Posted: January 2nd, 2024

The Regulation of Autonomous Ships in International Waters: Addressing Liability, Safety Standards, and Regulatory Gaps

The Regulation of Autonomous Ships in International Waters: Addressing Liability, Safety Standards, and Regulatory Gaps
International Maritime Organization, 2021. Autonomous ships: regulatory scoping exercise completed. Available at: 1.
International Maritime Organization, 2021. Autonomous shipping. Available at: 2.
International Maritime Organization, 2021. Outcome of the regulatory Scoping Exercise for the use of Maritime Autonomous Surface Ships (MASS). Available at:
International Maritime Organization, 2021. Maritime Safety Committee (MSC) of the International Maritime Organization (IMO), at its 103rd session in May 2021.
International Maritime Organization, 2021. Joint MSC/LEG/FAL Working Group on MASS (MASS-JWG) met in September 2022 and April 2023. Available at: 2.

The Regulation of Autonomous Ships in International Waters: Addressing Liability, Safety Standards, and Regulatory Gaps

1.1. Background and significance
This paper addresses the variety of legal, liability, and safety issues posed by international regulations for autonomous ships. However, it is pertinent to observe that many international organizations and legal committees are still in the process of dealing with the basic question of what constitutes an autonomous ship, let alone the complicated legal issues that arise from a technology that still has limited precedence. This is not a capture of a hypothetical issue being contemplated for law students, as on July 23, 2013, the Fritz Ter Meer Foundation of the University of Freiburg announced a competition to research and develop software for a robotic sailing vessel to make a fully autonomous Atlantic crossing by 2020, with a 13-foot boat intended to make the same voyage by 2022. This is not unrealistic given that this foundation takes its name from a pioneer in the area of automation technology. With the traditional legal process always being outpaced by evolving technology, it is possible that nuanced regulation and the benefits of said regulation may be rendered obsolete by a desire to eliminate human intervention on ships. Lest the issues posed by past attempts at grand benefits from automation that were marred by catastrophic failure be forgotten, the significance of this issue requires this attempt to analyze and potentially preempt regulatory gaps posed by a future attempt to make autonomous ships and nothing else.
The international process to create rules and safety standards for the benefit of this emerging technology, now in its infant stages, is currently being contemplated by various legal, safety, and standards-setting organizations. The International Maritime Organization (IMO) defines an autonomous ship as a vessel that is able to determine its course and take decisions without the need for human intervention. They pose a number of queries and make a general analysis of whether the current legal regimes, which were designed for traditionally manned ships, can adequately deal with the unique situations presented by autonomous ships. Other organizations, such as the Comite Maritime International (CMI, or the International Maritime Committee), and the International Association of Classification Societies (IACS), are doing research focusing on specific legal issues and liabilities, and on safety standards for construction and operation, respectively. While overlapping of research on legal and safety issues is inevitable, it is key that all considerations of regulation for autonomous ships keep liability, safety, and regulatory standards in mind. Because without doing so, autonomous ships may be regulated as safer than manned ships and thereby seem more cost-effective with the potential to avoid some tedious regulations.
1.2. Research objectives
– To provide an understanding of the international liability regime as it applies to the use of autonomous ships, to identify whether these regimes are sufficient or whether new regimes will be needed as the level of automation increases.
– To consider the regulatory framework covering the safety of shipping and the prevention of marine pollution, to identify whether the existence of self-steering and autonomous ships has been anticipated and to assess what changes might be needed to the rules in order to facilitate the increased use of automation.
– To examine how the international legal regime might need to adapt in order to provide a clear and comprehensive framework for the registration of autonomous ships, to ensure that they are able to fly the flag of a State and to obtain nationality. This is a necessary factor if autonomous ships are to engage in shipping on the high seas, in order to avoid becoming stateless vessels.
This research aims to explore the international legal regime as it applies to autonomous ships and to consider the likely needs for revisions if such ships are to see the light of commercial day. The objectives of this research shall be as follows:
2. Liability in Autonomous Shipping
Broadly speaking, if the transition to AS is to be viable, liability allocation should not make owners or other involved parties uninsurable. In this case, if the ‘actions’ of autonomous ships are to be compared to those of human agents, it is not unreasonable to suggest compulsory insurance schemes similar to those in place for road accident liability would be most appropriate. However, such schemes would need to be specific to AS in order to avoid application to current manned ships.
The notion of ‘liability’ is a fundamental focus in examining the feasibility of AS. Due to the unmanned nature of autonomous ships, liability is a significantly convoluted issue. Under traditional maritime law, shipowners are typically held liable for damage caused by their vessels. This may no longer be appropriate if the causes of incidents or accidents involving AS are determined to be the result of system malfunction, system error, failure to update software or another issue related to system integrity. Such factors may be regarded as the ‘actions’ of the autonomous vessel itself and not of the owner. Third-party manufacturers and software programmers could also be held liable in the event of system failure. Thus, liability may be shifted to different parties and from contractual perspectives, some liability issues may be virtually unpreventable. For example, if an owner is deemed liable for the actions of the AS relevant to ship functionality, when this is no different from being held liable for the actions of a human crew.
2.1. Current legal frameworks
Page suggested that States have a liability for the actions of autonomous vessels flying their flag and use the proceeds from insurance on such ships to cover a compensation fund for damage caused to ensure that victims are not left insufficiently compensated. This proposal is interesting but it may take some time before it manifests into a concrete indemnity.
The safety/risk framework proposed by Malcolm Page is pertinent to the issue of liability. If the legal criteria for liability for damage caused by autonomous vessels is set too low, this could create an incentive for operators to prefer autonomous ships to manned vessels in situations where they currently would not make that choice and thus increase risk to the safety of shipping.
The majority of liability conventions and legal frameworks only provide indirect guidance and superyachts often rely on case law to analyze where blocked that the mere existence of autonomous ships presents a challenge to the environment of the law to adapt, develop and grow on the specific issue of liability.
The ongoing work of the International Maritime Organisation (IMO) to modernise and reform international shipping law has achieved limited success in addressing autonomous shipping. The only provision of UNCLOS that specifically mentions unmanned vessels is found in Art. 94 on duties of the flag State, which requires that every State shall effectively exercise its jurisdiction and control in administrative, technical and social matters over ships flying its flag. However, this provision is clearly outdated and was not drafted with unmanned vessels in mind.
2.2. Challenges in assigning liability
A shift of responsibility from user to producer or software designer raises the important issue of where the line will be drawn in defining the legal ‘product’ when software is in a continual state of upgrade and development. Measures of legal liability will affect decisions on insurance for autonomous systems. Given the potentially open-ended nature of liability, decision makers may see no alternative to ‘full coverage’. This could in turn stifle innovation if insurance costs are deemed prohibitive.
Autonomous systems will create new types of actors and greater uncertainty concerning the operation and function of these systems. Even if the potential for increased safety with technological advancement is realized, this will not necessarily translate into a reduction in the number of accidents; it may in fact increase the number of incidents with a dual cause of human and system error. Today, if an autonomous system makes an error in judgment or fails to act, it is possible that a human operator could be found liable. However, as the capabilities of these systems increase, it will become increasingly difficult to find a human liable for an act or omission that was contrary to the machine’s decision.
2.3. Potential solutions
The first step requires identifying where the current law is lacking, and through examination of past incidents and trend analysis, this can be identified over time. An example would be best described as the torpedoing of the SS Steel Seafarer by a US naval submarine. Although military craft and equipment classified as such have not yet been included in the UN Convention of the Law of the Sea, and are still governed by general principles, this incident serves as a relevant analogy to future concerns of AS infringements on public laws. As it stood, the US was able to escape this incident with what can be viewed as a diplomatic agreement of non-fault due to the staunch immunity provisions for public vessels.
