03 July 2026
Bilbao
The introduction of autonomous ships into international maritime transport poses a major regulatory challenge for contemporary maritime law. The gradual replacement of crews by artificial intelligence, sensor technologies and remote control centres is reshaping the traditional legal framework, which was designed on the basis of a human presence on board.
Aitor Dominguez's doctoral thesis, Los Accidentes de la navegación y los buques autónomos, concludes that artificial intelligence has no legal personality and cannot hold rights or obligations. As a result, the shipowner will continue to bear primary responsibility for the ship and for any damage arising from its operation, preventing technological autonomy from becoming a means of avoiding liability. Although the shipowner will bear primary responsibility, the author of the thesis points out that they may subsequently exercise a right of ‘recourse’ (a legal term meaning to recover that liability) against any new parties that may emerge, such as AI specialists, software developers, shipyards, engineers, shore-based support personnel and supervisors.
In his thesis, Dominguez examines how automation affects three established principles of maritime law: collision, salvage and general average (where part of the cargo is deliberately sacrificed to prevent greater loss).
No captain on board
From both a technical and a legal perspective, navigation has traditionally been based on the expertise of the crew and the authority of the captain, who is responsible for the ship's operation and acts on behalf of the shipowner at sea. However, at the most advanced levels of automation, such as ships operated entirely by artificial intelligence or under remote supervision from shore, this traditional structure disappears.
The thesis highlights that, in the absence of personnel on board, operational responsibilities are transferred to remote control centres, software developers, cybersecurity specialists and shore-based operators. This shift introduces new risk factors that are not addressed in detail by current legislation. These include technical and operational risks, such as failures in sensor systems, loss of satellite connectivity, or the inability of algorithms to interpret unforeseen weather conditions. Another factor is cybersecurity, as reliance on connected systems makes cyberattacks a major threat to the security of maritime traffic. A third risk highlighted in the thesis is the traceability of evidence: the absence of a crew can hamper accident investigations, as there are no witnesses. Consequently, in the event of an incident, the judicial investigation will depend on the preservation and analysis of technical data and records of the decisions taken by the AI.
In the event of maritime salvage
The duty to render assistance to persons in distress at sea is a fundamental principle set out in international conventions. However, the research points out that an uncrewed ship faces significant practical limitations both in receiving assistance and in providing it.
If an autonomous ship requires salvage, the absence of a crew on board makes it difficult to carry out essential operations such as securing tow lines or containing fires immediately. Conversely, if the autonomous ship is required to assist another vessel, the current design of these ships—which lack accommodation, galley facilities and medical facilities—prevents them from providing appropriate shelter and care for survivors.
To address this scenario, Aitor Dominguez's research proposes making it mandatory for autonomous ships to be equipped with automated safe refuge areas, detection systems and direct communication systems linking them to shore-based control centres.
Decision-making at sea
Collision is the most common type of accident in maritime transport. The traditional legal framework is based on the International Regulations for Preventing Collisions at Sea (COLREGs), whose rules assume the presence of a human lookout. The thesis argues that computer vision systems should be certified by classification societies not only in terms of their mechanical components, but also through an assessment of the reliability of their decision-making algorithms. Similarly, it is suggested that specific signage be introduced to identify autonomous ships to the rest of the fleet at sea.
Finally, the study examines general average, the legal principle that governs deliberate sacrifices (such as jettisoning cargo) made to save the ship and the remaining cargo from a common peril. The thesis concludes that, although technology makes it possible to calculate with mathematical precision which assets should be sacrificed to stabilise the ship, the formal decision should still be confirmed by a human decision-maker on shore.
Some ships are already capable of operating fully autonomously, although their functions and operational scope remain very limited. These are primarily research and military vessels. However, the large autonomous merchant ships that are the focus of this thesis already exist. One example cited by Dominguez is the Yara Birkeland, a Norwegian autonomous, all-electric container ship.
This research therefore argues that the efficiency and reduction in human error promised by autonomous ships do not exempt the sector from a thorough review of the regulatory framework. The legal certainty of maritime trade will depend on legislators' ability to adapt the existing legal framework before this type of navigation becomes widely established on international shipping routes.
