Automobile Accidents Without a Driver?: The Insurance Liability of Highly Automated Vehicles

By: Ashley Mitchell

I. Introduction

In March 2018, an Uber fatally struck Elaine Herzberg.[1]  This accident was extraordinary because the driver was not operating the vehicle at the time of the crash; it was in autonomous mode.[2]  This tragedy raises many questions about the future of cars without drivers.  For now, Uber has removed self-driving vehicles from the roads.[3]  But as we look to the future of cars without a driver, we must ask: what is the insurance liability when a crash occurs?

Every 5 seconds there is an automobile accident and ninety-four percent are caused by human error.[4]  Auto accidents can cause property damage that is costly to repair, and bodily injury, which generates medical bills.[5]  Auto insurance covers the cost of property damage and bodily injury and is mandated in forty-nine states.[6]  Highly automated vehicles (HAV) are designed to eliminate drivers and rely on an automation system, subsequently reducing human errors and the number of auto accidents.[7]  Yet, it is unclear if insurance will cover the cost of damages or if the manufacturer will be held liable when a highly automated vehicle is involved in an accident.  Legal liability is one of the biggest obstacles to widespread utilization of HAV technology.[8]

HAV’s goal is to improve the operator’s quality of life.  In September 2016, federal guidelines were introduced to help legislators rethink insurance schemes and to call for states to develop uniform HAV policies.[9]  States must decide who—owner, operator, passenger, manufacturer, etc.—must carry motor vehicle insurance.[10]  The role of the federal government is to incentivize, and if needed, to direct the states to develop uniform policies regarding HAVs.[11]  For the time being, the guidelines are not binding and cannot be used in court to demonstrate reasonableness.[12]

The goal of this Note is to create a set of policies that lawmakers can use to balance public safety with industry liability concerns.  Progress in states like Michigan and Arizona provides evidence that clear and comprehensive guidelines for managing manufacturers’ expectations encourages the growth and use of HAV technology.[13]

II. Background
A. Definitions

A HAV is an automobile that relies “on the automation system (not on a human) to monitor the driving environment.”[14]  The human who operates the automation system is called an operator and “is not responsible for the driving task, but is still responsible for certain aspects of the journey (i.e., inputting a destination for the vehicle).”[15]  A driver is “the human operator of a [HAV] when it is not operating in a fully automated mode.”[16]

A crash is “an unintended event resulting in fatality, injury or damage to a vehicle or property, involving one or more motor vehicles, on a roadway that is publicly maintained and open to the public for vehicular travel.”[17]  Auto insurance operates on the premise that a crash is caused by a driver.  However, HAVs have operators, not drivers.

B. Insurance Coverage

Today there are two auto insurance schemes: tort liability and no-fault.[18] State legislatures determine whether a state will be a tort liability state or a no-fault state.[19]  Tort liability insurance mimics a negligence tort.[20]  The elements of a negligence tort are: the wrongdoer must have a duty to the victim, the wrongdoer breached that duty to the victim, and the breach caused the victim’s damages.[21]  In the automotive context, a driver has a duty to take reasonable care in the operation of his or her vehicle.[22]  If the driver does not take reasonable care and crashes, then the driver has breached a duty to the victim and caused the victim’s damages.[23]  The driver who does not take reasonable care and crashes is the wrongdoer in a tort liability state. The wrongdoer is the “at fault” driver.[24]  In a tort liability state, “a victim recovers compensation from another party (or, almost invariably, the other party’s insurer) if the victim can show that the other party was at fault for the accident.”[25]

When a crash occurs, if fault is not easily determined, the trier of fact must declare which driver was at fault.[26]  Generally speaking, drivers do not go to court; they subrogate their claims to their insurance companies.[27]  Subrogation is “the substitution of one party for another whose debt the party pays, entitling the paying party to rights, remedies, or securities that would otherwise belong to the debtor.”[28]  The litigation process—research, witness prep, and evidence review—takes time and money.  Moreover, the court system has a backlog of cases that are waiting to be presented.  Therefore, subrogation is the usual manner of determining the driver at fault.

