부패방지와 AI: 새로운 부패 위험과 법적 대응

Abstract Artificial Intelligence (AI) is a key technology that can enhance administrative efficiency and accuracy through the analysis and prediction of large-scale data. It is also attracting growing attention as a new policy tool in areas such as public procurement, fiscal execution, auditing, supervision, and investigation support for corruption prevention. Unlike conventional anti-corruption systems, which have relied primarily on post hoc detection and sanctions, AI enables a paradigm shift toward corruption prevention by proactively identifying abnormal transaction patterns and risk indicators. However, the adoption of AI does not in itself guarantee the effectiveness or legitimacy of anti-corruption measures; rather, it may generate new types of corruption risks that have not been adequately anticipated by existing legal frameworks. Against this backdrop, this paper analyzes the dual nature of the relationship between AI and anti-corruption and explores legal responses suitable for the AI era. First, it examines how AI can be utilized within anti-corruption systems and assesses its policy potential for strengthening early warning mechanisms, risk-based supervision, and transparency. Second, focusing on the capacity of AI to generate new forms of corruption risk, the paper analyzes structural threats inherent in the AI era, including algorithmic corruption, the concentration and abuse of surveillance power, cybersecurity vulnerabilities, and the manipulation or fabrication of evidence. In particular, it highlights the risk that opaque algorithms and automated decision-making structures may obscure accountability and function as tools for concealing or sophisticatedly reproducing corruption. Furthermore, this paper proposes three core pillars for legal and institutional responses to these risks: (1) ensuring the transparency and explainability of AI; (2) strengthening systemic coherence between anti-corruption legislation and AI-related laws; and (3) establishing multi-agency and multi-sectoral cooperative governance structures. It emphasizes that explainable artificial intelligence (XAI) is not merely a technical feature, but a fundamental requirement for reintegrating accountability and due process principles into AI-based anti-corruption systems. In addition, the paper underscores the need to clarify responsibility, remedies, and control mechanisms associated with AI deployment through the coordinated application of the Anti-Corruption Act, the Personal Information Protection Act, the Framework Act on Administrative Law, and the Framework Act on Artificial Intelligence. It further argues that the legitimacy and credibility of AI-based anti-corruption measures cannot be secured without multi-layered governance that integrates central and local governments, public and private actors, and administrative institutions and civil society. In conclusion, this paper contends that AI should not be regarded merely as a technical instrument for combating corruption, but as a normative force capable of reshaping structures of power and control. Anti-corruption in the AI era depends not on whether AI is adopted, but on the legal and institutional conditions under which it is governed and deployed. Future anti-corruption legislation should therefore be restructured around constitutional safeguards and institutional design that presuppose ongoing technological advancement.