The adoption of autonomous vehicles will have profound legal consequences for insurance claims and liability laws. As vehicles with increasingly sophisticated autonomous driving capabilities proliferate, industry experts expect liability for accidents to gradually shift from human operators to the manufacturers which design, build, and sell such cars. In the short term, insurance will likely continue to cover drivers as the primary holders of fault, but as drivers gradually cede control of their vehicles to software, manufacturers will assume greater responsibility under strict liability theories for faults in their transportation products. As the legal landscape of insurance laws changes on a state-by-state basis, there will likely be greater calls for federal-level regulatory action and perhaps even for legislation to create nationwide standards.
Overview of Autonomous Vehicle Adoption
According to current predictions, autonomous vehicles will roll out in five stages and gradually transition away from human control.1 Level 1 autonomous vehicles will have a single autonomous function, such as automatic emergency braking or lane-keeping functions. Level 2 vehicles will have at least two automated functions, but will still require substantial human input. Level 3 vehicles will handle many driving tasks with occasional human intervention required. Level 4 vehicles will be officially driverless in certain environments, such as freeways or urbanized areas. Level 5 vehicles will operate entirely without drivers, and will have no need for pedals, steering wheels, mirrors, or other artifacts of human control. As the level of driver control over their vehicles steadily reduces, so too will their fault for accidents.
The adoption of autonomous vehicles will likely sharply reduce accident rates2 because human error accounts for 90% or more of vehicle accidents.3 To address this, Level 1 technologies are hitting the market in increasing numbers: Nissan recently announced plans to equip many of its most popular vehicles with automatic emergency braking starting in 2018. The industry predicts that adoption of Level 1 autonomous vehicle technologies alone will profoundly reduce the occurrence of accidents. This prediction is already being borne out: Tesla’s introduction of partial self-driving functions has reduced accidents for models equipped with that technology by nearly 40%.4
Changes in car ownership patterns will also have a large impact on the car insurance industry. Car ownership, and the number of car owners requiring automotive insurance, will likely decline as the American population becomes increasingly urbanized. Access to better public transit and ride-sharing technology would obviate the need to own personal vehicles for some segment of the population. In contrast, the availability of automotive transport for minors, the elderly, and the visually impaired may increase. Some predict that the current mixture of privately-owned vehicles, cars for hire, and public transportation vehicles on American roads will be replaced by fleets of autonomous vehicles owned and operated by the manufacturers of such vehicles. Still others predict municipalities will purchase and operate fleets of vehicles themselves.5 In any event, experts predict the number and type of car owners requiring automotive insurance will decline.
Current Liability Laws Applied to Autonomous Vehicle Manufacturers
Before changes in car use and ownership patterns occur, however, autonomous vehicle manufacturers will continue to hold strict liability for defects in their products. As it currently stands, product liability doctrines provides some shield against claims arising from accidents caused by faulty autonomous driving technology. There are three categories of defects for which manufacturers may be held strictly liable.6
1. Liability Doctrines
In the first category, manufacturing defects, if a plaintiff proves the autonomous vehicle departed from its intended design even though all possible care was exercised in its preparation and marketing, then the manufacturer can be held liable. In general, plaintiffs in such actions must prove that the product that injured them did not conform to the manufacturer’s own specifications. For claims of manufacturing defects in the physical components of the car, this area of liability laws is unlikely to see substantial change: manufacturers will continue to be liable for defects in the physical construction of their vehicles, such as flawed raw materials or erroneous assembly. For claims of malfunction in the software algorithm controlling the car, manufacturing defect claims will be more difficult to prove: courts do not apply manufacturing defect doctrine to software because nothing is actually manufactured.7
In the second category, design defects, a plaintiff must prove that the design of the vehicle or algorithm was itself defective.8 Courts will either perform a consumer expectations test or a cost-benefit analysis. The consumer expectations test weighs the consumer’s reasonable expectation as to the safety of his vehicle. The introduction of autonomous vehicle technology will further complicate the application of this test, as some consumers will have unrealistic expectations of their autonomous vehicle’s capabilities. Under a cost-benefit analysis, however, courts will balance the societal benefits of introducing autonomous vehicle technology against the cost to manufacturers of developing and installing safer technology. It will not be difficult for a manufacturer to meet this burden, as a reduction in accident rates of even a few percentage points would result in hundreds of lives and billions of dollars saved.
