MarIus

3.1.a Setting the scene

Before addressing the four pitfalls outlined above, we shall first briefly discuss the features of remotely controlled vessels (scenario 1) and autonomously operating vessels (scenario 2) in order to qualify the analysis below. These features will be taken into account below where relevant for the discussion.

3.1.b Vessels being remotely controlled at the time of the incident: Still within the carrier’s scope of liability?

Looking first at scenario 1, if the crew, or a part of it, e.g. the bridge officers, are replaced by officers or other personnel remotely controlling the vessel from ashore, the risk of human error in the operation of the vessel is moved geographically away from the vessel. Also, new possibilities of technical malfunctions present themselves, connected with e.g. the transmissions of data and communications to and from the vessel’s steering systems etc.,. The central questions in case the vessel involved in the incident is operated by a land-based crew are: 1) from which location is the incident caused, and 2) by whom is it caused? The answers to both questions have ramifications for the analysis of the application of the unified maritime convention system.

If the land-based crew is within the owner’s company structure, the place of operations of the land-based crew will qualify as venue for lawsuit against the owner.(1) See e.g. Brussels Regulation 2012, Art. 7(5), Danish Code of Civil Procedure, § 237, German ZPO, § 21.Provided the control center is situated outside the flag state, the seat of the remote-controlling branch would therefore provide for an additional venue. If, on the other hand, the task is outsourced to a third party by the owners, the entity operating the land-based crew will provide an additional or alternative defendant, both for claims in contract and also for claims in tort connected with the operation of the vessel.

The answer to the question of the continued application of the maritime conventions in the case of a (fully) remotely controlled vessel will largely depend on: 1) if there are reasons that the land-based crew should be considered differently from the other servants of the carrier, and 2) in the case of outsourcing, whether the entity operating the land-based crew itself qualifies as an entity to which the maritime regulations apply.

3.1.c Autonomous vessels / vessels operating in autonomous mode at the time of the incident: A shift towards product liability?

The new autonomous/AI-based technologies have been argued as reducing or removing the human element from the chain of causation, thereby reducing the risk involved in the activity of shipping. Realistically, though, rather than making shipping risk-free, the new technologies will mean a shift in potential human errors from the incident-stage to the production, retrofitting or maintenance stage. Human errors must still be expected to occur regarding e.g. the coding of the automated/ autonomous AI software; the production of the hardware it uses; its interaction with other systems; its installation into the vessel, in case it is retrofitted; its maintenance, including running the updating of the system, or as regards when to use it, in cases where the vessel is not fully autonomous, but is set to autonomous mode at the time of the incident. Maritime incidents may therefore occur, not because the vessel was operated incorrectly, but because of an inherent risk in the vessel as such. Those risks are generally better dealt with under rules aimed specifically at dealing with such issues, i.e. the rules on product liability.

The delimitation between what is a question of product liability and what is a question of the liability of the owner under the maritime regime is not a clear one, and certain facts may – dependent on the circumstances – be covered by both regulations. Still, guidance may be given by considering whether the fact in question is most naturally placed within the producer’s sphere of liability or within the owner’s/ carrier’s sphere of liability. Below, in Fig. 4., a rough grouping of such risks has been attempted.

The original coding of software and production of the hardware required to run it are tasks that – together with the construction of a vessel in general – fall outside the sphere of application of the maritime conventions, as such tasks, considering their nature, do not fall “…within the shipowner’s ordinary course of business or area of expertise.”(2) T. Solvang, ”Shipowners’ vicarious liability under English and Norwegian law”, MarIus No. 571, p. 38f and 45ff.Maintenance of AI-systems already installed in the vessel, on the one hand, as well as the day-to-day use of such systems, on the other hand, both seem to fall squarely within the owner’s obligation to uphold the seaworthiness of the vessel and to operate her correctly. The integration of the AI system with other systems onboard or ashore, as well as the physical installation of the system onboard in cases of retrofitting, may however provide for overlap where both the rules of product liability and the owner’s obligation of seaworthiness may be triggered. In such cases, the claimant will normally be entitled to decide which regulation to rely on. As rules on product liability are not unified,(3) V. Ulfbeck, ”Produktansvarsskade i transportretten”, DJØF 2007, p. 17 & 107.this may result in an increased number of incidents being dealt with outside the maritime unified system and may therefore challenge the robustness of the maritime convention system.