Autonomous Vessels and the Future of Marine Claims

The navigational bridge of a commercial vessel.

On the 15th of February 2023, NYK and three of its group companies were awarded Japan's Minister of Land, Infrastructure, Transport and Tourism Award for their work on fully autonomous ships. The award recognised MEGURI 2040, a project run by the Nippon Foundation and carried out through the DFFAS consortium. The consortium brings together 30 companies, including shipbuilders, equipment makers and insurers, and in February and March 2022 it ran a trial simulating the actual operation of a fully autonomous ship between Tokyo Bay and Ise Bay. Its stated aims are to tackle labour shortages and reduce maritime accidents.

Much of what follows depends on a distinction that is often blurred. A ship with automated features still has a crew on board in command but relies on systems for particular tasks, such as collision avoidance, automatic manoeuvring or docking assistance. A remotely controlled ship is operated by people in a shore control centre and may or may not still carry some crew. A fully autonomous ship decides and acts on its own so that the system, rather than a master, is responsible for navigating the voyage. Only the first category is in widespread commercial use today, and each shifts risk, and the question of who answers for it, in a different way.

Insurers have not been standing still either. The Shipowners' Club launched the world's first autonomous vessel policy in 2018 and now covers around 80 small uncrewed craft of under 24 metres, without a single claim so far. Its Commercial Director, Mark Harrington, does not expect large uncrewed vessels to be commonplace before 2040, largely because international regulation must change first.

That change has begun. In May 2026 the IMO's Maritime Safety Committee adopted the International Code of Safety for Maritime Autonomous Surface Ships, known as the MASS Code, which took effect on 1 July 2026. It is non-mandatory and applies to cargo ships, and its purpose is to let states and industry gain experience before a mandatory code is introduced, which the IMO expects to adopt by July 2030 and bring into force on 1 January 2032. In the meantime, autonomy is arriving in stages. Automated features such as collision avoidance and docking assistance are already implemented on large ships, and AI is being applied to predictive maintenance and voyage optimisation. The realistic picture for the next two decades is a mixed fleet of crewed, partly automated, remotely controlled and, eventually, fully autonomous vessels, all of which insurers will have to price at once.

The first effect is likely to be fewer of the claims that dominate P&I today. Most are people-related, such as injury, illness and repatriation, so removing the crew removes much of that. Harrington expects larger fully automated vessels to bring a meaningful reduction in the cost of claims, though this is a projection rather than an observed result, since there are as yet no claims on vessels of that size to measure. Separately, IUMI has noted that automation in smart ports has cut operational accidents by up to 80 per cent, but that figure concerns port operations, not autonomous ships at sea, and so says little about what to expect on the water. However, fewer claims do not mean the same claims. Groundings, collisions and machinery damage will increasingly trace back to software faults, sensor failure or lost connectivity rather than navigational error, and establishing what went wrong may mean analysing code and data logs rather than hearing from a master.

Machinery claims may also begin to resemble product liability disputes, with owners pursuing manufacturers over defective systems rather than simply claiming under their hull cover. Fire remains a concern, and Harrington lists it alongside technology failure, cyberattack and public perception among the risks of uncrewed ships. Remote operation adds exposures of its own, since a lost data link or a delay in a shore operator's response could turn a manageable incident into a casualty. Pollution and wreck removal claims will not disappear either, but they may become harder to attribute.

The harder question is who is responsible. The collision regulations and liability conventions assume a master on board, and the MASS Code does not change that. It is a safety code, concerned with how such ships are designed, certified and operated, and it is voluntary for now. If the decisive error was made by a remote operator in a shore control centre, a software supplier or the algorithm itself, liability may shift between owners, operators and manufacturers, and recourse claims against technology providers will inevitably become routine. Cargo claims raise a similar issue: a carrier's duty to provide a seaworthy ship arguably extends to software and connectivity, while the traditional defence for errors in navigation by the master or crew sits awkwardly with an autonomous system. Charterers may also find themselves in disputes over off-hire and performance that turn on system data, a question inextricably linked to who owns and can access it.

For the wider market, the difficulty lies in pricing and accumulation. Premiums are built on decades of claims data, and for large autonomous ships there is almost none. Harrington has urged operators to share near-misses as well as claims, and IUMI has suggested a global risk data pool. Cyber risk adds an accumulation problem: ships running the same software share the same vulnerability, so a single flaw or attack could produce simultaneous losses across many owners, and a market used to largely independent losses cannot easily inoculate itself against that. IUMI describes cyber as a primary systemic risk, with one major attack potentially costing hundreds of millions of dollars. The P&I model, in which owners pool the risks of their own operations, also becomes harder to apply when responsibility is split between owners, remote operators, software vendors and ports. Perhaps this is why IUMI is calling for new products, such as Unmanned Equipment Liability and Cyber Business Interruption cover.

Ultimately, we are yet to see how quickly large autonomous vessels will arrive, but the direction is clear. Autonomy is likely to reduce human-error claims while replacing them with fewer, more complex and potentially larger losses, in which causation, liability and accumulation are harder to pin down. As Harrington points out, investment in autonomous shipping depends on insurers being able to offer financial security. The market should not wait for a major casualty to galvanise it into action; it needs to build the data, wordings and liability frameworks before large uncrewed vessels become commonplace.