Three types of defects commonly recognized in product liability cases are manufacturing defects, design defects, and informational (or failure-to-warn) defects. Deciding whether automated cars are defective under each of these theories will raise unique issues.

Manufacturing Defects

The first—manufacturing defects—is the most likely to remain unchanged in the face of product liability lawsuits involving automated or semi-automated vehicles. If the alleged defect is one of manufacture, the plaintiff will still have to prove that a manufacturing error occurred, resulting in an unreasonably dangerous condition that caused the plaintiff’s injuries. The presence of manufacturing defects, particularly ones that affect the designed choices the car makes when facing obstacles, is likely to result in greater injury. A driver cannot “steer” out of such a situation.

Design Defects

Design defects are likely to prove challenging, too. In a design defect case, a plaintiff generally must prove that the product had a defect in its design that rendered the product unreasonably dangerous, and that the defect caused the plaintiff’s injuries. Most states determine defect by reference to a risk-benefit test (asking whether the design’s risks outweigh the design’s benefits) or a consumer-expectation test (asking whether a reasonable consumer would expect the unreasonably dangerous condition). Applying these tests to questions such as whether the vehicle is defective and unreasonably dangerous because it was programmed to hit a cyclist who darts out into traffic, rather than veer into oncoming traffic and potentially create a multi-vehicle accident, is hard. Under the risk-benefit test, should a jury be looking to the risks and benefits of the motoring public at large, the user of the automated vehicle, or the cyclist? Veering into oncoming traffic may benefit the cyclist but could result in the injury or death of the user, as well as the occupants in the oncoming vehicle. (And if that vehicle is automated, its response to the situation also may play into the resulting outcome and further complicate the design defect inquiry.) Application of a consumer expectation test raises even more questions. An ordinary consumer may expect that the vehicle will avoid collision with the bicycle, but what if that course of action creates more harm to the user and occupants of other vehicles? Overall, establishing an unreasonably dangerous condition of design will be difficult because in many cases, the manufacturer’s programmed choices will most likely be reasonably safe, even if an injury results. Not every accident can be avoided simply by eliminating driver error. Myriad environmental factors exist—such as animals, cyclists, icy roads, or the presence of non-automated vehicles—that may cause accidents and injuries.

Failure-to-Warn Claims

Failure-to-warn claims also will raise complicated issues. Under a failure-to-warn theory, a plaintiff must prove that the manufacturer failed to provide instructions or warnings, the absence of which rendered the product unreasonably dangerous and caused the plaintiff’s injuries. As an initial matter, the contours of a manufacturer’s duty to warn users of automated vehicles needs to be established. Will a manufacturer have to disclose how an automated vehicle will respond in various situations? If it does, how would a plaintiff link that causally to whatever occurred? And which situations will require disclosure? Will that require proving that the likely human response would have resulted in a better outcome? Moreover, in vehicles that are not yet fully autonomous but require human intervention under certain circumstances, will there be warnings related to the amount of attention required to use the vehicle?

Allocation of Fault and Insurance

Lawsuits involving automated vehicles also will raise questions about allocation of fault. If the car was programmed to execute an action that resulted in an accident, it will be difficult for a car manufacturer, as a defendant, to assert that a driver, or user, bears responsibility. But in the case of vehicles that are not fully automated, how can fault be allocated between a driver and the vehicle? Additionally, in collisions involving a fully-automated and a non– or partially–automated vehicle, will a fact finder be more likely to find fault with the driver of the non– or partially-automated vehicle? Moreover, will jurors assume that an automated car should avoid all accidents and injuries, such that when an injury occurs, the manufacturer must be responsible (thus turning automobile product liability into a nearly res ipsa type analysis)?

Insurance likewise will become a complicated issue. Insurers may carve out responsibility for accidents involving automated cars on the pretense that the driver did not cause the accident. This may make the injured person’s recovery dependent on proving a product liability claim. It also may make car manufacturers virtual insurers of their products, something that product liability law has sought to avoid. These are legal issues with important public policy consequences that will need to be addressed.

Conclusion

Ultimately, automated vehicles hold great promise in reducing collisions and associated injuries and deaths. Automated vehicles, however, cannot eliminate all accidents or injuries, and thus product liability lawsuits are not going away. Therefore, product liability law will have to evolve to account for this changing landscape.