In the final category, “failure to warn,” plaintiffs must allege that the manufacturer’s warnings of the risks of using an autonomous vehicle did not provide adequate warning. This will likely become an area of substantial litigation for autonomous vehicles, since it is unclear how precise and direct a manufacturer’s warnings must be. The timing of warnings further complicates this issue: as cars become networked and capable of continuous communication with servers, manufacturers might be expected to send warnings nearly the instant they become aware of design flaws.
2. Product Liability Defenses
Depending on the jurisdiction, four defenses to product liability for autonomous vehicle manufacturers exist.10
First, a manufacturer can claim comparative negligence on the part of the driver. If the manufacturer proves the driver was operating the vehicle in a negligent manner, it can avoid or reduce its liability for a product failure. However, the primary function of autonomous vehicle systems is the removal of control from the driver, who no longer needs to pay attention to the road. It makes little sense to claim in litigation that a driver was negligent in not monitoring road conditions, and courts are unlikely to allow such a defense, particularly for Level 5 fully-autonomous vehicles.
Second, a manufacturer can claim the plaintiff was misusing the vehicle. If the plaintiff misuses the vehicle in an unforeseeable manner, the manufacturer can avoid liability. However, paying no attention to the road while Level 4 and 5 autonomous systems are in control of the vehicle will no longer be a misuse of autonomous vehicle technology. Nevertheless, misuse could be an available defense when the driver makes modifications to the autonomous systems that prevent them from working as the manufacturer intended.
Third, manufacturers can invoke the “state of the art” defense, which will likely be more availing than the first two defenses. When curative measures are not technologically or scientifically possible, manufacturers are not liable for accidents caused by defects in their products. An autonomous vehicle manufacturer could claim that autonomous vehicle technology was simply incapable of avoiding the accident in question. This could be a particularly potent defense for cutting-edge autonomous driving technology, especially when the product in question is the software algorithm controlling the vehicle. Manufacturers can easily assert that assessing new risks at the time the algorithm was written was technologically infeasible.
Lastly, manufacturers can claim that drivers assume the risk of accidents once they consent to the use of an autonomous car. Like any new technology, autonomous driving has inherent risks and imperfections, and so users of this technology can be found to have assumed the risk of accidents when opting to use it. “Assumption of risk” is thus a potent defense to liability claims against autonomous vehicle manufacturers, though not all jurisdictions allow it.
Short-Term Insurance Implications
As autonomous vehicles are adopted, strict liability for manufacturers will exist alongside driver liability. Nearly every state has adopted strict liability for products, or has arrived at the same result through warranty laws.11 As detailed above, product liability holds manufacturers responsible for defects in their products, and autonomous vehicles will be no exception. Some manufacturers have already accepted strict liability for crashes caused by their vehicles, while others have begun selling insurance as part of the price of their vehicles.12
Strict liability for manufacturers may discourage the introduction of autonomous vehicle technology. Alternatively, manufacturers may pass the cost of their increased liability exposure on to consumers. However, strict liability provides incentives to manufacturers to provide the safest product they can and reduce the danger of their products as much as they are capable.
In the short term, there will likely be growth in insurance markets, as three lines of insurance become prominent.13 According to a report from Accenture and the Stevens Institute of Technology, cybersecurity insurance will become a major new insurance market, covering remote theft, unauthorized entry, hacking, privacy breaches, and theft or misuse of personal data. Product liability insurance will also continue to grow and include in its coverage communications failures, software glitches, and the usual hardware breakdowns. Lastly, the report predicts that public infrastructure coverage will become a small but multi-billion dollar insurance market, defending claims against cloud server systems managing traffic and road networks, external sensors and signals and other system-level issues.
At the individual level, insurers may initially offer discounts for drivers of autonomous vehicles in order to encourage adoption. As autonomous vehicle technology gradually takes over driving functions, insurers may rely on telematics devices (“black boxes”) to monitor driver activity and ensure that no human took control of the vehicle prior to the accident.14 Some insurers already offer discounted rates to drivers who opt in to black box monitoring, but these are unpopular due to privacy concerns. In any event, telematics devices may no longer be necessary once autonomous vehicles manufactured without steering wheels or pedals become the norm.
