The government will expand the development environment and institutional support so that public officials can directly solve problems in the administrative field using artificial intelligence (AI). First, it will increase the simultaneous user capacity of the AI development platform for public officials, "AI Government Lab," to 6,000 next year.
The Ministry of the Interior and Safety reported the "plan to promote and expand direct development using AI by public officials" at a Cabinet meeting on the 8th. As a follow-up to the "strategy to realize the world's best AI democratic government" released on the 25th of last month, the core is to support public officials in directly using AI to solve problems in the field.
The "AI Government Lab," a development and testing space for public officials, will expand from an internet network focus to the work network. It will provide resources that are difficult for individuals to secure, such as AI coding tools, development environments, and graphics processing units (GPUs), and increase the number of simultaneous users from the current level of 200 to 6,000 by 2027.
Public data and administrative services will also be provided in standardized formats such as application programming interfaces (APIs) and the Model Context Protocol (MCP), enabling public officials to use the necessary data and functions for development.
According to the Ministry of the Interior and Safety (MOIS), participation by public officials has steadily increased since the pilot operation of the AI Government Lab began on July 3. As of the 3rd, 2,073 AI-developing public officials had signed up for the public GitLab, and 770 projects had been registered. Of these, 376 were made public so other public officials could use them.
However, the development environment, expense, and security were cited as obstacles. A survey of 244 attendees at the "AI Government Lab" briefing found that 68% named insufficient subscriptions for the development environment and AI coding tools as the biggest barrier. The view that applying security guidelines is difficult followed at 59%, and 40% said data linkage is difficult.
AI outputs created directly by public officials will be managed in three stages: "testing and verification → operations management → informatization project."
In the testing and verification stage, development can begin with minimal internal checks, without prior deliberation or separate project plan approval. When applying it to actual work later, the relevant department and agency will check accuracy, security, and personal information protection, and take responsibility for operations.
Rewards for achievements will also be prepared. Currently, in performance reviews for early promotion to Grade 5 in the advanced science and technology talent category, "AI Champion Black" certified high-skilled personnel are given extra points, and going forward, outstanding performers will receive special performance bonuses, training and learning opportunities, and additional learning expenses.
The Ministry of the Interior and Safety (MOIS) will establish the "guidelines for direct development using AI by public officials and management of outputs" in the second half of this year and set safety inspection standards for software vulnerabilities and personal information. In 2027, it plans to build the work-network AI Government Lab and a security inspection system.
Minister Yoon Ho-jung of the Ministry of the Interior and Safety (MOIS) said, "Public officials' ideas will be able to be tested quickly and freely, and when they are applied to actual work and public services, the organization will be responsible for security, quality, and operations, and we will expand verified outcomes so all agencies can use them together."