Conversely, the no-fault state does not assign liability for the crash.  There is neither an at-fault driver nor a victim.  No-fault insurance pays for the damages after an accident even if the insured is responsible for the accident.[29]  No-fault insurance reduces the volume of litigation, lowers court costs, and pays damages quickly because once the crash occurs, a subsequent determination of fault is not needed for damages to be paid.[30]

Self-insured retention insurance (SIR) is “a defined dollar amount specified in a liability insurance policy that must be paid by the insured before the insurance policy will respond to a loss.”[31]  Manufacturers that have a SIR in their liability policies are required to maintain a minimum dollar amount “X” available to pay damages if the manufacturer is found liable for an HAV crash.  When product liability claims are greater than X, the liability policy pays the remaining damages.[32]

C. Product Liability

Product liability law is a set of rules and guidelines that regulate the production of goods and their intended use.[33]  When a product does not perform its intended use, manufacturers are liable to consumers.[34]  Product liability is a tort.  A potential plaintiff must make a prima facie showing that the plaintiff was owed a duty of due care which was breached by the defendant and caused the plaintiff’s injuries.[35]  The plaintiff has the burden of proving that the product was defective.[36]  The defendant has the burden of proving alternative designs were not feasible and the benefits of the design outweigh the risks.[37]  There are three categories of product liability: (1) manufacturing; (2) design; and (3) failure to warn.[38]  As manufacturing processes improve and standardized techniques become more widespread, incorrect assembly defects are less likely to occur.[39]

There are two tests to determine if a design was defective: (1) consumer expectations; and (2) cost benefit/risk utility.[40]  The consumer expectation test is applied when the minimum safety assumptions are common knowledge of a lay juror.[41]  When applying the test, courts consider:

A product defect in design or formulation when it is more dangerous than an ordinary consumer would expect when used in an intended or reasonably foreseeable manner. Moreover, the question of what an ordinary consumer expects in terms of the risk posed by the product is generally one for a trier of fact.[43]

“Some courts, for example, use the term ‘reasonable consumer expectations’ as an equivalent of ‘proof of a reasonable, safer design alternative.’”[44]

The cost benefit/risk utility test is a balancing test that is applied when minimum safety assumptions are beyond the scope of common knowledge of a lay juror.[45]  “In such cases, the jury must consider the manufacturer’s evidence of competing design considerations [citation], and the issue of design defect cannot fairly be resolved by standardless reference to the ‘expectations’ of an ‘ordinary consumer.’”[46]  Courts weigh the benefits of the utility provided by a particular design against the costs or risks associated with the design.[47]

A failure to warn product liability claim alleges a hidden defect that the manufacturer had an obligation to warn consumers of the danger, but failed to.[48]  Manufacturers must consider all the ways HAV technology could be misused, and then publish a warning.  As it stands, there is no precedent on what constitutes a reasonable warning for HAV technology.

III. Analysis
A. Insurance Coverage

The tort liability auto insurance regime is not compatible with HAVs because HAVs have operators, not drivers.[49]  Tort liability insurance compensates victims of a crash after the “at fault” driver is declared.  So when a HAV crashes, who is the “at-fault” driver?

A no-fault insurance regime resolves the “at fault” driver problem, but avoids the fact-finding that occurs during litigation—which is problematic in the HAV context.  Parties involved in an HAV crash and the subsequent litigation would investigate the operation of the automation systems that monitor the driving environment to determine what caused the crash.  The investigation would uncover where the malfunction in the automation system occurred.  For example, a HAV failed to interpret and respond to the driving environment when it failed to detect a white trailer blended in with the horizon and failed to apply the brake resulting in a fatal crash.[50]  Even though the victims would receive damages in a no fault state, the malfunction in the operating system could go unnoticed for a longer period of time, without litigation and the accompanying investigation regarding fault.  A failure to diagnose and rectify the malfunction in the operating system could result in more accidents.  Thus, when considering insurance schemes, legislatures should balance compensating crash victims and encouraging public safety.

