Liability Rules for Autonomous Vehicles: How Traditional Legal Relations Encourage Modern Technological Innovation

Innovation and Economic Growth and Artificial Intelligence

Article Snapshot


Richard Epstein


Law and Economics Seminar, Stanford Law School, January 16, 2020


As autonomous vehicles (AVs) come into use, lawyers will guide courts’ choice of liability rules, the sanctions that apply when the rules of the road are broken. The best rule for many accidents involving AVs will be strict liability.

Policy Relevance

Traditional liability rules will serve well in AV accident cases. Regulators should avoid an overcautious approach that delays the benefits of AVs.

Main Points

  • Historically, liability rules have been stable over time, even though the technological transformation from stagecoaches, to railroads, to motorized vehicles was massive; by contrast, the rules of the road must change in response to innovation.
  • Only one set of liability rules is needed to cover all cases where harm has occurred in the past; however, in designing regulations to prevent future harm, different systems are needed to support innovation (for example, no one would apply the same licensing rules to drivers and to nuclear power plants).
  • As AVs come into use, the rule of the road will change, but virtually no change is needed in liability rules; traditional liability rules did not depend on technology and are consistent with all patterns of technological advance.
  • A negligence rule would require those responsible for AVs to use reasonable care; this rule would be impossible to administer, requiring examination of millions of lines of code to determine if any were not carefully written.
  • When harm occurs in an AV accident between strangers (parties with no contractual arrangement) the best liability rule is strict liability; however, if the injured party knew of the danger, they should take reasonable measures to avoid harm.
  • When harm occurs in an AV accident involving consensual contractual arrangements, strict liability is sometimes the best rule, but not always; for example, in determining liability for medical mishaps, the doctor’s disclosure of risks to the patient is often adequate.
  • Those who fear harm from road accidents in the future cannot identify the likely perpetrators in advance, so licensing is necessary; common law courts avoid enjoining activities with a low risk of harm, but complex projects might require a regime that includes public inspections and liability insurance.
  • Regulatory agencies might err by seeking to resolve all uncertainty about AVs early on, fixating on potential harms and blind to AVs’ potential benefits.

Get The Article

Find the full article online

Search for Full Article