AI 정부 구현을 위한 행정데이터 활용의 법적 쟁점

The advancement of digital transformation and artificial intelligence (AI) technology demands fundamental changes in government operations. Korea has evolved from early computerization through e-government to data-based digital government, and is now pursuing a transition to 'AI Government' that aims to provide predictive and personalized services based on AI. The government has declared 'World's Best AI Democratic Government' as a national agenda and is working to establish an institutional foundation, including efforts to amend the 「Act on Promotion of Data-Based Administration」 to the 「Act on Promotion of Artificial Intelligence and Data-Based Administration」. The core driver of AI government is data. Administrative data systematically accumulated by the government in the course of performing its duties is an irreplaceable core asset for AI learning and policy decision-making. However, the current legal system does not sufficiently support the smooth utilization of administrative data essential for implementing AI government. This paper aims to analyze legal issues surrounding the utilization of administrative data, a core resource for AI government implementation, and to propose legislative improvements. To this end, it first examines the concept of AI government and the significance of administrative data, and systematically reviews the limitations of the current legal system from a data lifecycle perspective. The main legal issues are as follows. First, the absence of data governance. In Korea, data-related functions are dispersed across multiple ministries, lacking strong and consistent leadership, and showing limitations in securing data standardization and interoperability. Second, the persistence of positive disclosure methods. While formally declaring a negative approach, many individual laws still maintain a positive method that restrictively enumerates provision targets. Third, the data silo problem. The current data registration system and public data sharing system constrain actual data sharing due to their voluntary and approval-centered structure. Fourth, conflicts with personal information protection. Tensions arise between protecting information subjects' rights and data utilization in the process of using personal information included in administrative data for purposes other than originally intended and outsourcing to the private sector. To address these issues, the following legislative policy improvements are necessary. First, to establish a substantial data governance system, the authority of the Chief Data Officer (CDO) should be strengthened and a control tower capable of coordinating inter-ministerial data sharing should be established. Second, the positive enumeration method remaining in individual laws should be converted to a negative disclosure system, and priority application provisions of general laws such as the Data-Based Administration Act should be clarified. Third, data registration should be mandated and the approval-centered sharing system should be redesigned into a structure of principle-based sharing and exceptional refusal. Fourth, requirements for using personal information for purposes other than originally intended should be clarified, substantial management and supervision systems should be strengthened when outsourcing to the private sector, and complementary safety measures such as pseudonymization and anonymization should be prepared. AI government signifies not merely technology adoption but a fundamental transformation of the administrative paradigm. Without a legal and institutional foundation enabling free flow and convergence of data, substantial implementation of AI government is impossible. It is hoped that the legal issue analysis and improvement measures presented in this paper will contribute to establishing legislative policies for AI government implementation.