Artificial Intelligence Regulation from the Perspective of Trust

The rapid advance of artificial intelligence (AI) has significantly impacted all aspects of society. AI has rasied complex ethical and legal issues. These issues cannot be easily addressed under existing legal frameworks. In particular, the autonomous decision-making by AI often unpredictable and opaque — necessitate an appropriate level of regulatory oversight. Establishing public trust in AI is crucial for ensuring its ethical, economic, and social sustainability. Trust presupposes knowledge of, or the ability to know, the subject to be trusted. However, where a high degree of technical expertise is required or an imbalance of information exists, verification by credible public institutions may play a key role in facilitating trust. Another essential element in building trust is the availability of effective remedies in cases of harm. Accordingly, regulatory obligations such as transparency, explainability, ex ante impact assessments, and human oversight are increasingly emphasized. The EU has adopted the world’s first comprehensive regulation on AI, ｢Artificial Intelligence Act｣, which applies a risk-based approach by imposing stricter requirements on high-risk AI systems. These include risk management systems, transparency, and human oversight. Similarly, Korea's ｢Framework Act on the Promotion of AI and Establishment of Trust｣ follows a comparable structure but requires further elements for building trust. For example, while the Act includes provisions on the duty to explain AI algorithms, it lacks clarity on the required level of explanation and does not provide specific sanctions for non-compliance. To strengthen legal remedies for AI-related harm, further discussions are needed regarding the introduction of punitive damages, the activation of AI liability insurance schemes, and the lowering or reversal of the burden of proof. These measures can enhance the accountability of AI systems and contribute to building public trust in AI.