A Study on the Composite Liability Structure for Level 4 Autonomous Vehicle Accidents - Focusing on the Compensation-First, Recourse-Later Model and Accident Liability Guidelines
Korean Insurance Law Association · 30.06.2026 · 20 (2) · 149-195
DOI: 10.36248/kdps.2026.20.2.149 ↗Kurz gefasst
Leitlinie in der Zeitschrift Korean Insurance Law Association, online erschienen am 30.06.2026. Verfasst von Young Kook Kim.
Abstract (Originalsprache)
The commercialization of Level 4 automated vehicles requires a fundamental reassessment of the traditional fault-based liability system centered on human drivers. Unlike conventional vehicles, Level 4 automated driving operates within a designated Operational Design Domain (ODD), where the Automated Driving System (ADS) performs the entire Dynamic Driving Task (DDT) and fallback functions without requiring continuous human supervision or intervention. As a result, the causes of accidents have become increasingly diversified, encompassing defects in automated driving systems, failures to implement software updates, remote operation errors, communication and mapping infrastructure malfunctions, and cybersecurity vulnerabilities. Consequently, accident liability must be restructured among multiple actors, including manufacturers, ADS developers, service operators, vehicle holders, and infrastructure managers. This study examines a hybrid liability framework in which operator liability under the Automobile Accident Compensation Assurance Act and manufacturer liability under product liability law function in a complementary manner in Level 4 automated driving accidents. Furthermore, it proposes a “first-compensation, then-recourse” model as a procedural mechanism for implementing such a framework. While a product liability-centered model has the advantage of internalizing accident costs to those who create and control technological risks, it suffers from limitations such as the victim’s burden of proof and delays in compensation. Conversely, an operator liability-centered model facilitates prompt victim compensation through automobile insurance, but may fail to ensure that ultimate liability is allocated to the party responsible for system defects or infrastructure failures. Accordingly, this study proposes a hybrid liability structure under which victims receive prompt compensation through the Automobile Accident Compensation Assurance Act and the insurance system, while ultimate liability is allocated on the basis of objective evidence obtained through accident investigations. Particular attention is paid to the distinction between accident causation and liability allocation. Accident investigation should be understood as a process of establishing technical facts through the analysis of ADS operational status, sensor perception data, communication records, and accident data. In contrast, liability allocation constitutes a separate legal and policy determination that considers factors such as risk creation, risk control capability, accident prevention capacity, and the degree of contribution to the accident. To enhance the effectiveness of the first-compensation, then-recourse model, this study explores a scientific accident investigation framework utilizing DSSAD data, vehicle logs, OTA update records, remote operation records, communication data, and accident reconstruction technologies based on 3D scanning and digital twins. It further proposes liability-sharing criteria based on risk creation, risk control capability, accident prevention capacity, and contributory responsibility, together with cost-internalization mechanisms incorporating liability insurance, risk pools, and public reinsurance schemes. From a comparative perspective, the insurer-first compensation model adopted in the United Kingdom, the data access and record-retention principles reflected in the EU Product Liability Directive and AI regulatory framework, and the risk-based liability regime under the German Road Traffic Act provide valuable guidance for the development of the Korean legal framework. Accordingly, a Level 4 automated vehicle accident liability guideline should initially function as a soft-law instrument in pilot projects and insurance markets, and subsequently be institutionalized through amendments to the Automobile Accident Compensation Assurance Act and the Product Liability Act as commercialization advances.
Zitieren
APA
Kim, Y. K. (2026). A Study on the Composite Liability Structure for Level 4 Autonomous Vehicle Accidents - Focusing on the Compensation-First, Recourse-Later Model and Accident Liability Guidelines. Korean Insurance Law Association, 20(2), 149-195. https://doi.org/10.36248/kdps.2026.20.2.149
Vancouver
Kim YK. A Study on the Composite Liability Structure for Level 4 Autonomous Vehicle Accidents - Focusing on the Compensation-First, Recourse-Later Model and Accident Liability Guidelines. Korean Insurance Law Association. 2026;20(2):149-195. doi:10.36248/kdps.2026.20.2.149
BibTeX
@article{Kim2026,
author = {Kim, Young Kook},
title = {A Study on the Composite Liability Structure for Level 4 Autonomous Vehicle Accidents - Focusing on the Compensation-First, Recourse-Later Model and Accident Liability Guidelines},
journal = {Korean Insurance Law Association},
year = {2026},
volume = {20},
number = {2},
pages = {149--195},
doi = {10.36248/kdps.2026.20.2.149}
}
EUVIMED