B. Product Liability

Manufacturers are responsible when an HAV crashes because of monitoring system failure.  The intended use of the HAV is to carry its passengers from start-point to end-point without any crashes.  The systems that monitor the driving environment are responsible for safely transporting passengers.  The HAV has not performed its intended use when there is a crash caused by the failure of the systems monitoring the driving environment, therefore allowing consumers to sue manufacturers under product liability theory.[51]  If manufacturers are concerned liability will be too costly, they will forego production of HAV technology.

The consumer expectation test and the cost benefit/risk utility test used in design defect cases will hinder the development of HAV technology.  There is no precedent for determining reasonableness in the context of HAVs, but there is a standard for assessing reasonableness in the context of non-HAV crashes.  In Pannu v. Land Rover North America, the California Appellate Court stated that the reasonable consumer expectation test applies when the vehicle is being used in a customary way and fails.[52]  Under these circumstances, an ordinary consumer can form a minimum safety expectation.[53]  The Court applied the test and found that a vehicle driven on a freeway was being used as intended, and a freeway crash was reasonably foreseeable.  Therefore, the ordinary consumer could form minimum safety expectations.  When a HAV is being used as intended and the resulting damage is reasonably foreseeable, the ordinary consumer can form minimum safety expectations.

Yet, the ordinary consumer’s minimum safety expectations could still result in a high level of liability for manufacturers.  Recall the crash that occurred after a HAV failed to detect a white trailer against the horizon.[54]  After the accident, Tesla released the following statement: “[n]either autopilot nor the driver noticed the white side of the tractor-trailer against a bright lit sky, so the brake was not applied.”[55]  If the human operator failed to distinguish the white side of a tractor-trailer, can a human program an automation system to distinguish the white side of a tractor-trailer?  There was an expectation that the Tesla autopilot feature was designed to pilot the car from start-point to end-point without a crash.  The Tesla was being used as intended, and it was reasonably foreseeable that a white tractor-trailer would turn in front of the Tesla.  The Tesla was unable to prevent the crash, thus manufacturers could be held liable for the accident.  Expected high levels of liability could deter manufacturers from adopting HAV technology.[56]

The cost benefit/risk utility design defect test relies on hindsight and could hinder the adoption of HAV technology.  The elements of the cost-benefit test are: how dangerous the design is; the likelihood that the danger would occur; the potential for an alternative design; the cost of an alternative design; and if the alternative design would alter the function of the product.[57]  In Pannu, the plaintiff argued: (1) his injuries could have been prevented if there were slight modifications to the center and width of the car track; (2) the slight modifications were relatively inexpensive; and (3) the defendant failed to show that an alternative design was not available or that an alternative design would reduce the function of the vehicle.[58]  The Court held the defendant liable for the plaintiff’s injuries.[59]  The plaintiff had the advantage of hindsight because he could determine the cause of the accident and how it could have been prevented.  Prior to vehicles being put onto the market, the vehicle must meet safety standards established by the federal government and undergo rigorous testing by the manufacturer.[60]  Rigorous testing cannot simulate every possible accident.  It is possible that a HAV could meet federal safety standards, be rigorously tested, and fail the risk-benefit test because the plaintiff, who has the benefit of hindsight, can claim that a minor change could have prevented their injuries.

Failure to warn could create high liability for manufacturers.  Failure to warn is a strict liability claim; when a manufacturer does not warn operators, the manufacturer is liable for the crash.[61]  It is impossible to anticipate every situation that could require a warning, especially when the instruction manual states the limitations of the HAV’s systems.[62]  But failure to warn in the context of a HAV probably would not be significantly different from liabilities that automakers currently face.

When considering product liability, legislatures should make liability cost predictable for manufacturers in order to encourage the adoption of HAV technology.

IV. Recommendation

In order for HAVs to drive on public roads, legislatures must balance the public interest with manufacturers liability expectations.  Legislation must achieve tort law goals: compensate victims, create incentives for safety, and provide victims with a societally sanctioned procedure to prove that they were wronged.[63] Regulations cannot create extreme financial burdens on manufacturers or manufacturers will forego HAV technology.  HAV technology is a public good because HAVs improve the general public’s quality of life.

A. The Future of Insurance Coverage

Legislatures should adopt a hybrid system of insurance to balance public safety in the short term and encourage technology that will reduce crashes in the long term.  Legislatures should create no fault insurance schemes for their state and HAV manufacturers should have liability insurance with an SIR clause.

A no fault insurance regime is preferable for HAVs because the no fault regime is efficient: victims of a crash receive damages quickly and litigation is avoided.  To still reap the public safety benefits of litigation—exposing the cause of the accident—without delaying damages paid to victims, insurance companies could pay their insured’s damages.  Policyholders should grant insurance companies subrogation rights.  Insurance companies can sue HAV manufacturers for product liability.  The HAV operators would be made whole again in the shortest amount of time, but malfunctions in HAV operating systems would still be investigated and exposed through litigation.
Manufacturers can manage their expected liability with a SIR clause written into their liability insurance policy.  SIR would create a maximum expected out-of-pocket cost for product liability lawsuits.  If manufacturers know that they will face limited liability, they will be encouraged to adopt HAV technology.

Rand Corp. suggests that Congress pass legislation requiring the operation of HAVs to be overseen by licensed drivers, but this would defeat the purpose of HAVs.[64] Granted, if licensed drivers oversaw HAVs, then states would not have to adopt new insurance schemes.  Requiring a HAV operator to be a licensed driver defeats a primary function of an HAV—fast travel for all, regardless of ability.  Additionally, recall the Tesla crash—both the Tesla and the licensed driver failed to detect a white tractor-trailer.[65]  Ultimately, the licensed driver was not able to prevent the crash.

B. The Future of Product Liability

To simultaneously protect public safety and minimize manufacturer liability, state legislatures should pass a statute that immunizes HAV manufacturers in product liability lawsuits.  The plaintiff must make a prima facie showing that the manufacturer was willful and wanton.  A statute requiring a showing of willful and wanton misconduct is a tougher standard than a showing of a duty of due care.[66]  A plaintiff would have to demonstrate that a manufacturer intentionally caused the plaintiff’s harm or that the manufacturer was unconcerned in protecting the plaintiff and the plaintiff’s property.[67]  To demonstrate a deliberate intention to cause harm, the plaintiff would have to show the manufacturer learned of the defect in the HAV’s operation system but failed to take any corrective action.  A lack of concern for the safety of the plaintiff and the plaintiff’s property is demonstrated when the plaintiff shows the manufacturer knew the HAV operating system was defective before the crash or that the manufacturer somehow caused the operating system to malfunction.  The tougher prima facie negligence standard of willful and wanton would still hold manufacturers liable for wrongful actions, but reduce overall liability.

Additionally, courts could still use the design defect tests.  Recall the consumer expectation test and the cost benefit/risk utility test used in design defect cases left HAV manufacturers open to high levels of liability.  But if the prima facie showing for design defect was raised to willful and wanton, then the design defect tests would only be applied to cases where the manufacturer knew there was a defect and failed to take action or the manufacturer somehow caused the operation system to malfunction.

Thus, the previous example—the human operator and the autopilot that failed to distinguish the white side of tractor-trailer against a bright-lit sky, resulting in a fatal crash—would not meet a standard of willful and wanton.  Tesla did not know the defect existed and Tesla did not cause the operation system to malfunction.  Therefore, Tesla would not be liable for the crash even though there was a design defect because the plaintiff could not make a prima facie showing of willful and wanton.  Because there is no prima facie showing, the Tesla operator would not have a product liability case against Tesla.  Therefore, the fact finder would not be given the chance to form a minimum safety expectation about the capability of the Tesla.  Although public safety is a top priority for legislatures, it is disadvantageous to punish manufacturers for defects that are beyond the capability of the HAV.

