<![CDATA[The development of fully autonomous ships challenges collision-avoidance rules that were drafted on the assumption that navigational judgment would be exercised by a master and crew. This article examines, first, whether artificial intelligence-based navigation can satisfy the open-textured standards in Rules 2,5,7, and 8 of the Convention on the International Regulation for Preventing Collisions at Sea 1972 (COLREGs), and, second, how legal liability should be attributed when an autonomous system generates the relevant decision. The study employs normative legal research using statutory, conceptual, and systematic approaches to COLREGs, the United Nations Convention on the Law of the Sea. It finds that autonomous systems can reproduce the formal sequence of lookout, risk assessment, and collision-avoidance action, but cannot automatically satisfy the contextual judgment embedded in good seamanship, special circumstances, and timely action. The MASS Code reduces this difficulty through operational envelopes, risk assessment, human oversight, override capability, traceability, and data logging. However, it preserves a gap between the system that factually produces a decision and the human or corporate actor to whom legal responsibility is assigned. Accordingly, the article argues that legal responsibility for fully autonomous ships should be assessed through five interconnected lenses: control over the autonomous function, organizational responsibility, technological causation, flag-State oversight, and insurance and risk-distribution mechanisms. These lenses provide a contextual framework for allocating responsibility according to the actor's legal duty, degree of control, causal contribution, and capacity to prevent or mitigate harm.]]>
Copyrights © 2026
