AI self-driving cars present a host of new legal issues

What San Francisco stopped as too dangerous, Los Angeles has now leaped into.

Last week’s Public Utility Commission’s approval of Google’s WAYMO self-driving taxis in Southern California opens a Pandora’s Box of legal issues. San Francisco has already confronted the problems of self-driving cars in the collapse of GM’s Cruise program. GM’s Cruise experimental fleet of self-driving cabs came to a screeching halt this past October when a Cruise robo taxi dragged an elderly pedestrian twenty feet after she was hit by a human-driven car and thrown into the path of the unsuspecting and computer driven Cruise vehicle.

Attempts to achieve a self-driving vehicle using artificial intelligence has long been a dream. Elon Musk promised in 2015 that it would be commonplace by 2017. That didn’t happen. Experimentation in self-driving trucks is just beginning to occur in places like Arkansas. A host of auto and truck manufacturers offer an abbreviation of AI driving in having a vehicle self-parking or steer in cruise control on highways. Not just Tesla, Apple, or GM, but also Ford are looking for more autonomous AI opportunities in driving vehicles. The path is not without its problems.

Neither the manufacturers, the regulators, the statutory legal structure nor the court’s common law structure are even close to being fully prepared for a world of vehicles driven by computers rather than humans.

First is the obvious question of safety and liability. San Francisco pulled the plug after that Oct. 2 accident. The simple fact is that artificial intelligence still has room for growth and increased sophistication. For example, a human being instinctively knows that if a ball bounces across the street there may be a little child running after it moments later, and so proceeds cautiously. AI-driven cars are not so prescient.

Moreover, our statutes have not developed to the point of dealing with the liability for AI-driven cars. While a vehicle owner as well as a vehicle operator not only have common law but statutory duties of care, as set forth in volumes by California’s Vehicle Code Sections 2400-28081, what about the passive individuals who rent one of these WAYMO AI cars? Can they sit in the comfort of the back seat and snooze, or must they be in the front, ready to grip the control at any moment?

AI vehicles currently provide for and rely on human intervention. For example, Tesla’s instructions for AI-driven vehicles require the “passenger” who turns the driving over to the computer to remain “fully attentive” and be able to intervene to take back control. Tesla successfully relied upon that operation instruction to achieve a favorable jury verdict when a Tesla owner was found to be inebriated and could not successfully “intervene” before the fatal crash.

Even limited operational AI vehicles currently have instructions that say, “Please keep your hand on the wheel. Be prepared to take over at any time.” So much for that relaxing AI taxi ride.

Indeed, GM’s Cruise operation reported one incident of the need for human intervention, requiring the passenger to assist every fifteen to twenty operations. By the same token, California Vehicle Code Section 21701 provides that no person may willfully interfere with the “driver” of the vehicle “in a manner that affects the driver’s control of the vehicle.” Who is the driver? The Google computer or the flesh and blood passenger?

And is the passenger who has the obligation to take back control required to have a valid California driver’s license? So much for having that Waymo taxi pick up your underage child from school and drop the youngster on your doorstep unaccompanied by a minor. Equally

problematic is using a Waymo or other unaccompanied AI vehicle to act as your “designated driver” when you leave that party.

Will auto and/or driver insurance policies cover AI vehicles? This is unclear and terms need to be written in policies.

Right now, the primary statute allowing autonomous vehicles is California Vehicle Code Section 38750, which require AI vehicles to have a permit from the California Department of Motor Vehicles to operate on public roads. It also dictates that the autonomous manufacturer “Require(s) the operator to take control of the autonomous vehicle.” Section 38750(c)(1)(C)(i), and that the vehicle comes to “a complete stop” if the operator does not or cannot take back control. Section 38750(c)(1)(C)(ii). However, the PUC approval specifically talks about “driverless” vehicles (which of course is the case when the Waymo cab drives from its last stop to pick you up, it is indeed “driverless.”)

No doubt there will also be a host of design “defect” litigation arising from the proliferation of self-driving cars. Google and before it, General Motors, were willing to take on those risks given the long-tern potential for AI vehicles. What about insurance companies? And how far up the development chain does the liability extend? Do the software subcontractors have liability? And more importantly if a start-up subcontractor, do they have assets or insurance to respond to liability?

Artificial intelligence promises to make our roadways safer and smoother. It can anticipate traffic delays and reroute cars automatically. It can save labor costs not only by replacing taxi, Uber, or Lyft drivers, but can also replace truck drivers. AI-driven trucks are being used in places like Arkansas, where Walmart is experimenting with AI-driven vehicles for short-haul freight operations between warehouses. Less traffic congestion, less need for overwhelmed parking facilities. Most of all, the avoidance of LA’s crowded freeways seems to be the dream of invoking AI in vehicles. But this still presents other legal problems.

What about privacy? In an AI cab, the computer will keep track of when you leave, where you go, and when you arrive. Granted, that access to data exists now when you use Uber or Lyft, but generally not in your own vehicle. Jurisdictional issues remain as well. In Northern California, certain communities vigorously fought the onslaught of self-driving cabs. The PUC experimental approval for Southern California includes not only Los Angeles but a host of other Southern California cities, such as affluent Beverly Hills, and allows speeds of up to 65 miles per hour. Can a city override the state PUC or DMV with its own regulations? Like no parking of a Waymo cab in the confines of Beverly Hills? There is at least one move afoot in the Los Angeles County Board to ask the Legislature to allow local jurisdictions to require valid local permits for AI taxicabs just like any other taxis. So far, the PUC has previously rejected protests by local communities such as San Mateo, but this issue has not been tested in the courts. Yet.

While AI provides tremendous promise and opportunity, in addition to the need for further technical expertise and sophistication, AI self-driving vehicles are still at the nascent age of development of legal rules and liability, all of which have yet to be devised not just by government regulators and the state legislatures but by the courts and us lawyers.