Self Driving Cars x Crime

When it comes to the impact of self-driving cars on crime, there are two general threads of discussion: the possibility of re-engineering cars for criminal purposes, and the opportunity to reduce traffic enforcement in the absence of speeding and parking tickets. One thing that has gotten less attention is the possibility for reduced anonymity in — and therefore greater prevention of — violent crime.

The FBI has voiced concerns that self-driving cars could free up criminals in a getaway car, or be loaded with explosives to carry out a car bomb without a suicidal driver (see this and this). While these are not scenarios to take lightly, both of these concerns generally presume that car ownership will not fundamentally change when cars become autonomous.

If Uber, Google, and even the City of LA have their way (as recent studies suggest), though, car ownership will decline dramatically when autonomous cars hit the road. Without the need for human drivers, Uber can cut costs and pass those savings onto the consumer, making it impractical to own a car.

If the norm is no longer to own a car, then two things will happen:

First, all trips would be registered in a database of some kind. Right now, every time I take an Uber or Lyft, those companies know exactly where I am for that period of time. They know where I was picked up, when I was picked up, the route I took, and my dropoff information. In a world where I rely on these services exclusively, my whereabouts could be determined at almost any point in the day. (This could happen even without a decline in car ownership. As cars become internet enabled, they’re already collecting data. Phones already track you too).

Second, those who decide to continue owning a car would fit into certain profiles. The car enthusiast, the hobbyist, the ridiculously frequent traveler, guy with too much money, the guy who lives in a car…and more insidious profiles: the hacker. With the decline of car ownership, many car dealerships would close, making it more difficult to own a car in the first place — and more noticeable when you do.

There’s an analogous set of profiles among non-owners that could develop, and would make certain behaviors more “noticeable”: the reduced need for car storage, and the rise of shared rides. As more and more rides can be shared, it would be both riskier to share a ride while committing a crime, and more noticeable when you opt to ride alone. (Note that for certain people, particularly the wealthy, riding alone will be the norm, and it’s difficult to know what carpooling rates would be in off-peak times.) Secondly, as most people don’t regularly need their trunk, most cars would shed their trunks and special cars would be ordered for travelers. This means that someone who orders a solo ride with a trunk would be less anonymous than your typical, trunk-less carpooling rider. (For a given hour and 2-block radius, this could be an identifying trait).

The reduction of anonymity might concern privacy enthusiasts, but I’m optimistic that both of these trends would make it much more difficult for a criminal to involve a car in certain types of crime.

  • Kidnappings: All cars that passed through a given location at a given time would be known. If a pickup was ordered in the vicinity of a victim’s last known location, the requester could be identified (in this case, an accomplice could drive through, but their whereabouts would be known as well as the fact that the car had stopped) (think about how helpful this could have been in the Lauren Spierer case)
  • Bank Robbery, any crime requiring a speedy getaway: Criminals would not be able to control the speed of the car, and the identity of the ride requester could be determined quickly
  • (Physical) Car Hijacking – (in the Tsarnaev fashion) – Cars that are re-routed could be identified, as well as any car passing through the vicinity. Had the Tsarnaevs relied on self-driving cars, they may have still successfully taken a guy hostage, but they would not have been able to zoom off at 90 miles per hour, and their constant, immediate location changes would have probably set off a crime sensor. The car would have also been programmed to stop for law enforcement.
  • (Technical) Car Hijacking/Hacking – Hackers were able to exploit a vulnerability in car software to remotely hijack a car. They were able to do this because they had two years to tear the car apart and dig into the car’s software. With a decline in car ownership, however, getting familiar with the software would be far more difficult to do under the radar. Not only would you have to tamper with the software for an extended period of time undetected, but you’d also need to test its interaction with the hardware, which would likely require a physical car to experiment with. If this becomes a national security concern, the government may decide to regulate who can physically own a car, and the owner of a compromised vehicle could be easily identified (which is already possible).
  • False Alibis – A suspect’s location could be verified in court proceedings if their ride history was obtained with a warrant (“Where were you on the night that Katelyn Stark disappeared?”)

Of course, there are simple workarounds for some of these tracking mechanisms. Phone theft would enable criminals to order rides without being tracked as themselves, but a simple phone password could at the very least deter a criminal from getting away as quickly as they could today. Simple security measures — verify a password when you get into a car, for example — could further thwart criminals.

Self-driving cars could help reduce these types of crime even without complicated crime detection technology. With more advanced crime determent systems, it would be even more powerful: auto-detecting explosives. Detecting when someone is tampering with the software. Detecting a body in the trunk. Using big data to track suspicious behaviors.

These changes could take years to unfold, and along the way we will discover that we haven’t even scratched the surface of how self-driving cars will change the nature of crime. In the meanwhile, the most important thing that autonomous car manufacturers can do is prevent new forms of crime from developing through hacking, hijacking, and re-engineering; take extra precautions to protect the vast amount of sensitive information they will be collecting, and take advantage of the data being collected to prevent accidents and save lives.