AI) detection technology to respond firmly to deepfake video and audio manipulation crimes ahead of the June local elections.
Minister of the Interior and Safety Yoon Ho-jung held a demonstration of the "AI deepfake detection model" at Government Complex Seoul on the 10th and stated accordingly.
According to the National Election Commission, requests to delete deepfake videos during election periods rose sharply from 388 during the 22nd general election in 2024 to 10,510 in the presidential election in 2025.
The detection model being introduced this time was built based on the results of the "AI detection model competition for responding to deepfake crimes" held in Dec. 2025. A total of 268 teams (1,077 people) participated in the competition at the time. The Ministry of the Interior and Safety (MOIS) and the National Forensic Service will provide the five final selected top-performing models to the National Election Commission and operate a support system to quickly examine suspicious content during the election period.
An official at the Ministry of the Interior and Safety (MOIS) said, "This model improves detection accuracy by conducting 'global analysis,' which analyzes the overall flow of the video, and 'local analysis,' which precisely determines manipulation traces in specific parts such as faces, at the same time," adding, "Actual verification results showed a high detection accuracy of about 92%."
The Ministry of the Interior and Safety (MOIS) and the National Forensic Service plan to gradually expand the scope of use of the AI deepfake analysis model going forward. They will work closely with related ministries and agencies, including the Ministry of Gender Equality and Family, the Korea Communications Commission, and the Korean National Police Agency, to build a dense, pan-government response system to digital crimes.
Minister of the Interior and Safety Yoon Ho-jung said, "Fake news using AI and deepfakes created with advanced technology is seriously threatening electoral democracy," adding, "The government will steadily strengthen its capacity to respond to false and manipulated information during the election process so that the precious rights of the people are not infringed, and will build a competent and trusted 'AI democratic government.'"