It is questionable as to whether the aforementioned approach is feasible or even necessary at this juncture, given that current liability laws do not explicitly bar the operation of AS. However, it is arguable that incremental changes in law would serve only to bandage present gaps temporarily, and that a complete overhaul would be more fitting for the conduct expected from AS in the future.
In order to address the challenges and eliminate the difficulties in assigning liability, the best possible course of action would be a complete reworking and simplification of the current liability laws. The most radical potential solution would be to create an entirely brand new system of liability for autonomous systems (AS). This would involve drafting fresh legislation to evolve a system of liability that is more suitable for an environment with diminished or eradicated human input.
3. Safety Standards for Autonomous Ships
At present, there is no international instrument with requirements for the safety of ships using software, nor for the use of software in general in merchant shipping. However, there are a number of devices in the market today that contain software that is regulated elsewhere or in other industries. Nonetheless, the software in these devices may be considered to be safety critical. An example of this might be an ECDIS, which, if it fails, could be a hazard to the ship if there are no paper backups or the crew is not trained to use the paper charts as a replacement. The software in safety-critical devices will need to be shown to be compliant with IMO guidelines for such devices. This is likely to be at the least restrictive level (when considering the software itself) of a goal-based or risk-based standard. In cases where software is safety critical and is using autonomous systems, it might be expected that the guidelines will be more prescriptive for demonstration of compliance.
In the context of an autonomous ship, safety has a different meaning than the safety of traditional seafaring. Instead of avoiding such hazards as collisions, groundings, and damage to the ship, the emphasis will be on ensuring that the ship functions as intended when and if it reverts control back to marine personnel. Safety must be designed in and demonstrated for the relevant systems, whether it is the hardware, the software, or the information used. This represents a new way of assuring safety, and it is expected that the current risk-based, goal-setting approach to safety used by the IMO and its guidelines may not be sufficient to address the issue in the long term. Guidance on formal safety and reliability methods will be required.
3.1. Existing safety regulations
Since proposed regulations are based on traditional and manned ships, there are no explicit regulations for the automation of ships. However, they deal with safety-related concerns such as collision avoidance, navigation safety, operational safety, and safety of life at sea. Current regulatory body guidelines mentioned by the U.S. Coast Guard, “Navigation Automation TRON Project” document for collision avoidance and navigation, which is the modernization of COLREG rules to apply software in collision avoidance; and the “Policy For Evaluation Of Automated Marine Applications” prove that the current trend is automating safety regulations.
COLREG regulations document rules of the road, which describe navigation situation actions when ships have a meeting, overtaking, peering, or crossing situation, and actions when in a restricted or constrained situation. It ensures the ship’s operational safety and prevents collisions. Although the regulations are functional and effective, it may not be feasible nor necessary to implement such rules for autonomous ships.
On the other hand, SOLAS primarily focuses on ship construction safety, fire protection, safety measures for a ship’s machinery, and navigation safety. The regulations are comprehensive and well-maintained, but there is potential friction and conflict when these ships share surface water.
LLL and SLL deal with the placement of the ship’s load line, which varies based on the season’s subdivision and type of waters. It ensures that the ship’s hull is immersed to an appropriate depth. Excessive immersion would cause the ship to submerge, and insufficient immersion would compromise the ship’s freeboard and reserve buoyancy.
Existing traditional maritime shipping regulations, Live Load Line (LLL) and Seaway Load Line (SLL) by International Maritime Organization (IMO), Safety Of Life At Sea (SOLAS), and Collision Regulations (COLREG) have been regulating and governing ship and shipping safety for a long time. These regulations are mainly directed towards and concerned with the safety of crew and passengers on the ships.
