While AI driving technologies are advancing rapidly, public acceptance moves only at the speed of trust. That is why the current race to deploy driverless vehicles is as much a race for public trust as it is a race for superhuman driving technology. Lawyers, policymakers, and regulators—just as much as engineers—may prove to be the most important variable in when and how autonomous vehicles (AVs) begin full-scale operations. The legal profession needs to help now in resolving novel questions about how we set AV safety standards, how liability should be apportioned, and how data may be used. Time matters. The nations and companies that are first to deploy successfully may well set the standard for all others. This could be the difference between an industry that stalls or is hijacked by other interests and an industry that achieves its potential of saving millions of people from needless death and injury, drastically reducing the cost and burden of private travel, easing traffic, freeing up leisure time, opening up urban spaces, and slashing carbon emissions, all without compromising privacy. This article addresses some of the most important issues that need to be resolved as AVs are deployed and offers some insights about the most promising approaches to meet this moment.
Lesson Learned About Policy, Trust, and New Technology
Technology always moves faster than the law. Automobiles began operating in the United States fifteen years before there was a single statewide traffic law in place. For decades, there were no speed limits, driving lanes, stop signs, brake lights, driver training courses, licenses, reckless driving laws, or virtually any other rules of any kind—just mayhem. In our own time, we’ve seen the same thing happen in the Wild West of cyberspace. Although most Americans have been living much of their lives in the virtual world for decades, the states and federal government have still barely begun to respond to an epidemic of privacy, security, information integrity, and cybercrime concerns that have infected the digital world.
Both automotive and cyber histories are converging in the AV space, where regulation is advancing but still lags behind the technology. The AV companies that are most likely to succeed will depend on the legal and policy communities to help manage the effects of this technology, develop internal standards, and anticipate social costs until regulators can catch up. The only way to build and maintain that sort of momentum is to earn the public’s faith in the AV category. Without it, the AV industry—and all its lifesaving and carbon-reducing effects—could be stymied. Move too slowly and it faces a regulatory stalemate that keeps it in regulatory limbo while advantaging less-responsible nations and operators. Move too quickly and it provokes preemptive regulation that overcorrects and saps the energy of the industry.
The challenge is further complicated by a normal human reluctance to concede defeat to machines in any task, from printing books, to washing clothes or playing chess. Overcoming a natural aversion to abdicating a human task to a machine will require showing how this technology—like a washing machine—improves the lives of the users and creates more and better opportunities for society as a whole. Fortunately, the AV industry is not the first to automate a human activity on a massive scale, so it can draw on past experience from other industries in earning public trust.
The scariest ghost of Christmas past for AVs is the “driverless elevator.” Passenger elevators in America were originally manned by elevator operators. Operating elevators was not easy, and overanxious passengers or distracted operators often produced grisly results. When reliable automatic elevators arrived on the market in 1900, the public should have embraced them, but instead were skeptical due to false rumors spread by elevator operator associations that automation brought increased risk. The industry failed to get out ahead of these issues by establishing their own clear safety standards, public education campaigns, and collaboration with regulators. As a result, automatic elevators languished for decades while the public chose to climb dozens of flights of stairs or move meetings to other buildings just to avoid them. Real estate developers listened to the public and went back to manually operated elevators. The result was a slow, expensive, costly, and dangerous system that remained long after it should, caused needless deaths and injuries, and distorted city planning. The public’s attitude finally changed only after the state of New York launched a public trust campaign in response to the 1945 elevator operator strike, which cost the city of New York $100 million in economic losses. The public trust campaign educated the public about how elevators functioned without a human operator, established a state system of annual inspections and certifications, and incentivized industry innovations to ease public anxiety, such as the emergency phone, stop buttons, alarms, and remote operators. Once driverless elevators were trusted and widely used, they dramatically reduced elevator deaths, transformed cityscapes with taller and more efficient structures, and changed forever how we live and work.
Like automatic elevator companies, AV companies will likely face resistance from commercial drivers and other incumbent products, as well as concerns from those who benefit from the current insurance and liability regimes and have an interest in sowing public doubt about the security of AVs and the potential for abuse and exploitation of their personal data. Companies developing commercial AVs—ride-hail, delivery, and long-haul transport—need to address these concerns head on with responsible principles for assuring safety, security, compensation, and privacy. The goal is to do what automatic elevator companies failed to do: generate public trust and draw a road map that helps policymakers find their way quickly after this inevitable period of lag and uncertainty.
