인공지능 구축 및 이용에 있어 책임법리

The civil principles of liability concerning the provision and use of Artificial Intelligence (AI) need to be reconstructed based on a precise technical understanding of AI, as an extension of the existing legal framework. Although some argue that AI's autonomy, unpredictability, and unexplainability make liability determinations difficult, these characteristics merely stem from the technical essence of AI—processing vast amounts of information statistically. Such traits are not of a nature that calls for the complete exclusion of existing criteria for determining negligence. Since AI is a value-neutral tool without its own purpose and inherently possesses “error-proneness,” developpers, providers, and users of AI should bear the duty for proper verification and management based on the premise of this inherent fallibility. In this regard, the traditional theory of fault-based liability remains applicable. On the other hand, regulatory approaches that target AI technology itself based on its intended use, such as the EU AI Act, conflict with the technical value-neutrality of AI and suffer from low regulatory effectiveness. It is necessary to evaluate liability for AI-driven actions and their outcomes within the existing framework of civil law, grounded in an accurate understanding of AI technology. Excessive regulation of AI technology not only hinders industrial development but also risks distorting the distribution of liability under civil law.