However, the report says, "There is no question that electronics of the car will become a target for malicious hacking attacks. This becomes increasingly feasible when the car is wirelessly connected to various types of networks."

Automotive risk managers will have to ensure products are reinforced with hardware-based, tamper-proof security, as well as highly sensitive software and data integrity controls that detect unauthorized access. 

"The self-driving car will have large programs and complex software code that will require extensive testing to make the software reliable," says IHS. "It is extremely difficult to test the SDC software for all events that could happen. The solution is to use a software architecture that has built-in fault tolerance to limit bad events if something goes wrong."

Only three states currently have laws for safety testing for self-driving cars, posing legal questions that must be untangled and established before a framework for lawmakers and insurers is established.

For example, how will liability for accidents between SDC and manually driven cars be determined? Currently, autonomous capabilities in most vehicles are at what IHS calls "level 0," or simple driver-assist warnings for lane departures or forward collisions, or "level 1" with lane keep assistance or autonomous braking features to avoid crashes.

As the technology becomes more intuitive with time, the lines between decisions made by driver and machine will blur. 

IHS says self-driving cars are already a question of "not if, but when", however, there are many miles to go before insurers and agencies have to take the wheel in offering SDC coverage to the average consumer.