This article was written by Mark Schaub(partner), Atticus Zhao(senior associate).
The development of advanced automated vehicle safety technologies, including fully self-driving cars, will be the greatest change to personal transportation since the introduction of the personal automobile nearly a century ago.
Although humans love cars they do not seem to be particularly adept at driving. Surveys have found that some ninety percent of motor vehicle crashes are caused at least in part by human error. Accordingly moving the driving from humans to autonomous vehicles may improve safety and save lives.
As with every ground breaking technology, automated vehicle technologies have the potential to bring great social, economic and environmental benefit but will also lead to new legal issues that will need to be considered.
Perhaps the most basic and most important is in respect of liability. As driving functions become increasingly automated there will be a shift of responsibility from driver to vehicle. The question, will then arise as to who bears responsibility if there is a collision occurs - the driver? The owner? The manufacturer? The automation system provider? Or the insurance company? Or do they share liability and if yes then to what extent?
This article seeks to provide some preliminary views on liability issues relating to automated vehicle technologies and how this is likely to develop in China.
In the first part, we will consider some of the latest legislation internationally in respect of liability for automated vehicle technologies. In the second part, we will consider the responsibility of driving on the basis of SAE level of automation. In the third part, we provide an overview of the liability regime of China and how it may decide to apportion liability in the future.
Current legislation or regulatory efforts in certain leading countries
Automated vehicle technologies are developing at a break neck speed.In his TED Talk in April 2017, Elon Musk, the CEO of Tesla, said that he believed that the technology behind fully autonomous vehicles is only "about two to three years away".
The legislative response has not kept pace with the development of the automated vehicle technologies. However, some jurisdictions notably Germany, the United States, the United Kingdom and Australia have enacted laws or issued guidelines to seek to deal with the legal challenges the new technologies will bring. One core issue in these legislative efforts is to deal with the issue of liability.
Home to some of the world's largest car manufacturers such as Audi, BMW, Mercedes-Benz, Porsche and Volkswagen and also to many of the world's most innovative auto-suppliers - Germany is at the forefront of developments in self-driving technology.
In 2015, German government devised a Strategy for Automated and Connected Driving (the "German Strategy"), for the purpose of maintaining Germany's leading role in the field of automotive innovation.
On 12 May 2017, the German parliament approved the amendments to the German Road Traffic Act (Strassenverkehrsgesetz, StVG) and allowed automated driving as of 21 June 2017 ("New Law"). Germany is one of the first few countries to enact a law that allows automated driving.
It is important to note that the New Law does not change general liability concepts under German law. Both driver and owner remain liable even if the vehicle is in automated driving mode but drivers are able to avoid liability if they are found to have lawfully used the automated driving mode. Under the New Law, automated vehicles must be equipped with a "black box" to identify whether it was the driver or the system which had control at the time of an accident. This will assist to determine whether the vehicle caused the accident which is relevant under German product liability rules.
The key provisions of the New Law are as follows：
- Highly or fully automated vehicles are defined. The New Law allows for the registration and use of vehicles providing a highly or fully automated driving system. The new definition does not include so-called "autonomous vehicles".
- The use of automated vehicles is permitted within the limits of the intended use as defined by the individual car manufacturers. The system must inform the driver if his/her use exceeds the limits of intended use (e.g. if a driver leaves the driver's seat during automated mode).
- The New Law does not require drivers to remain focused on the road at all times, but they must be able to react "without undue delay" if the system prompts them to do so, or if they themselves realize this is necessary. 
- The New Law raises the maximum liability limits under German Road Traffic Act from Euro 5 million to Euro 10 million for death or injury and from Euro 1 million to Euro 2 million for damage to property.
Under the New Law, if the vehicle is in automated driving mode then the driver and not the car manufacturer may be held directly liable. The driver may have recourse against the manufacturer but he/she will be the initial party to be held liable. In addition, it is worth noting that under the German Road Traffic Act the owner of the vehicle is subject to strict liability i.e. the owner's liability is established not based on fault but based on ownership of the vehicle that caused the accident. Accordingly, if the owner assumes liability, he or she may be able to take recourse against the manufacturer.
2. United States
The United States and its technology giants are also playing a leading role in the development of automated vehicles.