The stakes could not be higher. The effects of failing to build trust in the areas of safety, liability, and privacy will be much worse than just a lost opportunity. Other nations, particularly China, are already working to establish their own AV regimes—supported by government subsidies, lax liability laws, and advanced infrastructure—to define markets worldwide. Relying on lower safety and privacy expectations in less-developed parts of the world, other nations could drive the industry in intolerable directions and empower authoritarians by converting vehicles into means for surveillance. A delay in the U.S. AV deployment, therefore, would not only cause U.S. companies to miss out; it would also result in good technologies being hijacked altogether. The race to deliver self-driving technology to the world will depend on open societies and companies accelerating their own trust and collaboration in safety, liability, and privacy.
AVs are already well on their way to achieving “superhuman” performance, i.e., performing at or better than the level at which the average driver operates in terms of driving smoothly, efficiently, and safely. Indeed, many industry stakeholders have been focused on finding the point at which AV safety is so reliably better than human driving safety that they can pronounce AVs safe enough. This is complicated not by science but by perspective. Logically, as soon as AVs perform even slightly better than human drivers, they should be deployed because, by definition, they will begin saving lives. Emotionally, however, the public is much more forgiving of human error than an error caused by a machine that is supposed to be safe. A traffic fatality in Arizona caused by a distracted safety driver in a test-AV led to a massive shutdown of Uber’s program and people shooting weapons at Waymo AVs.1 There is, in fact, no magic point at which the public will suddenly accept that AVs are safe enough. Instead, just as cars and airplanes went through a period of gradual acceptance as they grew safer, the same will inevitably happen with AVs.
Trust, however, shouldn’t depend on contrasting definitions of “superhuman enough” because, in reality, the safety case for AVs is much broader and deeper than simply comparing an AV “brain” to a driver. An “apples to apples” comparison of the safety of an autonomous driving system (ADS) to a human driving system (i.e., a person) may miss the real power of AV safety, which promises to revolutionize safe travel. Besides reducing the total volume of crashes and injuries, AVs do three things that automobile regulation has never contemplated. First, they immediately reduce the pool of potential crash victims in commercial vehicles. Second, they remove the factor that is the least correctable variable in traffic incidents and replaces it with a system that can be monitored, improved, and replicated. And third, they alter the driving environment itself, including the behavior of other vehicles, to reduce the circumstances that lead to crashes. Refashioning regulation to account for these massive improvements in safety would make it easier for the public to recognize the immediate value of AVs, and help regulators accelerate their efforts to responsibly guide AV companies to achieve appropriate standards.
Removing Human Drivers from Commercial Vehicles Improves Road Safety
In America, more than ninety people die in car crashes every day, and another three million are injured each year.2 The most likely victim is the driver. In fact, 52.1% of traffic fatalities are the driver—more than passengers, pedestrians, and occupants of other vehicles combined.3 Around two million more drivers in car crashes experience permanent injuries.4 Although commercial vehicles vary as a percentage of traffic by region, freight and commercial transportation vehicles drive one out of every eight miles in the country. Ride-hailing vehicles alone can travel up to 400% more than a personally owned car in a given year.5 And by simply removing the human driver—the one statistically most likely to suffer injury or death in a crash—commercial AVs immediately produce a safety benefit. This improvement holds, even if commercial AVs are operating only at an equivalent crash rate to humans. Driverless purpose-built vehicles for ride-hail or delivery are therefore, by definition, a safety improvement over equivalent human-driven vehicles because they reduce human exposure to severe-injury crashes.
Current regulations do not account for this safety improvement in traditional applications for new AVs. However, National Highway Traffic Safety Administration (NHTSA) regulations may give regulators the space to do so. NHTSA’s regulations allow manufacturers to introduce commerce vehicles that do not meet current design requirements under some circumstances if they are deemed to make operation as safe or safer, or “the applicant is otherwise unable to sell a vehicle whose overall level of safety or impact protection is at least equal to that of a nonexempted vehicle.”6 Thus, a manufacturer may seek an exemption to produce a vehicle that has, for example, no steering wheel, rearview mirror, or other manual controls and seating needed by a human driver because it is designed to prevent a human commercial driver from operating it. The regulations require the manufacturer to identify the features that improve the safety of the vehicle (here, the more reliable autonomous driving system “brain”) and “[o]ther arguments that the overall level of safety or impact protection of the vehicle is at least equal to that of nonexempted vehicles.”7 Although it relied on a different exemption provision to approve a low-speed delivery vehicle developed by Nuro, NHTSA did acknowledge that one of the features that led it to approve the petition was the fact that it did not contain any occupants.8 “Given that both an exempted and compliant [vehicle] would have no occupants and would operate without a human driver, compliance with the three requirements from which Nuro seeks an exemption would not provide a safety benefit.”9 This suggests one means by which designing vehicles to reduce the necessity of a driver could be recognized for what it is—a safety feature that dramatically reduces the likelihood of the most commonly injured victim of a vehicle crash even being present.
Replacing and Making More Reliable the Most Varied and Uncorrectable Feature of the Vehicle
The human driver, in addition to being a potential victim of a collision, is the most variable and dangerous aspect of the vehicle itself. NHTSA’s 2008 National Motor Vehicle Crash Causation Survey concluded that human error is the critical reason for 93% of crashes.10 Specifically, the three major factors most frequently reported included improper lookout, excessive speed, and inattention. If there was ever any doubt that human operators are the most significant factor in vehicle crashes and fatalities, traffic statistics during the pandemic have provided sad proof of this. Although the same kinds of vehicles operated on the roads in 2020 as in 2019, and traffic volume dropped dramatically because of reduced travel and activity, deaths spiked because of poor drivers. According to NHTSA, total traffic volume fell 16% on U.S. roads during the first half of 2020, but the fatality rate rose by 30%. NHTSA believes that the pandemic has pushed people towards riskier behavior related to distraction, use of drugs and alcohol, and greater opportunities for speeding and stunt driving.11 As a practical matter, a purpose-built commercial AV removes a feature that the manufacturer can’t control with one that can be evaluated, improved, and refined by manufacturers. Rather than guessing how drivers may or may not use a design feature, and hoping they will use it appropriately, AV manufacturers and their regulators can ensure consistent, reliable, and improvable operation. In effect, it is the equivalent of including an early version of safety restraints that over time improved because they were stronger and more reliable than people throwing their arms in front of one another.
Altering and Improving the Threat Landscape
Finally, AVs provide a third safety feature that immediately provides an improvement over traditional commercial vehicles. The only thing as dangerous and unpredictable in your car as a human driver is the human driver in another car. The ADS “brain,” however, is part of a system that extends beyond the car. AVs function by perceiving traffic conditions, predicting behavior by other road users, and then planning movements that permit the journey to proceed safely. In effect, unlike a human driver, they can constantly monitor and predict the behavior of other vehicles, and by rerouting and other behaviors, they also influence the behavior of other vehicles. By allowing more vehicles on the roads that are directly responding to external sensors, regulators would immediately increase the percentage of vehicle miles traveled involving consistent and predictable responses that make safety events less likely. The system of sensors, camera, radar, and lidar allows the vehicle not only to remove a potential victim from a vehicle, it also has the capacity to anticipate the actions of other vehicles and respond appropriately to reduce the number of hazardous situations.
The Safety Act12 may offer an additional means for NHTSA to recognize these safety enhancements appropriately and, as NHTSA has recently announced, apply its authority to “create a level playing field” for fledgling companies to develop domestic vehicles with full driving automation.13 Section 30114 of the Safety Act, as recently interpreted, would allow NHTSA to permit AVs to be deployed on public roads through a pilot program to inform NHTSA’s judgment about how best to modify existing regulations to address the unique safety features of AVs.14 This approach could both improve public confidence and acceptance of AVs and facilitate faster regulatory engagement once vehicles achieve roughly human performance levels.
Rethinking Commercial Liability
The second principal challenge for regulators in the AV space concerns the allocation for risk and fault when AVs are involved in some form of accident or incident. Because AV technology is moving far faster than efforts to develop a more appropriate liability regime, initially participants in the self-driving industry will likely operate under the existing liability regime: a system that is poorly suited to allocating costs and burdens in the AV context. Our existing regime either turns on things like mental states (negligence or recklessness) and “reasonable” human reactions/responses that don’t apply to AVs or imposes strict “product liability” in which any accident resulting from an interaction between a human and a product is the product’s fault. Accordingly, even as AVs operate within existing laws, AV companies will need to start laying the foundation for the future that will produce incentives for AVs to improve their design and not incentivize humans to drive irresponsibly around AVs.
To change how our society manages traffic risk and compensates for losses—and to avoid efforts by entrenched interests to preserve the status quo—the self-driving industry needs to develop an alternative that the public can understand and recognize as more trustworthy.
Initially, the self-driving industry can expect parties affected by an accident to assert traditional claims for personal injury and/or property damage under existing theories of recovery, with some combination of negligence and strict products liability being the most likely. The handful of lawsuits that have been filed in the self-driving space to date bear this out.15 Variations will largely depend only on things like the level of vehicle autonomy, the vehicle’s ownership model, and the jurisdiction where the incident occurred. For example, vehicles with lower-level autonomy that require some level of human engagement will likely raise more “foreseeable misuse” claims than higher-level AVs that operate with full autonomy. Likewise, whether a vehicle is owned privately or is part of a fleet operated for use by drivers could produce different results in states such as California that subscribe to a common carrier standard of care.16 This standard would likely be deemed more applicable to fleet-owned than privately owned AVs.17 AVs will, therefore, need to operate initially using whatever ownership models, autonomy restrictions, and geographic limitations that keep them from being put at an unreasonable disadvantage regarding liability claims vis-à-vis human-operated vehicles. Longer term, though, they will need to engage the public and lawmakers in a rational effort to use unique features of AVs to reduce administrative costs, facilitate fair and predictable outcomes, encourage safer behavior, and allow full mobility.
The first step toward a more suitable liability regime is explaining why traditional theories of liability are not a good fit for AI-powered AVs. The current framework of auto liability largely allocates responsibility to either the operator or the manufacturer of the vehicle. But with self-driving technology, perception, prediction, planning, and maneuvers are all performed by the vehicle itself, blurring boundaries around operator and manufacturer, and around responsibility and accountability. A programmer who creates a vehicle operation does not specifically tell the vehicle how to react in every situation. Instead, it creates the capacity for the vehicle itself to recognize situations and make choices, just as human drivers do, based on certain parameters (i.e., a desire to avoid a collision while also complying with traffic requirements). Moreover, the kind of inquiry into “what was that driver thinking” would become much more expensive and complicated in the AV context because AVs can’t express their thoughts, and interpreting their code—as the Toyota unintended acceleration cases showed—is extensive, expensive, and usually inconclusive. Finally, AI technology itself presents challenges to traditional notions of fault. A vehicle may do things “better” than a human, such as stop faster or take unexpected action to avoid a hazard that a human wouldn’t have perceived. If that causes a trailing vehicle to hit it because the driver expected a “normal human” action, who should be at fault?
AV companies are already contemplating alternative systems that might better allocate fault to incentivize improvements. Many are modeled on existing regimes for workers’ compensation or other predictable sources of injury, such as no-fault insurance pools and capped compensation programs. It is too early to say what system would be best, particularly until enough cases occur to help identify patterns, but the industry and regulators can at least agree on the key features of a better regime. In particular, policymakers might agree that any system must incentivize relevant parties to take appropriate precautions, allow AVs adequate time and resources to improve their systems to protect much larger groups than just the individuals involved in a specific incident, and leverage the data self-driving vehicles generate to streamline causation determinations. Lawyers and policymakers who engage in these efforts can have an outsized influence in framing the reforms and helping develop incremental milestones.
Adopting early principles like these would also help AV companies generate trust that will facilitate whatever regime comes next. Being transparent about safety data, mitigation strategies, and ways in which AV technology may streamline and improve the accuracy of investigations will help give the public confidence that incidents involving AVs afford them the chance to have a more rapid, accurate, and fair accounting of events.
Finally, most writing about AVs’ impact on privacy has tended to focus on issues that aren’t unique to AVs or aren’t features at all. Analysts and advocates have raised concerns about potential risks like consumer profiling, government surveillance, malicious hackers, and more. These are not novel concerns, and they are not unique to AVs. Ride-hailing and delivery companies already collect extensive data on individuals via consumer and driver apps, and they and other consumer product companies have faced and will continue to face scrutiny for their data-related practices, including consumer profiling, government access, and data security. But more importantly, these generalizations miss the critical point that AVs are far from homogenous, and the privacy implications largely depend on several features that vary widely across the technical, business, and design domains. For example, privacy issues will differ for autonomous freight, ride-hail, personal, and delivery vehicles. They will differ for “self-contained” AVs and “interdependent” AVs. And at a more profound level, they will differ depending on the identities and values of market leaders. Market leadership in this space is more than an issue of commercial success—it will affect important societal values depending on which nation’s AV technology becomes the market standard. If China were to win this race and establish itself as a global AV industry leader, there is little doubt that AVs will be designed not simply as a means of mobility but also as a means of surveillance.
Nevertheless, there are some novel privacy considerations related to AVs that warrant public attention and discussion and that AV companies and policymakers together can lead. While the full range of these issues is beyond the scope of this piece, two examples are worth discussing, namely the use of information gathered by external sensors and the appropriate type and amount of data collected and retained from rides.
Third-Party Data Collected by External Sensors
While automobile manufacturers are increasingly installing sensors on consumer cars to assist drivers, AVs use a multitude of different sensors to inform the machine rather than the person. These sensors collect various types of information about the external environment, such as radar and lidar point clouds and physical imagery. Imagery data tend to get the most attention, but AVs are hardly the first example of cameras in public spaces, or even of cameras mounted on vehicles. Google Street View launched around thirteen years ago, and Microsoft StreetSide and Apple LookAround are two of several comparable alternatives. The missteps and pitfalls that those providers experienced serve as valuable lessons today for AV makers and operators, especially in the international context (see, for example, local reactions to Street View in Germany18 and Japan19). While in the United States the law does not recognize a reasonable expectation of privacy in public spaces, the public and regulators expect AV companies to serve individuals’ privacy interests rather than put them at risk, and what those mean can vary considerably among cultures. At the very least, however, companies can expect the public to demand transparency about their own specific processing within particular operating areas, including how they will use sensor data, how long they will retain the data, how they will control access and enforce deletion policies, and how these data fit within the applicable privacy regime.
Classification of Data
A common refrain is that AVs generate a massive amount of data, so the public and regulators should be concerned. It’s true: Sensors, vehicle data, logs of model outputs, routing instructions, and more all generate a substantial amount of data per AV trip. But volume does not equal sensitivity. In general, the most significant concern with data volume is cost, so AV companies have an inherent interest in reducing data collection to what is needed to allow a customer to travel safely and comfortably from one point to another. The real question is not how much data does an AV generate, but how much of the data is associated or reasonably associated with an identifiable person. Companies can vary widely in how they approach this issue. Some may decide to embrace this concept and link as much as possible to consumers in order to provide personalized experiences, boost valuations, and provide diverse paths to profitability, including data monetization. Others may decide to go the opposite route and link only the minimum necessary to provide requested services to win customer trust.
Once again, there is no single right approach yet, but working with regulators and developing clear principles will accelerate deployment across global markets. One thing we know for sure, though, is that the sooner AV companies formulate their privacy policies, the sooner they can gauge market and regulator reactions and adjust. In the race to deploy, knowing what the public needs will require a different form of trust—trusting consumers to advise what will and will not fly regarding their data.