Rand Corp. argues that raising the standard for tort liability would not achieve the goals of tort law.[68]  If state legislatures create a no-fault, SIR product liability insurance scheme, raising the standard for tort liability would still achieve the goals of tort law—compensating victims, creating incentives for safety, and providing victims with a societally sanctioned procedure to prove that they were wronged.  First, raising the standard to willful and wanton does not prevent victims of a crash from filing an auto insurance claim.  Awards in product liability suits would not be to compensate victims, but to reimburse insurance companies for paying victim’s damages.  Manufacturers must produce safe vehicles to not be found willful and wanton.

Second, raising the standard does not eliminate an incentive for safety.  Federal regulations will provide the biggest incentive for safety.  Today “manufacturers bear the responsibility to self-certify that all of the vehicles they manufacture for use on public roadways comply with all applicable Federal Motor Vehicle Safety Standards (FMVSS).”[69]  The FMVSS will not disappear, as HAVs grow more popular.  Consequently, minimum safety requirements for the manufacturing of HAV operation systems as set by the federal government would create a base line incentive for safety.  Raising the standard would create additional incentives for safety because knowledge of a defect and failure to take action or causing the defect would create liability.

V. Conclusion

As HAVs move beyond the testing phase and become more commonly used on public roadways, state legislatures have a responsibility to create a workable liability system.  Legislation should balance encouraging the growth of technology while simultaneously protecting public safety and penalizing manufacturers who do not protect public safety.  Auto insurance must adapt the language it uses to assess liability and pay damages—no fault insurance pays damages without a declaration of fault.  Raising the standard for suing HAV manufacturers to willful and wanton allows victims a societally sanctioned process to air their grievances but shields manufacturers from unrealistic expectations.

[1] Matt McFarland, Uber Self-Driving Car Kills Pedestrian in First Fatal Autonomous Crash, CNN Money, (Mar. 19, 2018; 1:40 PM),

[2] Id.

[3] Id.

[4] Sonari Glinton, The Transportation Officials to Issue Rules for Self Driving Vehicles, NPR (Sept. 20, 2016),

[5] See, 5 Reasons You Need Car Insurance, ESURANCE, (stating that in 2013, the average property damage cost was $9,300, and accidents that caused disabling injuries were $80,700 on average).

[6] See JAMES M. ANDERSON, ET AL., AUTONOMOUS VEHICLE TECHNOLOGY: A GUIDE FOR POLICY MAKERS 140 (2016) (stating that every state except New Hampshire requires drivers to have auto insurance).

[7] Id. at 114 .

[8] Self Driving Cars and Insurance, INS. INFO. INST. (July 2016),

[9] Cecilia Kang, Self-Driving Cars Gain Powerful Ally: The Government, NY TIMES (Sept. 19, 2016),

[10] U.S. Department of Transportation, Federal Automated Vehicles Policy 45 (2016).

[11] Kang, supra, note 9.

[12] Kang, supra, note 9.

[13] See Aarian Marshall, Michigan Just Embraced the Driverless Future, WIRED (Dec. 9, 2016, 6:38 PM), (“The fact of the matter is 75 percent of all the companies that are doing research and development in this space are in southeast Michigan.”); Matt McFarland, Uber Self-Driving Car Kills Pedestrian in First Fatal Autonomous Crash, CNN: MONEY (Mar. 19, 2018, 1:40 PM), (stating that Arizona is “a hotbed of self-driving car development” after Governor Dough Ducey updated an executive order that allows autonomous vehicles to drive on state roads without a driver behind the wheel).

[14] U.S. Department of Transportation, Federal Automated Vehicles Policy 84 (2016).