In September 2016, the National Highway and Transportation Safety Administration (NHTSA) issued an updated guidance for automated vehicles ("Automated Vehicles Policy"). In the Automated Vehicles Policy, NHTSA makes it clear that the US Department of Transportation (DOT) and Federal government will be responsible for regulating motor vehicles and motor vehicle equipment whereas the states will be responsible for regulating the human driver and most other aspects of motor vehicle operation including vehicle insurance and liability.
On 19 July 2017, a US House panel reached bipartisan agreement in respect of major aspects of legislation to address the testing and deployment of autonomous vehicles. The panel aims to clarify the federal and state roles for regulating highly automated vehicles. The bill would retain the states' authority to regulate, inter alia, registration, licensing, insurance and liability for highly automated vehicles.
The hotchpotch of different jurisdictions in the United States will mean that state tort law will play a central role in resolving liability disputes arising out of crashes and liability will vary depending on the state (i.e. some states base auto liability on traditional negligence, others on no-fault liability and some on strict liability). In addition to auto related liability it is very likely that the United States' fearsome product liability regime will also likely play a role with respect to liability for automated vehicles.
To date only a handful of states have enacted new bills or amended current laws to clarify liability issues in respect of automated vehicles. Michigan, no doubt influenced in large part to its auto manufacturing base, enacted Senate Bill No.633 on December 27, 2013, which precludes manufacturers from bearing any liability for alleged damages unless the defect from which the damages resulted was present in the vehicle at the time of manufacture. Furthermore, manufacturers will not be liable under product liability for damages resulting from the modification of equipment installed by the subcomponent system producer to convert a vehicle to an automated vehicle unless the defect from which the damages resulted was present in equipment when it was installed by the subcomponent system producer. In 2017, Tennessee which like Michigan has a strong auto sector (including Nissan, GM, VW and over 900 auto suppliers) drafted similar provisions in its House Bill 1131, which is still pending.
Massachusetts which has no auto industry to speak of since 1989 has drafted a somewhat different bill in 2017.The Massachusetts bill stipulates that manufacturers shall assume liability for incidents in which the automated driving technology is at fault. On the other hand New York has opted for a much wider ambit of liability and holds every manufacturer, owner and operator of the unmanned motor vehicles strictly liable and responsible for death or injuries to person or damage.
Accordingly, already at this early stage it is clear that different US states will approach liability very differently. Car making states like Michigan and Tennessee will try to limit liability against manufacturers; Massachusetts law will hold manufacturers liable when at fault whereas New York seeks to adopt strict liability to hold everyone liable for an accident involving an unmanned vehicle.
3. United Kingdom
Despite Brexit looming it seems the UK legislature has been able to take time out of its busy schedule to address issues in respect of automated vehicles.
On 22 February 2017, the Vehicle Technology and Aviation Bill (the "VTA Bill") introduced new insurance rules for self-driving cars.
Under the VTA Bill, insurers would be primarily responsible for paying out damages arising out of accidents caused by automated vehicles. The VTA Bill proposes to extend compulsory motor insurance requirement to include automated vehicle owners. This means whether a car is controlled by a human driver or is in automated mode the victims will be able to follow the same route for remedy. The victim will have a direct right against the motor insurer and the insurer in turn will have a right of recovery against the responsible party to the extent there is liability under existing law, including under product liability law.
The VTA Bill lays out how the liability for accidents involving automated vehicles should be assigned and includes the following key terms:
- The UK government is responsible for preparing and keeping up-to-date a list of all automated vehicles. Only listed vehicles are considered 'automated vehicles' and will be subject to new insurance and liability provisions.
- Insurers will, by default, be liable for damage arising out of accidents caused by 'automated vehicles' in self-driving mode if the vehicle is insured at the time of the accident.
- Insurers have a right of recourse to recover the cost of the pay-out from the manufacturer.
- The amount of an insurer's liability may be reduced if an injured party is also wholly or partially responsible for the damage caused. Insurers are not liable if an accident involving an automated vehicle has been caused by the person in-charge's negligence to allow the vehicle to drive by itself when not appropriate.
- Insurers are not liable for damage resulting from accidents caused by automated vehicles if the vehicle is not insured. In such cases the owner of the vehicle will be liable.
- Insurers are entitled to exclude or limit liability for damage caused by automated vehicles resulting from unauthorized or prohibited alterations to the vehicle's operating system by the insured owner.
- Exclusions or limitations on liability are allowed if the insured owner of an automated vehicle has failed to install software updates to the vehicle's operating system that their insurance policy requires to be installed. 
The VTA Bill indicates the UK government position is to ensure a victim can always claim compensation from the insurance company even if the car acted autonomously. If the manufacturer is found to be liable, the insurer will be able to pursue a subrogated claim against the manufacturer under existing common law and product liability arrangements and recover their costs from such manufacturer.
Under the VTA Bill, insurers essentially act a bridge between manufacturer and the victim. The victim in an accident involving automated vehicles will need to claim against the insurers and not directly from the manufacturer. From a liability apportionment perspective, according to the VTA Bill, if an accident occurs where an automated vehicle is in self-driving mode and is not due to the fault of the operator or owner then the manufacturer is ultimately liable.
The Australian government has also been considering how the development of automated vehicles in Australia will play out. The Australian government has recognized the need for a nationally consistent regulatory framework that while embracing innovation also ensures automated vehicles are safe.
Since 2015, the Australian government has started to identify whether there are any regulatory barriers associated with the introduction of more automated road and rail vehicles in Australia. In November 2016, one of the key policy bodies for road and vehicle use in Australia - the National Transport Commission (NTC) - issued a Policy Paper on regulatory reforms for automated road vehicles ("Policy Paper").
NTC stated in the Policy Paper that assigning fault is likely to become more complex in accidents involving automated vehicles. This added complexity was primarily ascribed to the following factors:
- complexity of the automated vehicle operating environment – multiple parties could be responsible for an accident
- continued interaction between human and machine – it is likely that some automated vehicles will require humans to take over the driving task at different times and other automated vehicles will likely require the human driver to monitor the automated driving system.
- new kinds of accident causes – although human error may be reduced automated vehicle may have accidents caused by new threats such as cybersecurity breaches, software bugs or failing sensors
The Policy Paper concluded that no changes are suggested in the near term for the current Australian liability regime for automated vehicles, since the current liability framework is believed to be sufficiently adaptive to accommodate automated vehicles. However, it does stress that liability risks could prevent the deployment of automated vehicles, or severely reduce their functionality or scope of operations, if manufacturers become excessively cautious.
Another reason the Policy Paper mentions that no legislation is recommended for the time being for Australia is that the complexities in respect of liability will likely increase as highly and fully automated vehicles emerge and it is likely to be premature to seek to legislate in such regard.
The Policy Paper also indicates that the greatest uncertainty in relation to liability is the issue of who is in control of a vehicle with different levels of driving automation. As such, if the meaning of 'driver' and 'control' can be clarified then the greatest point of uncertainty in relation to liability is likely to be solved. Clarifying who is the driver and who is in control of the vehicle will require access to relevant data.
An initial analysis of driving responsivity based on SAE level of automation
The SAE definitions for levels of automation have been widely recognized and adopted. The below table sets out a high-level summary of the SAE definitions as to the levels of automation:
|Human monitors driving environment|
|0||No Automation||the full-time performance by the human driver of all aspects of the dynamic driving task, even when enhanced by warning or intervention systems|
|1||Driver Assistance||the driving mode-specific execution by a driver assistance system of either steering or acceleration/deceleration using information about the driving environment and with the expectation that the human driver perform all remaining aspects of the dynamic driving task|
|2||Partial Automation||the driving mode-specific execution by one or more driver assistance systems of both steering and acceleration/deceleration using information about the driving environment and with the expectation that the human driver perform all remaining aspects of the dynamic driving task|
|Automated driving system monitors the driving environment|
|3||Conditional Automation||the driving mode-specific performance by an automated driving system of all aspects of the dynamic driving task with the expectation that the human driver will respond appropriately to a request to intervene|
|the driving mode-specific performance by an automated driving system of all aspects of the dynamic driving task, even if a human driver does not respond appropriately to a request to intervene|
|the full-time performance by an automated driving system of all aspects of the dynamic driving task under all roadway and environmental conditions that can be managed by a human driver|
A better understanding of the different levels of automation helps determine who is in control of the vehicle and therefore responsible for driving. This in turn helps to establish liability in the case of an accident. Below is a preliminary analysis of the driving responsibility based on the different level of automation:
1. Level 0 - 1
Level 0 means vehicles with no automation at all. The driver takes the full control of the vehicles.
Level 1 refers to driving assistance. Under such circumstances, human drivers are still taking control of most of the driving tasks and the automated system can only assist the driver in certain situations. The driver will still be fully responsible for driving.
2. Level 2
Level 2 is defined as partial automation. Vehicles with automation Level 2 still require drivers to be attentive and monitor the driving environment all the time. Drivers should take back control of the vehicles at any time when they realize or should realize the automated driving system can no longer handle the current driving situation.
While the automated system in certain driving modes can take control of the vehicle completely, human drivers should keep monitoring the driving environment and are obliged to intervene at any time. Accordingly in such cases, the driver should remain responsible for driving. One exception may be that if a driver has fulfilled his or her obligations to monitor the driving environment and an accident occurs solely because of the malfunction of the automated system.
3. Level 3
Level 3 refers to conditional automation. The difference between Level 2 and Level 3 is that Level 3 vehicles not only take control of steering and accelerating/decelerating but can also monitor the driving environment in certain circumstances. In such cases human drivers only need to take over as the automated system requires. This distinction is significant and is often seen as the breakthrough between driving assistance and automated driving.
It is clear that if a driver fails to respond to the system's request to take control of the vehicle, then the driver will be responsible. Accordingly, if the system has not required the driver to take control and an accident occurs while the vehicle is under the control of the system, the manufacturers will likely be held responsible for the accident. Drivers may be imposed on more duty to monitor the environment. For example, under the New Law of Germany, if drivers realize or should realize that the automated system is not functioning safely (even if the system does not give any instruction or warning) then drivers are required to take back control of vehicles.
Naturally, life will also throw up far more complex situations in which liability will be difficult to settle. For example, if the driver reacts to the system's request, but the accident occurs nevertheless. Who should be responsible for the accident - the driver or the manufacturer? The key to answering this question may depend on whether the automated system provided the driver with reasonable time to react. If the driver has been provided with reasonable time before an accident occurs but fails then the driver may be liable – if time not reasonable then the manufacturer would bear the liability. The determination of "a reasonable time" itself may be a challenging technical issue and will very likely depend on the circumstances – a reasonable time on a highway may be very different from a driveway.
4. Level 4
Level 4 means high automation. Under this driving mode, vehicles could reach a state of autonomous driving subject to limitations in certain conditions. In this level it is the driving system that drives the car rather than the human. Consequently, there generally should be no fault of the operator (passenger), unless he or she intentionally disturbs the automated system. In such cases it would be reasonable to shift the entire responsibility completely away from the human drivers to the manufacturers.
However, as automation Level 4 is still subject to certain conditions it should be noted that human driver control will still be required when the vehicle is operated beyond the permitted driving condition. As such, if an accident occurs as a result of unpermitted use of the driving system then the human drivers should still bear primary responsibility for the accident.
5. Level 5
Level 5 applies to full automation, which is the highest level of automated driving. Basically the vehicle can drive itself and it has passengers not drivers. The most notable difference between Level 4 and Level 5 is that at this stage, the automated vehicle must be able to deal with all conditions that a vehicle may face. Accordingly, the driving responsibility will shift from human drivers to manufacturers.
Exceptions to the responsibility allocation at this level could be that an accident is caused by hacking or other improper intervention by a human. Cyber security issue is likely to be a significant threat to safe driving of autonomous vehicles. To prevent non-authorized external access (e.g. hacking) to the autonomous vehicles, manufacturers must "guarantee the secure encryption of data and communications". Manufacturers must ensure that there is sufficient protection against manipulation and misuse both of the technical structure and of the data. If manufacturers have fulfilled their obligations, however, and hackers nevertheless invade into the automated driving system and cause an accident, the hackers rather than the manufacturers will bear primarily responsibility for the accident (in both civil and criminal aspects).
In summary, so long as human drivers control steering, acceleration, deceleration and will need to react to emergencies either alone or in conjunction with the automated driving system then it is likely the human drivers will be primarily responsible for the driving (basically automated systems from Level 1 through 3). However, if the human driver's role in respect of the control of the vehicles has been largely or completed passed on to the automated driving system then the responsibility and liability for driving will likely shift from drivers to manufacturers.
What will China Do?
1. Current liability regime
Currently, China's liability regime is based on tort and product liability. In the event of a road accident, the driver/owner that caused the accident will be held liable under PRC Road Traffic Safety Law ("Traffic Law")  and PRC Tort Law. If the accident is caused due to defects of the vehicles then the manufacturer or seller of the defective vehicle will be liable.
