By Kyle Durch


Think back to the last time you drove on a freeway. The vehicles around you were all driven by people. Were any of those drivers distracted? Were there any speeders. Were you speeding? Assuming the best intentions, people just want to get from point A to point B in the quickest, most convenient way possible. To make transportation safer for everyone, shifting the driving burden from people to autonomous vehicles (AVs) is now feasible, and the transition must happen as quickly as possible with the guidance of smart regulation.


Poor decision-making on the road is one of the leading causes of injury and death. The National Highway Traffic Safety Administration (NHTSA) reported that over 36,000 people were killed in motor vehicle crashes in 2018.[1] And although traffic decreased by forty-one percent nationwide thanks to the COVID-19 pandemic—including a decrease of more than sixty percent in urban areas—the remaining drivers made riskier decisions, contributing to a spike of over thirteen percent in traffic fatalities during the summer of 2020.[2] AVs strike at the core of this problem, allowing drivers to safely engage in non-driving activities. Although critics argue that current systems have difficulty dealing with unusual driving circumstances, they acknowledge that driver assistance/replacement systems show great promise in preventing accidents that result from common situations like fatigue and distraction.[3] But how does this “sci-fi” technology work?


AVs are functionally the same as any other vehicle on the road, differing only in the use of advanced sensing and control technology. Self-driving technology can be built into electric cars, gas cars, and even into tractor trailers.[4] AVs are equipped with sensors that feed data to an onboard computer, which in turn controls the vehicle. Different kinds of sensors are currently in use, including radio range finders, or radar; light detection and ranging, or lidar; and suites of cameras and ultrasonic obstacle sensors.[5] Combined with Global Positioning System (GPS) location awareness, the computer analyzes the sensor data to produce a simulated representation of the world.[6] The computer then uses that representation to react to the world by sending control signals to the vehicle’s steering, acceleration, and braking systems.[7]


But the vast amount of data collected outstrips the computing power available in each vehicle, requiring the use of off-board processing. For example, Tesla uses a data center to run a centralized neural net—a type of artificial intelligence program—that processes data sent to it by its global fleet of cars.[8] The learning produced is then transmitted to every vehicle on the network. In this way, a single Tesla encountering a new stretch of road shares its experience with every other Tesla. Like the Borg in Star Trek, AVs operating on a neural net share a “hive mind” that proactively adapts to the world.[9] Given enough time and experience, these systems have the potential to far exceed the capability of any human driver. That said, connection to the “mothership” is essential.


Consistent secure communications enable the neural net to function. Cellular networks provide the bulk of this connectivity.[10] Like any other use of wireless communication, or network connections in general, these networks may be exposed to attack. Hackers may jam signals, insert false inputs, or even directly control vehicles.[11] Preventing these attacks will require standardization, based on an intelligent approach to cybersecurity.


Unfortunately, the current regulatory framework is woefully unprepared for this new paradigm. The NHTSA has regulatory authority over all highway vehicles, but no authority over cybersecurity.[12] Lacking the requisite authority and expertise, the agency must rely on the Federal Communications Commission (FCC).[13] The FCC in turn approaches cybersecurity more generally, with efforts such as spectrum availability for transportation and internet of things devices.[14] Without a single entity focused on the best application of cybersecurity to AVs, there is a strong likelihood that uncoordinated implementations by different manufacturers will result in a fragmented, less-secure mess.[15]


On the bright side, the NHTSA is aware of its lack of expertise in cybersecurity and is taking initial steps to address the problem. Currently, the agency seeks comments in an advanced notice of proposed rulemaking on Automated Driving System (ADS) safety, now extended to Apr. 1, 2021.[16] Although early in the process, this rulemaking is a critical opportunity to ensure that the NHTSA sets proper benchmarks and limits for the industry. Technology-specific rules based on current capabilities would stifle innovation. On the other hand, overly broad technology-neutral rules may fail to guide industry into safe and secure AV development. The agency needs to adopt a middle-ground approach, requiring industry to use best-practice security measures while incentivizing implementation of any AV system with proven effectiveness. In fact, any manufacturer who fails to implement AV systems in its vehicles should be held accountable for failing to take steps to protect the public health and welfare.


Between the NHTSA and vehicle manufacturers, there are many entities who play a role in highway safety, and they all have a responsibility to the public. The development of the autonomous vehicle is arguably the most important highway safety innovation since the seatbelt and the airbag. And like those innovations, the implementation of AVs will see a long, bitter policy battle.[17] But in the end, a minimum baseline of “responsibility to implement,” along with security guidelines and standards, would go a long way toward reducing suffering on America’s roads.


[1] Automated Vehicles for Safety, NHTSA, (last visited Feb. 26, 2021).

[2] Jacob Baumgart, Pandemic Revs Up Bad Driver Behavior in Maryland, Patch (Feb 12, 2021, 8:33 PM),

[3] See Peter Hancock, Are Autonomous Cars Really Safer than Human Drivers?, Sci. Am. (Feb. 3, 2018),

[4] See Michael Hicks & Michelle Fitzsimmons, Self-Driving Cars: Your Complete Guide to Autonomous Vehicles, TechRadar (June 7, 2019),; Semi, Tesla (2021),

[5] See What is an Autonomous Car?, Synopsys (2021),

[6] See, e.g., Future of Driving, Tesla (2021), (visualizing the car’s recognition of objects in the world, as compared to the driver’s view).

[7] Id.; see also Dirty Tesla, FSD Beta Gets Updated with V8.2 Software–First Impressions Downtown | 2021.4.11.1, YouTube (Mar. 5, 2021), (demonstrating capability of the most-recent Tesla beta release of autonomous driving software, including many examples of reactions to poor human driving and degraded road infrastructure).

[8] See Maarten Vinkhuyzen, Tesla Dojo Supercomputer Explained—How to Make Full Self-Driving AI, CleanTechnica (Nov. 21, 2020),

[9] See Borg, Star Trek (2021),

[10] Shusuke Morimoto et al., Cybersecurity in Autonomous Vehicles, Intro. to Applied Informatics, May 2017, at 1–8.

[11] See id.; Rahul Razdan, Tesla Decepticons? Is Automotive Cybersecurity a National Defense Issue?, Forbes (May 2, 2020, 7:33 AM), (recounting demonstration of hacker control of a Jeep through its telematics system).

[12] Framework for Automated Driving System Safety, 85 Fed. Reg. 78,058, 78,064 n. 31 (proposed Dec. 3, 2020) (to be codified at 49 C.F.R. pt. 571).

[13] Id.

[14] See Jennifer Johnson & Thomas Parisi, IoT Update: FCC Proposes New Spectrum Plan for Vehicle Safety and Unlicensed Uses, Inside Tech Media (Dec. 4, 2019),

[15] See Jemima Meyers, Self-Driving Cars: How Automakers Can Overcome Cybersecurity Issues, Tripwire (Dec. 30, 2018),

[16] Framework for Automated Driving System Safety; Extension of Comment Period, 86 Fed. Reg. 7523 (proposed Jan. 29, 2021) (to be codified at 49 C.F.R. pt. 571); see also 85 Fed. Reg. at 78,058.

[17] See generally Anthony Branch, National Traffic and Motor Vehicle Safety Act, Britannica, (Mar. 16, 2018).

Image Source: