The Evergreen Marine v Nautical Challenge case has brought fresh scrutiny to the application of the International Regulations for Preventing Collisions at Sea (COLREGs) as the maritime industry grapples with the integration of autonomous shipping. This landmark legal debate underscores critical challenges tied to rules such as the crossing rule (Rule 15) and the narrow channel rule (Rule 9). Since their inception in 1977, the COLREGs have relied on human interpretation, often incorporating subjective decision-making, such as the “ordinary practice of seamen.” However, in the era of autonomous vessels, maritime stakeholders face the daunting task of translating these nuanced regulations into deterministic algorithms capable of ensuring safe navigation without human oversight.
Research indicates that autonomous vessels have the potential to meet or even exceed human capabilities in complying with collision avoidance regulations. For instance, Rolls-Royce’s MAXCMAS project has claimed that its artificial intelligence can emulate good seamanship, even when encountering vessels that fail to act properly. While promising, this development raises complex implementation questions, particularly regarding the identification of "give-way" vessels and defining automated standards for "good seafarer behavior." The Evergreen case itself highlights the ambiguity present in interpreting Rule 9 and Rule 15, with judicial commentary noting the absence of definitive precedent for prioritizing these rules in specific scenarios. The intricacies of human judgment, historically relied upon to navigate such ambiguities, may prove exceedingly challenging to replicate in AI-driven systems.
Another layer of complexity emerges from the operational environment of mixed fleets. For the foreseeable future, autonomous and manned vessels will co-exist, creating a necessity for regulations to accommodate both seafarer decision-making and machine-led processes. The COLREGs’ distinctions between vessels "in sight" and those operating in "restricted visibility" may place heavy demands on AI developers to create adaptive systems that account for variable scenarios while maintaining compliance. Furthermore, while autonomous shipping holds potential to mitigate human-error risks—such as inattentiveness or misinterpretation of environmental signals—it introduces technological vulnerabilities. Programming deficiencies, sensor system failures, and unanticipated edge cases may lead to accidents, opening unresolved accountability questions. As yet, there is no established case law to address liability for collisions involving autonomous vessels.
The ethical dimensions of maritime AI add further complexity. When programmed systems face life-and-death situations, questions arise regarding how decision-making pathways can integrate risk mitigation while balancing ethical considerations. Such scenarios underscore the importance of establishing maritime-specific guidelines for autonomous operations, tackling not just practical technology implementation but the philosophical underpinnings of AI conduct.
Why this matters
The implications of this debate extend across operational and regulatory boundaries, with shipowners, developers, and insurers facing significant challenges. For operators, managing a transitional period where autonomous vessels share waterways with manned ships necessitates a reevaluation of navigation rules to prevent conflicts. The liability risks associated with programming errors or AI decision-making could elevate insurance premiums and shift the risk calculus for shipowners in an industry already navigating stringent cost pressures.
From a regulatory perspective, the ambiguity currently embedded within COLREGs means enforcement remains murky for incidents involving autonomous technologies. Developers are under increasing pressure to translate complex human-based regulations into programmable guidelines, which could require years of iterative advancements and potentially expose non-compliant tech to litigation risks. Classification societies and flag states may also need to establish new certification pathways for autonomous vessel systems to address compliance gaps.
This discourse further impacts ports and coastal authorities. High-traffic areas with narrow channels—such as the ports of Ho Chi Minh City and Hai Phong in Vietnam—may experience disruptions without clear distinctions on how autonomous and manned vessels coordinate movements. Operational frameworks for mixed fleets could include restrictive measures until broader adoption consensus is achieved.
The maritime industry’s integration of autonomous vessels, while offering the promise of reduced collision risks, is contingent on modernizing regulatory frameworks to account for AI decision-making. As highlighted by the Evergreen Marine case, the interpretative nature of current COLREGs cannot seamlessly translate into programmed behavior without significant adjustment. Shipowners, developers, and regulators will need to work collaboratively to resolve ambiguous navigation rules and establish a liability framework that reflects the operational realities of autonomous shipping. This evolution of maritime law will significantly shape the safe and efficient deployment of autonomous vessels globally.