Long-term Insurance Implications
1. State-Level Reform
In anticipation of liability shifting away from drivers and toward manufacturers, states have begun instituting long-term reform. States have been especially active in passing legislation enabling the testing and use of autonomous vehicles. At present, eighteen states have passed legislation and four governors have issued executive orders relating to autonomous vehicles.15 Like vehicle codes, insurance is state-regulated.
State liability laws typically take one of two forms.16 In a no-fault regime, insurers pay the injured party regardless of who caused the accident. These regimes have the advantage of quickly compensating victims and obviating the need to prove fault. Because this type of insurance regime does not require the assignment of legal fault, it may become more popular as the question of fault becomes obfuscated by the introduction of autonomous technology.
In states that regulate their insurance industries according to tort law, payments are predicated on driver liability for operating a vehicle in a negligent or abnormally dangerous manner. Negligence liability will become a murky and highly contentious area as drivers cease bearing full responsibility for control of their vehicles. For negligence inquiries, the operative question remains whether or not the driver took reasonable care in the operation of the vehicle. However, the use of autonomous driving technologies will cause societal shifts in what is seen as reasonable: it will be “reasonable,” for instance, to take one’s eyes entirely off the road when traveling in a Level 4 or 5 autonomous car.
For drivers who operate their vehicles in an “abnormally dangerous” manner, however, strict liability applies. This area of the law will be subject to dynamic change as society adopts autonomous vehicles. Because custom defines “abnormally dangerous” activities, their definition will change with time: operating an autonomous vehicle before their use has become widespread will at first be considered abnormally dangerous, but when autonomous vehicles are widely adopted their nonuse will be considered abnormally dangerous.
2. Federal-Level Reform
Widespread adoption of autonomous vehicles and new insurance challenges may lead to calls for greater federal regulation of auto insurance. The federal government can preempt state laws to establish a national insurance standard, although it has been largely content to leave regulation to state governments and will meet resistance on federalism grounds if it chooses to act. Federal action could include legislation or regulation through agencies.
If the federal government were to intervene in automobile liability laws, it could pass legislation authorizing its agencies to promulgate national regulations of autonomous vehicles. The National Highway Traffic Safety Administration (NHTSA) will likely be the predominant source of any federal regulation of autonomous vehicles, but it has thus far promulgated very little such regulation.17 If NHTSA adopted comprehensive regulation, manufacturers would likely argue state tort law claims are preempted. Regulations authorized under federal legislation have been found to impliedly preempt state tort law in previous lawsuits against the automobile industry, including litigation over a manufacturer’s failure to install air bags.18
In the absence of regulatory action, the federal government could well pass legislation limiting automotive industry liability generally.19 Some precedent exists for federal action limiting industry liability through legislation and treaty, including the Nuclear Industries Indemnity Act and the accession to the Warsaw Convention of 1929 limiting airline liability. More rarely, Congress will create comprehensive no-fault compensation schemes for manufacturers providing highly valuable public goods. For instance, when the vaccine industry was faced with a potential inundation of lawsuits over vaccine injuries, Congress passed the National Childhood Vaccine Injury Act, providing a mechanism for no-fault compensation of vaccine injuries and preempting lawsuits over those injuries. If the provision of autonomous vehicles comes to be seen as a public good, the government may protect it through legal limits on liability and subsidies through a no-fault compensation scheme.
The gradual adoption of autonomous vehicles with increasingly sophisticated driving functions will likely be met by patchwork changes to liability laws at the state level, and perhaps eventually federal action. As autonomous driving becomes standard, driver liability for accidents will gradually decline, replaced by greater claims of manufacturer liability. In the meantime, insurance coverage will rise to meet new demand in covering liability for autonomous systems and infrastructure. One thing is certain: liability laws and insurance coverage will be profoundly different when all the cars on the road drive themselves.
The authors would like to recognize summer associate Tom Kinzinger for his contributions in drafting this article.