[15] Id. at 85.

[16] Id. at 83.

[17] Id.

[18] JAMES M. ANDERSON ET AL., The U.S. Experience with No-Fault Automobile Insurance, RAND 7 (2010).

[19] ANDERSON ET AL., supra note 6 at 113.

[20] Id. at 112.

[21] ANDERSON ET AL., supra note 18 at 8.

[22] ANDERSON ET AL., supra note 6 at 112.

[23] No-Fault Insurance and Fault Insurance, DMV.ORG (last visited Dec. 4, 2016),

[24] Id.

[25] ANDERSON ET AL., supra note 18 at 7.

[26] How Do The Police (or a Court) Determine Who’s at Fault in an Accident if Both Drivers Insist it Wasn’t Them, JUSTIA, (last visited Apr. 2, 2018)

[27] Car Insurance Subrogation Defined, ESURANCE, (last visited Apr. 2, 2018)

[28]Subrogation, BLACK’S LAW DICTIONARY, West (2011).

[29]No-Fault Insurance and Fault Insurance, supra note 23.

[30]ANDERSON ET AL., supra note 19.

[31]Self-Insured Retention—What it is & Isn’t, ALIGNED INS. (last visited Mar. 23, 2018),

[32]See Lewis Bass & Thomas P. Redick, What is Product Liability Insurance?, PROD.LIAB.: DESIGN AND MFG. DEFECTS § 22:1 (2d ed. Sept. 2017) (explaining that insurance companies already offer product liability insurance).

[33]What is Product Liability?, FINDLAW, (last visited Dec. 4, 2016).


[35]Bass & Redick, supra note 32, at § 18.25.



[38]ANDERSON ET AL., supra note 19, at 123.


[40]Id. at 124.

[41]Soule v. Gen. Motors Corp., 8 Cal. 4th 548, 567 (Sup. Ct. 1994).


[43]ANDERSON ET AL., supra note 19 at 123.


[45]Soule, 8 Cal. at 567 (1994).

[46]Pannu v. Land Rover N. Am., Inc., 191 Cal. App. 4th 1298, 1313 (2011).

[47]ANDERSON ET AL., supra note 19 at 125.

[48]Id. at 127.


[50]Bill Vilasic & Neal E. Boudette, Self-Driving Tesla Was Involved in Fatal Crash, U.S. Says, NY TIMES, (June 30, 2016),

[51]What is Product Liability?, FINDLAW (last visited Dec. 4, 2016),

[52]Pannu v. Land Rover N. Am., Inc., 191 Cal. App. 4th 1298, 1299 (2011).


[54]Vilasic & Boudette, supra note 50.


[56]ANDERSON ET AL., supra note 19 at 125.

[57]Pannu, 191 Cal. App. 4th at 1313 (2011) ( “[T]he gravity of the danger posed by the challenged design, the likelihood that such danger would occur, the mechanical feasibility of a safer alternative design, the financial cost of an improved design, and the adverse consequences to the product and to the consumer that would result from an alternative design.”).



[60]Akweli Parker, How Car Testing Works, HOWSTUFFWORKS (Mar. 11, 2009),

[61]Pannu, 191 Cal. App. 4th at 1316 (“[T]he reasonableness of the defendant’s failure to warn is immaterial”).

[62]ANDERSON ET AL., supra note 19.

[63]Id. at 142.

[64]Id. at 144.

[65]Vilasic & Boudette, supra note 50.

[66]See generally, Illinois Tort Immunity Act, 745 ILCS 10/1-210 (2018) (stating willful and wanton conduct means “an actual or deliberate intention to cause harm.”).

[67]Id. (“‘Willful and wanton conduct’ as used in [this Act] means a course of action which shows an actual or deliberate intention to cause harm or which, if not intentional, shows an utter indifference or conscious disregard for the safety of others or their property.”).

[68]ANDERSON ET AL., supra note 19, at 142.