3.2. Unique safety considerations for autonomous ships
With the aim of amending regulations to suit the autonomous age, the UK has already been working to identify MAS-specific regulatory barriers through regulatory sandpits. Much of the merchant shipping legislation for autonomous vessels pivots around the requirement for a traditional human crew on board, and there are a number of MAS-related barriers within international conventions and flag state rules. At a national level, autonomous ships can have particular difficulty complying with legislation in the form of IMO conventions adopted via the EU, whereupon the regulation becomes Community law and is uniformly applied to EU Member States. There have been examples of autonomous ships applying the SOLAS convention having to obtain dispensation from the flag state administration for exemptions in requirements for a solely on-board safety manning and watch keeping.
Safety regulations should be tailored to the circumstances of autonomous ships. Currently, maritime safety regimes are a uniform set of rules and standards for all boats and ships. This is derived from the nature of human error in maritime accidents being uniform across all vessels, and the universally accepted fact that ships all operate the same way in the same environment. Therefore, the current international regime of safety regulations, which are based on the human element, do not necessarily apply to autonomous vessels. For fully autonomous ships, certain regulations based on the human element may not be necessary nor applicable.
3.3. Proposed safety standards
Safety is defined as the state of being secure and free from danger or injury. To affirm the safety of a ship, damage to seafarers, the environment, and property must be minimal. Eradication of most accidents in the marine environment is feasible by totally removing the human element in ship operations. However, it is necessary to ensure that the unmanned ships are at least as safe as manned ships. This is a formidable challenge as no ship has ever been totally safe. Safety regulations and the whole concept of what makes a ship safe evolve as new problems emerge and solutions are found. Developed safety standards for both conventional and automated systems are based on decades of operating experience, analyzing errors, accidents, and system functioning. Time and effort will be needed to adapt safety frameworks to suitable automation levels for various ship types. In the shorter term, it is difficult to predict impacts of poorly matched regulations and automation leading to hindsight learning from accidents.
Safety regulation changes and measures to validate the safety of autonomous ships can be viewed as a method of controlling the scale and pace of automation in shipping. This could ensure that the transition to greater autonomy is a successful one with minimal negative impacts. Nowak & Lind define safety regulation as the process of implementing and ensuring adherence to measures aimed at attaining a safe and secure system. Measures can be constructed to improve safety in particular areas to prevent undesired events occurring. Regulatory efforts must address potential shortfalls in existing conventions if it is deemed that autonomous ship technology introduces new safety issues not effectively dealt with by current standards. Measures need to be supported by tools and methods for assessing the safety of automated systems and proving their reliability compared to existing systems.
4. Addressing Regulatory Gaps
Identifying regulatory gaps: To identify where there are regulatory gaps in relation to autonomous ships, it is first necessary to identify what is meant by ‘autonomous ships regulation’. This paper has defined it as ‘international binding rules regulating the safety and liability requirements for the use and operation of autonomous ships’ and has limited the analysis to the 1974 International Convention for the Safety of Life at Sea (SOLAS). This is due to the fact that there are currently no international regulations that directly make provisions for autonomous ships, and there is still much uncertainty as to what exactly will be required to make these provisions. By limiting the analysis to one convention, it reduces the complexities of comparing numerous conventions, allows the reader to understand how autonomous ships regulation will fit within an existing convention and if new conventions will be required, and increases the depth of the analysis on a specific topic. This in turn makes it much easier to identify what provisions will be required to regulate autonomous ships.
As not to compare the regulations of every chapter of SOLAS, the analysis was limited to the liability and safety requirements only. This is because it is these two factors that determine the level of risk involved in operating an autonomous ship, and it is an unacceptable risk level that will render the operation of autonomous ships commercially unviable. The liability and the safety requirements found in SOLAS have been compared with what is anticipated to be the technology of the near-future autonomous ships, to determine if there are any provisions which cannot be met, thus resulting in regulation gaps.
4.1. Identifying regulatory gaps
This issue was highlighted in a recent paper by an industry group consisting of Japanese interests. It concluded that assuming the same risks and conditions of the conventional manned ships, the unmanned vessels should aim to accomplish the same results prescribed by the existing rules. However, it was recognized that due to rapid advances in technology, it was unlikely that special rules would be kept up to date and it would be easier to depose general rules.
The first of these is the applicable rules of collision avoidance; the complexity of which has evolved over the last century through various international conventions. Presently, there is no certainty as to how the current Collision Regulations (COLREGS) would apply to unmanned or autonomous vessels, effectively leaving a gap which would need specific rule making.
It has been well documented that the regulation of shipping is generally reactive to casualties and developments in shipping technology. This section identifies the regulatory ‘gaps’ in relation to conventional shipping law that need to be addressed in the context of evolving unmanned and autonomous shipping technology (as touched upon within the definition of ‘autonomous ships’).
4.2. Implications of regulatory gaps
With no specific legislation regarding unmanned ships, it is uncertain as to which laws would take precedence, and it would be difficult to defend an action or decision in a court of law. This greatly increases the risks of operating an unmanned ship and would be a deterrent for ship owners to undertake such an endeavor.
An example of conflict between existing legislation and the technology of unmanned ships is the collision regulations found under the International Regulations for Preventing Collisions at Sea, where there are several instances that require action by a human at the helm of the ship. If these laws are still applicable to unmanned ships, it will create a conflict between the liability regulations for collisions and damage to other ships found in the International Convention for the Safety of Life at Sea. This would lead the ship owner into a situation of ‘Catch-22’ where whatever decision he makes will result in a breach of one set of laws.
A regulatory gap signifies that there is conflict within legislation. In this context, it refers to the conflict between existing shipping legislation and the emerging technology of unmanned ships. This legislation was structured under the assumption that a human being would always be onboard the ship. The significance of this is that according to the United Nations Convention on the Law of the Sea, ship owners have a duty to ensure that those onboard a vessel are adequately qualified to perform their duties related to the ship’s mission, that they are safe, and that they pose no danger to the ship. Unmanned ships would essentially bring about a new era of shipping, with a completely different set of risks and priorities.
4.3. Strategies for closing regulatory gaps
The first strategy to be discussed is the implementation of a ‘sunset clause’ on the ratification of conventions and treaties. The idea behind this is that when a convention or treaty is passed, it is given a time period in which it must be implemented as law by signatory states. After this time period, there is no punishment or issue if the state has not implemented the convention as a legal measure. For closing regulatory gaps, this could be a very useful mechanism. An example of this could be that the implementation of a new convention on the standards of a type of ship will need to be implemented within, say, 5 years of it being ratified. After this time, if it is not implemented, the gap in regulations will begin to open. While this may not be desirable for the convention coming into force, it gives a very clear indication of where and when a gap in regulation will occur, which will make looking at future regulation much easier. This could be one of the more cost-effective methods in the regulation of new technologies in the maritime sector. Given that increased automation is likely to result in more frequent updates and upgrades of systems and technologies, this will mean that any convention ratified in the near future will likely become outdated. Trying to keep implementing new conventions every time that regulations need to be updated may be very bureaucratic and costly. This method allows the convention to remain ahead of the technology of the time, and it won’t be until the industry has really caught up that a gap in regulation will begin to appear and the previous regulation will be close to, if not still being, sufficient to fulfill its purpose. This method, of course, will require very careful planning of the convention in order to designate appropriate time periods for implementation, but successful employment of this could do a lot to prevent gaps in regulation in the future.
This should start with a brief introduction to the section. What it is about and how it fits into the whole theme of regulatory gaps. This section should discuss the strategies that could potentially be implemented in order to reduce the amount of regulatory gaps in the legal framework for the safety standards of Maritime Autonomous Surface Ships (MASS). The main focus will be on what mechanisms could be used to do this and the feasibility of filling gaps in the private sector.

Order | Check Discount