The State of Play with the New Administration and Congress
The AV technology industry, like the state and federal laws and regulations that govern it, is still developing. Now is the time when it can build trust or suspicion and either help gain support or build a backlash. The choice remains fairly wide open. Although legislation was introduced in two successive Congresses, Congress has not yet enacted a comprehensive legal framework for AVs.20
Without a comprehensive legal framework for AV technology at the federal level, a patchwork of local and state autonomous laws have developed, making it more difficult for AV companies to operationalize and scale. Twenty-nine states have enacted some type of legislation relating to AVs, and the governors of eleven states have issued executive orders relating to AVs.21 The legal frameworks established by these states have varied greatly. For example, some states only allow for testing activities on public roads with a human backup driver in the vehicle, while others allow for testing and deployment of fully driverless vehicles on public roads. In California, home to many of the leading AV technology companies, AV companies seeking to transport passengers are governed by two regulators—the Department of Motor Vehicles and the Public Utilities Commission, which recently paved the way for AV companies to accept fares from passengers.22 Texas enacted a comprehensive AV law in 2017 that allows for the testing and deployment of driverless vehicles on the state’s public roads.23 Similarly, Florida’s AV law allows for fully driverless testing and deployment of the technology.24 Both the Texas and Florida laws prevent a patchwork of AV laws from developing within the states by preempting local governments from enacting their own AV laws. Other states, such as Arizona, have opted to implement the state’s AV legal framework via Executive Order.25 Finally, in Pennsylvania, the state’s Department of Transportation has issued regulatory guidance allowing AV companies to test the technology within the state, as long as a human backup driver is present in the vehicle.26
As new legislators and a new administration begin their terms in 2021, approaches to safety, liability, and privacy will again likely guide any proposed AV legislation. Given Congress’s repeated efforts to legislate in this area, and the U.S. Department of Transportation’s efforts to develop guidance documents on AVs,27 an opportunity exists for the industry to find its regulatory footing.
Silicon Valley’s playbook of “move fast and break things” seems unlikely to work this time. As the proverb advises those about to begin a journey: “If you wish to go fast, go alone. If you wish to go far, go together.” The self-driving industry and its regulators have a responsibility to deliver on the benefits of life-saving AV technology, and the path to achieve those benefits is paved with trust for those who want to go far.
1. Simon Romero, Wielding Rocks and Knives, Arizonans Attack Self-Driving Cars, N.Y. Times (Dec. 2018), https://www.nytimes.com/2018/12/31/us/waymo-self-driving-cars-arizona-attacks.html.
2. Motor Vehicle Injury, Ctrs. for Disease Control & Prevention (Oct. 4, 2019), https://www.cdc.gov/transportationsafety/index.html?CDC_AA_refVal=https%3A%2F%2Fwww.cdc.gov%2Fmotorvehiclesafety%2Findex.html.
3. Car Accident Statistics: Fatalities Broken Down by Role, Aceable, https://www.aceable.com/safe-driving/car-accident-statistics.
4. Nat’l Highway Traffic Safety Admin. (NHTSA), Early Estimate of Motor Vehicle Traffic Fatalities for the First Quarter of 2019, Traffic Safety Facts (Aug. 2019), https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/812783.
5. Michael Nicholas, Peter Slowik & Nic Lutsey, Charging Infrastructure Requirements to Support Electric Ride-Hailing in U.S. Cities (Int’l Council on Clean Transp., Working Paper No. 2020-09, Mar. 2020), https://theicct.org/sites/default/files/publications/Charging_infrastructure_ride_hailing_US_03242020.pdf.
6. 49 C.F.R. § 555.6(d); see also 49 U.S.C. § 30113(b)(3)(B)(iv) (“compliance with the standard would prevent the manufacturer from selling a motor vehicle with an overall safety level at least equal to the overall safety level of nonexempt vehicles”).
7. 49 C.F.R. § 555.6(d)(1)(v).
8. Nuro, Inc.; Grant of Temporary Exemption for a Low-Speed Vehicle with an Automated Driving System, 85 Fed. Reg. 7826, 7827 (Feb. 11, 2020).
9. Id. at 7827.
10. Nat’l Highway Traffic Safety Admin., National Motor Vehicle Crash Causation Survey Report to Congress (July 2008), https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/811059; see also Nat’l Highway Traffic Safety Admin., Critical Reasons for Crashes Investigated in the National Motor Vehicle Crash Causation Survey (Feb. 2015), https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/812115.
11. News Release, Nat’l Highway Traffic Safety Admin., 2019 Fatality Data Show Continued Annual Decline in Traffic Deaths (Oct. 1, 2020), https://www.nhtsa.gov/press-releases/2019-fatality-data-traffic-deaths-2020-q2-projections; Nat’l Highway Traffic Safety Admin., Early Estimate of Motor Vehicle Traffic Fatalities for the First Half (Jan–Jun) of 2020, Traffic Safety Facts (Oct. 2020), https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/813004.
12. 49 U.S.C. § 30114.
13. Pilot Program for Collaborative Research on Motor Vehicles with High or Full Driving Automation, 83 Fed. Reg. 50,872, 50,882 (Oct. 10, 2018).
14. This approach would also further NHTSA’s efforts consistent with 49 U.S.C. § 30182, which confers authority on NHTSA to conduct motor vehicle safety research, “including activities related to new and emerging technologies.”
15. See, e.g., Nilsson v. Gen. Motors, No. 4:18-cv-00471 (N.D. Cal.) (plaintiffs alleged only a negligence theory); see also Huang v. Tesla, No. 19cv346663 (Santa Clara Super. Ct.) (plaintiffs alleged both negligence and strict liability theories against Tesla).
16. A common carrier is an entity that holds itself out to the public as able to transport persons from one location to another. See, e.g., Squaw Valley Ski Corp. v. Super. Ct., 2 Cal. App. 4th 1499, 1507 (1992). Common carriers are subject to a higher duty of care than traditional negligence: They must use the “utmost care and diligence for [passengers’] safe carriage, must provide everything necessary for that purpose, and must exercise to that end a reasonable degree of skill.” Lopez v. S. Cal. Rapid Transit Dist., 40 Cal. 3d 780, 785 (1985).
17. In states that have rejected the common carrier doctrine, courts have held that “the appropriate standard of care in negligence actions by passengers against common carriers is the objective, reasonable person standard in traditional negligence law.” Nunez v. Prof’l Transit Mgmt. of Tucson, Inc., 229 Ariz. 117, 122 (2012).
18. Eric Ho, Alas, There Will Be No More Google Street View in Germany, Time (Apr. 11, 2011), https://techland.time.com/2011/04/11/alas-there-will-be-no-more-google-street-view-in-germany.
19. Reuters, Google Reshoots Japan Views After Privacy Complaints, Reuters (May 13, 2009), https://www.reuters.com/article/us-google-japan/google-reshoots-japan-views-after-privacy-complaints-idUSTRE54C22R20090513.
20. In the 115th Congress, H.R. 3388 (SELF DRIVE Act) and S. 1885 (AV START Act) were introduced. In the 116th Congress, a variation of the SELF DRIVE Act was introduced as H.R. 8350.
21. Autonomous Vehicles: Self-Driving Vehicles Enacted Legislation, Nat’l Conf. of State Legislatures (Feb. 18, 2020), https://www.ncsl.org/research/transportation/autonomous-vehicles-self-driving-vehicles-enacted-legislation.aspx.
22. Decision Authorizing Deployment of Drivered and Driverless Autonomous Vehicle Passenger Service, Dec. No. 20-11-045 (Cal. Pub. Util. Comm’n Nov. 23, 2020), https://docs.cpuc.ca.gov/PublishedDocs/Published/G000/M352/K185/352185092.PDF.
23. S.B. 2205, 85th Leg. (Tex. 2017), https://capitol.texas.gov/tlodocs/85R/billtext/pdf/SB02205F.pdf.
25. Ariz. Exec. Order No. 2018-04, Advancing Autonomous Vehicle Testing and Operating: Prioritizing Public Safety (Mar. 1, 2018), https://azgovernor.gov/sites/default/files/related-docs/eo2018-04_1.pdf.
26. Pa. Dep’t of Transp., Automated Vehicle Testing Guidance, Pub. 950 (2020), https://www.penndot.gov/ProjectAndPrograms/ResearchandTesting/Autonomous%20_Vehicles/Documents/PUB_950_9-20.pdf
27. See, e.g., Nat’l Sci. & Tech. Council & U.S. Dep’t of Transp., Ensuring American Leadership in Automated Vehicle Technologies: Automated Vehicles 4.0 (Jan. 2020), https://www.transportation.gov/sites/dot.gov/files/2020-02/EnsuringAmericanLeadershipAVTech4.pdf.