The introduction of autonomous ships will raise numerous legal questions in the years to come. In this article, we focus on some central liability rules in Norwegian maritime law, and assess whether they can be applied if damage is caused by autonomous ships.

The introduction of autonomous ships will likely be a hot topic in the Norwegian maritime industry in the coming years. Several autonomous cargo ship projects are currently in development in Norway. The two perhaps most prominent projects are the electrical container vessels for Yara (the “Yara Birkeland”) and ASKO respectively. Both projects aim to change local and regional cargo logistics from truck transport to autonomous short-sea shipping.

In this article, we will focus on the not so distant scenario of fully autonomous ships in Norwegian territorial waters. Will the shipowner be liable if something goes wrong and an autonomous ship causes damage?

Liability rules in Norwegian Maritime Law

The liability rules in the Norwegian Maritime Code (“NMC”) are based on the principle of negligence. Vicarious liability for the shipowner pursuant to section 151 is conditional upon negligence on the part of the master, crew, pilot, tug or others performing work in the service of the ship. The collision rules in NMC chapter 8 are (in accordance with the 1910 Collision convention) also fault-based and thus conditional upon a crew maneuvering the ship.

Due to the international character of shipping, the Norwegian liability rules are quite similar to rules in other jurisdictions around the world. However, there is one important exception: More than 100 years ago, and in response to the industrial revolution, the Norwegian Supreme Court developed a doctrine of strict liability for damage caused by certain industrial, dangerous activity. The doctrine has no direct parallel in jurisprudence from other countries, including the other Nordic countries.

The classic case is from 1875, when a dynamite factory exploded and caused considerable damage to the surrounding properties. The Supreme Court reasoned that the owner of the factory, who had created the danger for the surroundings, should also bear the financial risks for such damage caused thereby, irrespective of proof of negligence. From an economic point of view, it was argued that the factory could include the risk of such liability in the operating costs: Such accidents will happen from time to time with certain statistical certainty. Later, the doctrine expanded to other areas of life, and is today an important part of Norwegian tort law.

In 1921, the Supreme Court applied the strict liability doctrine for the first time to a shipping case in Rt. 1921 p. 519 (Neptun). A ship crashed into a bridge and caused damage after suffering technical failure. As neither the shipowner nor the crew had acted negligently, the shipowner was not held liable based on vicarious liability. However, the Supreme Court referred to the doctrine of strict liability for causing danger to the surroundings, and found the shipowner liable for the damage to the bridge. This first case was followed by a similar case in Rt. 1952 p. 1170 (Sokrates), this time involving a ship crashing into a pier due to technical failure. In the Sokrates case, even though the basis of liability was not-statutory, the Supreme Court allowed the shipowner to limit his liability according to the statutory rules in the NMC. The Neptun and Sokrates cases have not been overruled.

The doctrine of strict liability has also been argued in shipping cases that did not involve a technical breakdown, but with reference to the more general rule regarding danger to the surroundings. However, the Norwegian Supreme Court has been careful not to expand the doctrine to shipping cases outside those involving technical failure, and the doctrine has not been applied as basis for liability since 1952. For example, the Supreme Court decided in Rt. 1948 p. 1044 (Sirius) that a shipowner is not liable when the damage is caused by the ship due to bad weather conditions (force majeure) rather than technical failure. In conclusion, in shipping cases, the strict liability rule is restricted to damage caused by technical failure.

When considering an application of the doctrine, the Supreme Court has referred to the Maritime Code, and pointed out that there is not much room for a strict liability rule besides the statutory liability rules – which require negligence. For example, it does not apply in collision cases, which are regulated exhaustively by the NMC. This is illustrated by the case in Rt. 1971 p. 843 (Marna Hapsø). Similar to the course of events in Neptun and Sokrates, the crew on board the vessel was not at fault, but lost control of the ship due to technical failure when arriving at port. The ship then crashed into a moored ship. The claimant relied on the Neptun and Sokrates cases as precedents. The Supreme Court rejected this contention, and referred to the incident as a collision, triggering the statutory collision rules. The shipowner was not liable, as the collision was caused by technical failure and not by negligence.

The doctrine thus applies in parallel with vicarious liability under section 151 of the NMC, i.e. in cases where negligence by the shipowner or the crew cannot be established, and the damage is caused to the surroundings due to technical failure on the ship.

Risks associated with autonomous shipping – fault or technical failure?

Today, human errors are the main cause of maritime accidents. It is therefore expected that the introduction of fully autonomous ships will reduce the general level of risk for marine accidents. However, the risk will never be fully eliminated. There will always be a risk of “classical” technical failure, such as breakdown of machinery.

In addition, there are inherent risks that already exist in highly automated vessels today, such as sensor faults and software bugs. An example from aviation is the Boeing 737 MAX accidents in 2018-19, in which the aircraft’s angle sensors provided erroneous information to the aircraft’s altitude control system. The system was difficult to override manually – and the aircraft crashed.

The risks mentioned above may all be defined as risks of technical failure. Other risks may be more difficult to define. On fully autonomous ships, artificial intelligence will manoeuvre the ships instead of humans: Data from the surroundings will be gathered from sensors, which will then be analysed and used by algorithms to make decisions. Similar to humans, algorithms may make mistakes due to lack of training in relevant marine environments. On the other hand, errors in the programming algorithms may also be caused by mistakes by the programmer. Accordingly, in the existing liability regime, damage caused by faulty algorithms may potentially be classified as a product defect (and thus as a technical failure), but also (depending on the root cause) as negligence.

Liability for autonomous ships – can the existing rules be applied?

As the Marna Hepsø case illustrates, the collision rules require negligence as a condition for liability.  However, it seems plausible that the collision rules, with some amendments, can be applied to autonomous shipping. If two autonomous ships collide, it is possible to share the costs proportionally based on the level of “fault” considering the navigations rules (COLREGs): Even if the root cause is technical failure, the ship that acted “objectively wrong” may be deemed at fault. However, this obviously presupposes action on the part of the legislators to amend or add to the collision rules.

The existing rules on statutory vicarious liability appears less suitable to regulate liability for autonomous shipping. One could discuss whether the faults lie within the vicarious liability of the shipowner, for example if the fault lies with the manufacturer, programmer etc. However, the application of this rule on autonomous ships will likely be complicated and unpredictable.

Due to the introduction of autonomous ships, the long “sleeping” doctrine of strict liability for technical failure might prove to be very relevant in the near future. For the purpose of deciding liability, most or all failures on autonomous ships can be defined as technical failures. Accordingly, the strict liability doctrine, as developed by the Supreme Court, will be applicable to damage caused to the surroundings by autonomous ships.

Compared to vicarious liability, a strict liability rule would also be a more natural and predictable rule when autonomous ships causes damage: The shipowner will be liable as long as the damage is caused by technical failure, which will include more or less all failures on the ship. A strict liability rule should not be of concern for the shipowner: Firstly it follows from the Sokrates case that the shipowner can limit liability according to the NMC, which is in line with the LLMC Convention of 1976 (1996 Protocol). Secondly, an extended P&I insurance will most likely be available on the market for autonomous ships.

The subject of this article is based on the master thesis of our associate Arne Moss Westgård, which was published in MarIus 527 (2019). The subject was also explored in an article recently published in Morgenbladet that Arne co-authored with his supervisor, Professor Trond Solvang at the Scandinavian Institute of Maritime Law. For further reading on the subject, Professor Solvang has recently co-authored a book called “Autonomous Ships and the Law”, Routledge (2021).

  • Published: