Self-Driving Cars in Belgium: A Clash Between Innovation and Liability?
By Steven De Schrijver and Olivier Van Fraeyenhoven of Astrea
"As a matter of fact, innovation is ultimately bound by existing legal principles. Hence, the completely self-driving car project might have to be put “on hold” if the current legal framework does not offer sufficient legal certainty."
Imagine you start your car, configure your GPS, then sit back, relax and wait until you arrive at your requested destination. Indeed, the self-driving vehicle is a hot topic nowadays. Take for instance the Google self-driving project or Tesla’s recent announcement of its newest software update, with which their vehicles would be able to drive on their own. It can surely be predicted that the popularity of the self-driving car will only increase in the future.
Although innovation should be applauded, car manufacturers themselves are fully aware of the potential stumbling blocks that might hinder the launch of a completely self-driving car. With this article we tend to explore, from the perspective of the driver as well as that of the car manufacturer, whether and, if so, how Belgian law deals with liability issues in the case of self-driving cars. As a matter of fact, innovation is ultimately bound by existing legal principles. Hence, the completely self-driving car project might have to be put “on hold” if the current legal framework does not offer sufficient legal certainty. Below, we will first discuss the liability of the driver and the product liability of the car manufacturer. Subsequently, we will suggest a solution that might remedy some of the current shortcomings of the liability regime.
The Liability of the Driver
The driver’s liability is laid down in the Belgian Road Traffic Code on the one hand and, on the other hand, in some provisions of the Belgian Civil Code. Article 8.1 of the Belgian Road Traffic Code clearly indicates that “every vehicle [in motion] must have a driver”. Further, the applicable provisions of the Belgian Civil Code for tort are Articles 1382 and 1383. Again, these provisions contain a direct reference to a human person (“every act of a human person (…)”). An alternate option could be to rely on Article 1384, which contains liability for “things under custody”, but this would not change much as the human person would still be liable for an act that has been committed by his or her car (ie, the thing under his custody).
It is evident that the liability regime of the driver does not offer much scope for new technologies such as the self-driving car. The current regime makes the introduction of completely self-driving cars rather impossible, as this technology is not in accordance with the law. It therefore even seems illegitimate to introduce a completely self-driving car. This is a pity, as it has generally been acknowledged that self-driving cars would be the safest cars, because of their incredibly fast reaction speed, something a human does not have. Self-driving cars, for instance, offer systems of pedestrian detection, automatic parking, lane monitoring and vehicle-to-vehicle communication, by which cars can communicate with each other and exchange speeds with one another in order to avoid accidents. However, as the current regime does not distinguish between both, Tesla, for example, took its precautions by informing all their vehicle owners that they still need to keep their hands on the steering wheel. This essentially means that the driver is de iuri still in charge. This might, in turn, trigger a counterproductive consequence that drivers are not willing to use the self-driving software: drivers will not only be responsible for their own human mistakes, but might also be liable for mistakes of the car due to, for instance, software defects. However, regarding these software defects, it is important to stress that the car manufacturer can be held liable on the basis of product liability, which we will discuss further below.
The Product Liability of the Car Manufacturer
The liability of the car manufacturer for his products includes both a civil and a penal liability. The penal liability is incorporated in Article IX.2 of the Economic Law Code with the obligation for the car manufacturer to exclusively introduce safe products on the market. In this regard, it must be highlighted that if an unsafe product was introduced on the market, the government could not only prosecute the company of the car manufacturer, but also its directors. Further, the civil liability regime, established on a no-fault basis, is contained in the Belgian Product Liability Act of 25 February 1991 (hereinafter “the Act”) which is the implementation of the Council Directive of 23 July 1985 on liability for defective products (hereinafter “the Directive”). Article 1 of the Act holds that “the producer shall be liable for damage caused by a defect in his product”.
The car manufacturer, as the main producer, will be liable in the first place if the self-driving car acts in a way that could not reasonably be expected by the driver, since the defective product concerns the car as a whole. Alternatively, the driver could address the software implementer directly, in his capacity as a software producer. The fact that software rather concerns an “immaterial” moveable, which seems to fall outside the scope of Article 2 of the Act, does not appear to cause problems. It has generally been accepted that if software is contained in a tangible medium, it could be considered a “material” movable, in accordance with Article 2 of the Act. However, turning to the software implementer will only happen in very rare circumstances. We will therefore adopt the approach that the car manufacturer is fully liable in case of a defect.
