Automated Vehicles and the Product Liability Act of South Korea

The focus of the global automobile industry is on developing self-driving vehicles in addition to electric vehicles. According to the Global Autonomous Vehicle Market Status Report 2024–28, published by the market research firm Technavio, the autonomous vehicle market is expected to reach USD974.5 billion in 2028 at an annual growth rate of 58.78%.

Below, the authors examine the main issues related to automated vehicles.

Safety of automated vehicles

According to the US National Highway Traffic Safety Administration (NHTSA), the frequency of airbag deployment for cars equipped with autonomous driving (autopilot) mode is one per 1.3 million kilometres, whilst for cars not equipped with autopilot mode the rate is substantially higher, at one per 0.8 million miles. It is estimated that the rate of fatal collisions for automated cars would be more than 40% lower than that for non-automated cars, and even early-stage automated cars are expected to be more than 50% safer than non-automated cars. In particular, 93% of non-automated car accidents are caused by human-related factors such as negligence. The foregoing statistic suggests the potential for the automated driving system (ADS) in automated cars to dramatically reduce accident rates, hence empowering the view in favour of restricting and eventually banning human driving.

As automated car development projects approach completion, it is expected that countries will designate and expand regions and roads open only to automated cars. In this process, difficulty is likely to arise in establishing autonomous driving infrastructure in the midst of mixed traffic between automated and non-automated driving cars. In response to the mixed traffic, it has been proposed that separate road lanes be implemented for automated and non-automated cars.

Six stages of automated driving systems

The US NHTSA distinguishes six levels of automation.

• Level 0: no automation – the driver performs all driving tasks.
• Level 1: driver assistance – whilst the driver controls the vehicle, such as by steering, accelerating or decelerating and hitting the brake pedal, some assistance to driving is provided by the advanced driving assistance system (ADAS).
• Level 2: partial automation – the ADAS controls the steering, acceleration/deceleration and brake pedal, and the driver assists the ADAS and monitors the environment.
• Level 3: conditional automation – the ADA is engaged in driving tasks, including monitoring the environment, but the driver must be ready to take immediate control of the vehicle in case of emergency.
• Level 4: high automation – the ADS is capable of operating the vehicle completely autonomously within limited service areas, and the driver is not needed other than, for instance, to set the destination and route.
• Level 5: full automation – the ADS is fully autonomous and capable of performing all driving functions under all conditions without the driver. Consequently, level 5 self-driving cars do not require a steering wheel or a brake pedal; the human “driver” becomes only a passenger.

Generally, a car is effectively classified as an automated vehicle starting at level 3, which requires the driver’s intervention in emergency situations whilst the ADS mainly controls the driving. To illustrate, Hyundai Motor Company of Korea is currently testing the prototype of a level 4 automated vehicle, but its commercialised automated cars equate only to level 2 automation. This gap likely owes to insufficient risk assessment for vehicles with automation above level 2.

Technology used by automated vehicles

Most automated vehicles currently under development are equipped with artificial intelligence (AI), such as the ADS. The ADS is evolving into an autonomous system that interconnects with other vehicles to exchange traffic information and judges the traffic environment, going beyond a self-contained system.

The technologies that constitute the ADS are as follows:

• the ADAS;
• wireless connection (Wi-Fi, Bluetooth, 5G mobile communication);
• vehicle-to-everything (V2X) communication – precise location measurement and high-definition 3D mapping; and
• deep-learning to adapt to various driving environments via the Drive Data Platform (DDP) and experience with different driving environments.

Causes of accidents involving automated vehicles

The following causes have been identified for automated vehicle accidents:

• defects in the ADS and vehicle equipment or devices;
• operational defects in the electronic control unit (ECU) software that is responsible for controlling the ADS and driving-related devices/functions;
• errors in V2X communication or the DDP, or the exchange of incorrect information with other vehicles or road infrastructure systems;
• failures of traffic lights, roads and facilities for automated driving;
• natural disasters – eg, radio interference due to sunspots; and
• hacking of automated vehicles and related systems.

Given these various possible causes, it is difficult to precisely pinpoint a single, specific cause of automated vehicle accidents, which can be caused by a combination of factors.

Liability issues for automated vehicles

Drivers

“Driver” refers to the human being who sits in the driver’s seat and operates the vehicle. For automated vehicles, determining the degree of duty of care owed by the driver would be the central issue.

It is expected that the same duty of care currently in force would apply to drivers of automated vehicles with a maximum of level 2 automation (partial automation).

For automation levels exceeding level 3, it is expected that the degree of duty of care would be lower. For level 3 (conditional automation) and level 4 (high automation) automated vehicles, the driver is likely to be found to have breached the duty of care if the accident is caused by his or her failure to take measures despite a warning to switch to manual driving mode.

For level 5 (full automation) automated vehicles, it would be unnecessary or irrelevant to consider the driver’s duty of care.

Operators

For non-automated vehicles, “operator” is a legal concept that has been used to refer to a person who is not actually driving the vehicle but is effectively in control of it and benefits from the driving thereof. An operator is legally obligated to retain liability insurance for driving, and examples include the vehicle’s owner or a lessee who is entitled to use the vehicle. In the case of car leases, both the lessor and lessee are considered operators.

In the imminent era where the use of automated vehicles will dominate, it is likely that rental car companies will enter into contracts with various parties for the shared use of automated vehicles owned by the rental companies much more frequently. Through these contracts for the shared use of automated vehicles, the relationship between the rental car companies and the counterparties would be akin to a service contract, rather than a rental or lease agreement. This is because a lessee has been deemed as an operator for having de facto control of the vehicle, but it may be farfetched to apply the same reasoning when the use of automated vehicles is based on a contractual relationship in the form of a service agreement.

