Insuring Automated Vehicles – An Analysis of Travelers Institute White Paper

August 16, 2018


Autonomous vehicles (“AVs”) are constantly in the news. A range of technology companies, such as Google and Uber, as well as automotive manufacturers, are leading the development of vehicles with autonomous features. The technology promises improved convenience and safety but questions have naturally arisen about how to insure vehicles under computer control (we will refer to the computer controlling a vehicle as the “autopilot”). Properly insuring AVs and individuals operating AVs is essential to ensure the development and use of the technology.

Travelers has joined a host of other insurers, consultants, attorneys and think tanks by publishing a whitepaper on insuring autonomous vehicles. Insuring Autonomy, How auto insurance can adapt to changing risks, Travelers Institute White Paper (July 2018), available at (the White Paper). Traveler’s White Paper includes two key proposals: (1) that traditional tort liability and automobile insurance will be able to efficiently handle risk transfer for accidents involving automated vehicles; and 2) auto insurance for AVs will need to include a cyber-insurance component. This blog addresses Traveler’s proposals on assigning tort liability and insuring loss caused by an AV’s autopilot, as well as Traveler’s acknowledgment that auto policies in the future will need to cover cyber-risks, such as hacking.

AV Technology

The Society of Automotive Engineers (SAE) has designated six levels of automation for automobiles, numbered 0 through 5. Level 0 AVs have no automation; level 5 AVs require no input from the operator except entering the destination. Level 3 AVs are just arriving on the market, with Audi and Cadillac introducing vehicles in some areas with advanced autopilot functions that allow hands-free operation under certain conditions, albeit with constant human monitoring of the function. Level 4 vehicles will offer the option of full automation or human control; level 5 vehicles will be driven only by the autopilot. The timeline for the entry of Level 4 and 5 AVs into the market is uncertain.

AVs will make transportation more convenient because AV owner/operators will be able to put driving time to other uses. In addition, AV technology promises to significantly decrease accidents. Human error is causes 94% of all motor vehicle accidents in the United States. Replacing a human with a computer, experts believe, will reduce the frequency and severity of accidents.

The Travelers’ White Paper discusses why the auto insurance industry should remain relevant with respect to AVs. This is in reaction to a number of commentators, including Warren Buffett, who have predicted a vastly diminished role for auto insurers as vehicle safety improves.

We doubt the need for auto insurance will disappear. AV owners, like every other auto owner, will face risk from collision and weather hazards, among others. Up to the point where vehicles are entirely automated, there will be a need for liability coverage for the driver. Even when level 5 AVs enter the market, owners will need liability insurance to cover them in the event their faulty maintenance or neglect of a vehicle leads to an accident, and to cover loading and unloading of a vehicle.

Liability for Accidents Caused by the Autopilot

When an autopilot is in control of an AV and an accident occurs, difficult insurance questions arise. Most states’ tort systems assign liability to the person or persons responsible for an accident. The insurer of the party at fault pays, subject to a reduction if the plaintiff is found to have contributed to his or her own loss and the comparative or contributory negligence scheme in place. When a human is not controlling a vehicle, it is unclear how liability should be apportioned and how that liability will be insured. There are no answers yet.

Assuming that the owner of the AV properly maintained the vehicle, the autopilot was engaged within its design parameters and the accident was caused by the autopilot, how should liability be apportioned, and what insurance should respond to the claim? Travelers argues that traditional risk transfer methods (that is – auto liability policies and the traditional tort system) will be sufficient to handle these situations. Travelers notes that traditional auto insurance has the advantage of familiarity to consumers and flexibility to adapt to changes brought on by AVs. Travelers, however, offers no specifics on how the risk transfer would occur in these circumstances. Travelers also suggests the auto industry will need to modify its insurance products, but does not recommend any specific revisions.

Other commentators have suggested that the manufacturer of the vehicle should insure loss causes by an AV autopilot as a products liability claim. This is logical, as the autopilot is the automaker’s product, and the failure of that product has caused injury. Travelers points out, however, that products liability insurance is not designed to quickly compensate victims, and the ponderous nature of products liability litigation will delay compensation for victims of autopilot caused accidents. On the other hand, Travelers does not explain how traditional auto insurance will result in faster compensation for victims when the automaker’s product causes the accident.

The Rand Corporation has suggested a number of schemes for allocation of fault in the event of an AV accident caused by the autopilot, including a no-fault system. In the alternative, Rand suggests a federally mandated no-fault compensation fund, similar to the National Childhood Vaccine Injury Act. This act protects vaccine manufactures from liability for injuries resulting from vaccines. Both schemes offer the advantage of speedy compensation for victims of accidents. No-fault insurance, however, has not proved popular among the states, and a number of states that adopted no-fault schemes have abandoned them. A no-fault compensation fund may prove difficult to reserve, at least initially. There is little data on AV accident rates to support an actuarial analysis. Although AVs are expected to be safer, a chance also exists that accident rates could rise for a time, while drivers learn to use the technology and undiscovered engineering issues are uncovered. It is likely states would also resist federal intervention into auto insurance, which states have traditionally regulated.

A possible compromise between traditional auto liability and products liability would be to treat the autopilot as an insured driver under the owner’s liability policy, or under a separate policy issued by the automaker on the vehicle. Along these lines, Mercedes, Volvo and Google’s affiliate Waymo have reportedly agreed to accept full liability for vehicle accidents while their vehicles are using automated technology. The scope of these automaker’s acceptance of liability is unclear. It is also unclear if these companies intend this acceptance of liability to provide a permanent solution to AV liability. This scheme, however, would likely promote rapid resolution of AV accident claims. It might also allow automakers to cap their liability at the policy limits set for the autopilot.

Travelers’ argument in chief is that “auto insurance should play the same primary risk transfer role in that world as it does now for non-AVs.” The White Paper at 13. We agree role for traditional auto insurance, including liability insurance, will exist for owner/operators of AV. The introduction of fully autonomous autopilots, however, will require a significant modification of traditional auto policies to account for liability attributable to the autopilot. As noted above, insuring the autopilot as an additional driver is one potential solution.


Travelers notes that AV owner/operators will need cyber-insurance. Travelers’ focus, however is on data protection. While insuring owner/operator data is important, insurers such as Travelers need to focus on another potential risk. The hacking of a vehicle or other cyber-security threats puts owner/operators, passengers and others at risk of suffering bodily injury or property damage. Several years ago, “white-hat” hackers demonstrated the ability to hack into a vehicle and control it remotely. The ability to take control of a vehicle suggests the ability to steal the vehicle or cause an accident. Clearly, a need exists for insurance coverage to protect drivers, passengers and third-parties from property damage and bodily injury resulting from the hacking of an AV.

The insurance industry, however, has sought to avoid insuring property damage or bodily injury arising from cyber-risks. Most cyber-insurance policies we have examined specifically exclude coverage for bodily injury and property damage. Many standard general liability insurance policies bar coverage for property damage or bodily injury that result from the loss of use of or corruption of electronic data – which insurers will argue bars coverage for a hacking losses.

We are not aware of any auto policies that exclude coverage for loss resulting from damage to, corruption of or inability to access data, similar to CGL policies. The first auto policy to insure the owner/operators of AVs, issued by Adrian Flux in the U.K., offers only limited coverage for hacking, however. If other auto insurers offer only limited coverage for cyber-risks, there could be a significant gap in protection from hacking that causes loss to AV owners/operators.

Coverage for automobiles is compulsory and highly regulated. To adequately insure AV owner/operators from cyber-risks, traditional auto policies must evolve. Auto insurers must affirmatively insure against bodily injury and property damage that result from the hacking of an AV’s computer systems. To guarantee that the insurance industry adequately covers cyber-risks associated with AVs, state insurance commissioners should take the lead in drafting model legislation to adequately protect AV owners and the general public.

AVs hold the promise to significantly improve road safety. For AVs to succeed in the market place, consumers must have confidence that appropriate insurance exists to protect them from risk, including cyber-risk, or they will refuse to purchase AV technology. Accordingly, if cyber-risks are to be managed through traditional auto insurance policies, it is critical that insurers offer adequate coverage for this risk.