Tuesday, March 28, 2017
Mark Geistfeld has posted to SSRN A Roadmap for Autonomous Vehicles: State Tort Liability, Automobile Insurance, and Federal Safety Regulation. The abstract provides:
The vast majority of motor vehicle crashes are now caused by driver error. Autonomous vehicles will eliminate driver error and prevent thousands of fatalities and serious bodily injuries, making a compelling safety case for policies that foster the rapid development of this technology. Considerable technological advances have occurred over the past decade, but there is widespread concern that the rate of development is hampered by substantial uncertainty about potential manufacturer liabilities for the crash of an autonomous vehicle. This uncertainty is compounded by apparent variations in the requirements of state tort law across the country, which make it even more difficult for manufacturers to assess their liability exposure in the national market. The considerable uncertainty and potential patchwork of state laws has prompted calls for the federal safety regulation of autonomous vehicles.
The uncertainty stems from the complexity of driving behavior. Autonomous vehicles will not be perfectly safe; in this respect, they will inevitably fail at times. Given the complexity of driving and the inherent limitations of coding that behavior, how can courts reliably determine whether such a failure — the crash of a fully functioning autonomous vehicle — subjects the manufacturer to tort liability?
Framing the liability problem in this manner masks one central behavioral factor that considerably simplifies matters. Autonomous vehicles will transform the individualized behavior of human drivers into a collective, systemized form of driving. In effect, the entire fleet will be guided by a single driver — the operating system comprised of the hardware and software that determines how this class of autonomous vehicles executes the dynamic driving task. Systemized driving is properly evaluated in relation to the performance of an entire fleet of autonomous vehicles with the same operating system, an inquiry that is often quite different from the case-by-case analysis of crashes involving conventional motor vehicles.
When a crash was caused by the fully functioning operating system, the autonomous vehicle was engaged in systemized driving performance that should be evaluated with aggregate driving data for the fleet. Regardless of the particular circumstances of the crash, the programming or design of the operating system would be reasonably safe under widely adopted rules of products liability if the aggregate, premarket testing data show that the autonomous vehicle performs at least twice as safely as conventional vehicles. To avoid liability for the crash of a fully functioning autonomous vehicle, the manufacturer must also adequately warn consumers about this inherent risk. Once again, the risk involves systemized driving performance, and so aggregate driving data provide the appropriate measure. Based on these data, auto insurers can establish the risk-adjusted, annual premium for insuring the vehicle. By disclosing such a premium to consumers, the manufacturer would satisfy its obligation to warn about the inherent risk of crash, eliminating this final source of manufacturer liability for crashes caused by a fully functioning autonomous vehicle.
The collective learning of state tort law can then inform federal regulations governing the reasonable safety of automated driving technologies. The foregoing analysis is based on tort rules that have been widely adopted across the country. States that do not follow the majority approach might reach different conclusions. To ensure that manufacturers face uniform obligations within the national market, the National Highway Transit Safety Administration could adopt two federal regulations that clearly fit within its recently announced regulatory approach, each respectively derived from the associated tort obligations concerning adequate premarket testing and disclosure of the inherent risk of crash. These regulations would largely retain the role of tort law, because regulatory compliance would also satisfy the associated tort obligations in most states. The regulations would promote the federal interest in uniformity in a manner that minimizes the displacement of state tort law, thereby optimally solving the federalism problem.
The federal regulations would be supplemented by state tort law in important instances, yielding a comprehensive regulatory approach. Within this legal framework, a regulatory compliant autonomous vehicle would subject the manufacturer to tort liability only for crashes caused by malfunctioning physical hardware (strict products liability); malfunctions of the operating system due to either programming error (same) or third-party hacking (strict liability again, with an important caveat); or the manufacturer’s failure to adequately warn about safe deployment of the vehicle (an ordinary products liability claim). A manufacturer would also be subject to tort liability for not complying with the federal regulations (negligence per se).
By eliminating driving error and otherwise reliably performing in accordance with system standards, autonomous vehicles will not subject manufacturers to tort liability. Autonomous vehicles can be regulated in a manner that ensures reasonably safety without impeding the rapid development of this life-saving technology.