Manufacturers of automated vehicles

An increase in automated driving is likely to be accompanied by a higher rate of manufacturing defects in automated vehicles as the cause of accidents. Due to the emergence of automated vehicles, manufacturers are becoming the most important parties with respect to the issue of liability; in the past, the manufacturers of non-automated cars were rarely concerned with liability.

The general consensus is that software is not considered a “product” under the Product Liability Act of South Korea; however, that a product equipped with software is a “product” is undisputed. In this sense, there would be no question that automated vehicles equipped with AI would also be considered a “product” under the Product Liability Act of South Korea.

The manufacturers of automated vehicles may be unable to avoid liability for damages that could have been avoided if a reasonable alternative design had been used. In the early stages of use of automated vehicles, it will be difficult to develop reasonable alternative designs. In this sense, it appears that it would at first also be difficult to hold the manufacturer liable for manufacture or design defects, but as the use of automated vehicles becomes dominant, manufacturers may find it extremely difficult to avoid liability on the basis of the lack of a reasonable alternative design.

Traffic AI system management

As discussed above, automated vehicles receive traffic information from a traffic AI system that allows for safe self-driving. However, if an accident occurs due to errors in the traffic AI system, the injured party may seek to recover damages from the central or regional government, because an error in the software used by the traffic AI system is more likely in this case than a physical defect in the road, etc.

Automated vehicles in the context of the Product Liability Act of South Korea

Burden of proof

The most debated issue for automated vehicles in the context of the Product Liability Act of South Korea is the burden of proof in proving the defect itself, and the causal relationship between the defect and the loss.

The South Korean Supreme Court determined the following in a case involving the explosion of a television that was being used normally:

It is extremely difficult to perfectly prove, both scientifically and technically, the defect and its causal relation to the alleged damages. In the event of an accident or explosion of a television in normal use, the consumer is required only to prove that the incident occurred within the manufacturer’s exclusive domain of control and would not usually occur without a defect in the product. If the consumer proves these circumstantial facts, it is presumed that the product had a defect, unless the manufacturer proves that the incident occurred due to a cause other than a defect in the product. It is reasonable and in accordance with the principle of damage compensation that the consumer bears the lighter burden of showing the defect that caused the incident (Case No 98Da15934).

In a case involving the explosion of a battery, the Seoul Central District Court followed the same logic adopted in the foregoing Supreme Court judgment, as follows.

Even if it is not proven that the battery explosion was caused by a specific kind of defect in the battery, the fire occurred when the battery was used normally in accordance with the manufacturer’s instructions and while under the defendant’s exclusive domain of control; therefore, it is reasonable to presume that the battery was defective, not having the safety, durability or performance that it should normally have. Hence, the defendants must prove that the fire was caused by a factor other than the defective battery (Case No 2020Na78249).

State-of-the-art defence

An automated vehicle is a product posing a high risk of harm, as an accident may potentially harm not only the driver but also the drivers of other cars and pedestrians. This high risk is especially highlighted by the fact that an error in the software may render it completely impossible for the driver to control an automated vehicle. In this regard, it would be reasonable to apply the highest level of scientific and technological standards for the state-of-the-art defence of an automated vehicle.

The South Korean Supreme Court determined the following in respect of the state-of-the-art defence in a case involving victims of defoliants during the Vietnam War.

If the manufacturer could not have discovered the defect in the product based on the science and technology available at the time of manufacture, it is deemed that the manufacturer could not have foreseen the accident unless there were special circumstances; thus, the manufacturer would be freed of liability for the defective product. However, this case involves Vietnam War veterans who were repeatedly exposed to a 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD)-containing poisonous defoliant, and it was thus foreseeable that the defoliant would harm the life or body of the veterans. The manufacturer breached the high duty of care to prevent such damages to the veterans, and on this basis, the court cannot accept the defendants’ state-of-the-art defence (Case No 2006Da17539).

Insurance in the context of automated vehicles

As automated vehicles become commercialised, it is expected that, in the long-term, the overall number of car accidents will decrease, and that there will be fewer policyholders as car-sharing becomes more common. For the time being, however, it is likely that, as automated vehicles co-exist with non-automated vehicles, any decrease in the accident risk or insurance premiums is unlikely to be rapid. Rather, it is likely that insurance companies will increase insurance premiums or refrain from selling insurance policies in the short term due to the lack of data on accidents involving automated vehicles.

The commercialisation of automated vehicles requires changes in the insurance system. Under the current system, car owners are obligated to purchase liability insurance, likely owing to the car owners’ civil liability in car accidents. The situation will be much different when automated vehicles are commercialised and issues of product liability become central in disputes. In this respect, it is likely that car manufactures will be compelled to carry liability insurance from the standpoint of product liability, similar to the current situation where car owners are obligated, as operators, to purchase liability insurance. This may lead to a world where car manufacturers begin to engage in insurance business or sell cars and insurance together as a package.

As automated vehicles become mainstream, there will be a long-term need to discuss the establishment of a new compensation system for car accidents. For instance, the government may establish a public compensation fund to indemnify victims of automated vehicle accidents, where the operator of the fund would pursue a recourse claim against the party liable for the accident. This approach of applying the framework of public insurance in place of the traditional framework of tort liability for personal injury cases may be an innovative way to obviate the difficulties and inefficiencies faced by individual litigants aiming to prove the cause of an accident.