China mainly embraces three types of liability, namely at-fault, presumed-fault and strict liability. As per the China Traffic Law, in cases of an accident between motor vehicles then at-fault liability will apply. On the other hand if there is an accident involving a motor vehicle and a driver of a non-motor vehicle or a pedestrian then presumed-fault or strict liability will apply.
Under the Traffic Law, an insurance company of the owner of the vehicle which caused the accident will first pay damages to the victim(s), at an amount capped by compulsory third party liability insurance. If the damages paid by the insurer are not sufficient to cover the losses suffered by the victim(s) under the compulsory third party liability insurance, the driver/owner shall pay further damages to the victim(s) for the difference.
2. Possible liability apportionment for automated vehicles
Naturally China's liability regime was established well before anyone was thinking of introducing automated vehicles.
China's view is more likely to reflect Germany's or Michigan's views. China has taken steps to embrace automated vehicle technology but has not yet tabled a substantial legislative response to deal with new legal conundrums stemming from the use of automated vehicles. It is likely that China as a major auto manufacturing power would prefer to adopt a "wait and see" approach to ensure the government does not regulate too early. Premature regulations may create artificial barriers that could retard growth and development of these emerging technologies. Naturally, regulating too late may result in safety-issues but it is likely that existing liability and product quality laws could address these concerns until developments are clearer.
It is clear from above that different jurisdictions will regulate liability differently. Although there will be different models such as the UK's "single insurer model" it is clear that the trend will be that responsibility and liability for accidents will gradually shift from the human to the manufacturers. This trend will be as human transitions from actual driver to passenger. China's future allocation of liability would also reflect this shift in responsibility.
Accordingly, it may be practical for China to adopt an approach similar to Germany. That is, regulating new legal issues of automated vehicles within the current liability regime where the driver/owner of the automated vehicles (or the owner's insurer) will be primarily liable for accidents and the driver/owner or the owner's insurer may subsequently take recourse against manufacturers in case of manufacturer's liability.
We expect that insurance will play a critical role in resolving the complex liability issues involving automated vehicles. Insurers may act as the middlemen for manufacturers, whereby victims can gain quick access to compensation and insurers can develop processes to handle subrogated claims quickly and easily in the background. However, legislative actions will likely be required to better assess risk and such measures may include regulating the registration of automated vehicles (to ensure safe access to market by automated vehicles) and requiring the installation of a black box in automated vehicles to identify whether the driver or the system was in control at the time of an accident.
Page 5, Federal Automated Vehicles Policy issued by US National Highway Traffic Safety Administration (NHTSA) in September 2016.
SAE International Standard J3016, issued by SAE International
Hopefully not a poor choice of words
Para.1, page 3, Strategy for Automated and Connected Driving
Section 7 of German Road Traffic Act
Page 38, Federal Automated Vehicles Policy, DOT, September 2016.
The draft is seen to be a significant federal legislation aimed at speeding self-driving cars to market
Page 15, Products Liability and Driverless Cars: Issues and Guiding Principles for Legislation, John Villasenor, April 2014.
1961 PA 236 Sec.2949b (1).
1961 PA 236 Sec.2949b (2).
2017 Ma H 3422 of Massachusetts, which is still pending.
2017 NY A 7243, which is still pending.
Sections 1-5 of Part 1 of the Vehicle Technology and Aviation Bill and https://www.out-law.com/en/articles/2017/february/new-uk-laws-address-driverless-cars-insurance-and-liability/
This could be an issue for anyone who does not regularly update their mobile phone software!
Page 59, Policy Paper
Page 59, Policy Paper
Page 62, Policy Paper
Page 60, Policy Paper
SAE International Standard J3016
Section 1b para.2, German Road Traffic Act
Page 22, Strategy for Automated and Connected Driving
Article 76 of the Traffic Law
Article 48 of the Tort Law
Articles 41 and 42 of the Product Quality Law and Article 12 of Interpretation on Several Issues Concerning the Application of Law in the Hearing of Cases Involving Compensation for Damages in Road Traffic Accidents issued by China Supreme Court 27 November 2012
Chinese scholars have different opinions on which theory of liability is adopted by the PRC Traffic Law in this regard.
Article 76 of the Traffic Law
E.g, "Made in China 2025", Auto Industry Mid and Long Term Planning and Establishment of National Standards System of Telematics Industry
Note: Special thanks to our intern Shan Wenyu for his contribution to this article.