Article 5 of the Act establishes the ultimate condition that the product in question needs to be “defective”. A product can be considered defective “when it is not offering the safety that a person might expect from it”. Emphasis has been placed on the representation of the product, the normal and foreseeable use of the product and the moment the product is put into circulation.
In the case of self-driving cars, the representation of the product, together with the legitimate expectations the vehicle owner might have within the framework of normal and foreseeable use, might be a stumbling block for the car manufacturer. As mentioned above, self-driving cars are generally portrayed as being among the safest products, since they have the ability to react a lot faster to unexpected traffic situations. Presenting your product in such a way might trigger product liability quite easily when eventually the car reacts in a way that was not expected by the driver. The driver, for instance, presumes that within the context of the normal and foreseeable use of the car, the car will react properly on all sorts of “foreseeable” circumstances that might happen in traffic (eg, unexpected items on the road, a driver that does not give right of way to the right). In other words, there are many “foreseeable” circumstances that might be somewhat unforeseeable for a self-driving car, and on which it might react differently than expected by a normal driver. This of course means that car manufacturers must bear a high liability risk. Tesla, for instance, tried to hedge this risk by announcing that their software is still in a “beta-phase” and that the hands of the driver should be kept on the steering wheel at all times. The crucial question then, of course, is whether this “disclaimer” is sufficient to hedge the risk as you can never be sure people will comply with your disclaimer. We might indeed reasonably expect that people will comply during the testing period, but what will happen afterwards?
As a result, although self-driving cars have the ability to be among the safest vehicles in the world, car manufacturers are somewhat reluctant to introduce them properly as they have to bear a high liability risk, both civil and criminal. In his paper, Schellekens calls this “the chilling effect of liability law”, as the delay in introduction is mainly due to a liability regime not able to cope with innovation. According to Schellekens, car manufacturers will only introduce self-driving vehicles when “they meet a higher standard”. Then the question is: what standard can be considered high enough?
Based on the foregoing, it seems that there are still a lot of open questions on the matter of self-driving cars. The driver’s liability regime is unadjusted to introduce completely self-driving cars and the product liability risk the car manufacturer has to bear is rather harsh, which makes the introduction of these cars less probable. We would therefore like to suggest some solutions that hopefully have the potential to have more self-driving cars on our highways in the very near future.
Proposal to Adjust the Current Framework
A first necessary intervention is the adaptation of the Road Traffic Code, which currently contains the obligation that every vehicle needs to have a human driver in charge. As a result, the Road Traffic Code leaves no room for completely self-driving cars. Only intermediate models, in which the car can do some tasks on its own but where ultimately the driver is in charge, are achievable. This needs to change in order for completely self-driving cars to be in accordance with the law.
As the product liability regime indeed puts a high burden on the car manufacturers, a possible solution could be to launch the completely self-driving software in successive stages, thereby relying on a risk-based approach. This would then mean that car manufacturers would, for example, first launch the completely self-driving car in traffic situations where it is plausible that the drivers will behave in a similar way, such as on highways. In a second stage, car manufacturers could, for instance, expand to secondary roads, where the risk for deviating behaviour is somewhat higher, but not as high as in busy city centres. This way, the car manufacturer’s liability risk is lowered to a certain extent without stifling innovation. Furthermore, after completion of all the different stages, we expect self-driving cars to be quite advanced, and as a result, the risk to be borne by the car manufacturer would be rather low, as few (or no) accidents would occur by then. In addition, insurance also plays a large role in this matter, and it will need restructuring as well with a view to self-driving cars.
We therefore suggest tailoring the current driver’s liability regime to self-driving cars so that completely self-driving cars would not be in breach of the law. Our analysis also showed that the product liability risk of the car manufacturer is quite burdensome. We would therefore propose a solution that introduces the self-driving car in successive stages. In that regard, it would be possible for the car manufacturer to move on to subsequent stages gradually once the car provides a standard that can be considered “high enough” for that specific stage. However, it cannot be denied that the liability regime is not the only issue. Many other factors need to be reconsidered as well if the completely self-driving car makes its debut on the market, such as whether and how the current road infrastructure and traffic rules need to be reviewed, and in what way the actual insurance policies should be adjusted. In other words: it is clear that much still remains to be done in order for the completely self-driving car to conquer the world.