02.07.2026

Self-Driving Cars: A New Revolution for Motor Insurance

Self-driving cars are becoming a rapidly growing way to drive to your destination. With the benefits that come with driverless cars, such as safer and easier driving, the motor insurance world is stepping into a new realm of possible claims. Who is liable for a crash? Was there an intervention by the human? Was there a breach in cyber security? With these questions at hand, the implementation of self-driving cars could complicate the way motor insurance works. This article will tackle why driverless cars matter to actuaries, and the impacts this futuristic technology could have on insurance. The self-driving level of a car is measured through the SAE scale, ranging from level 0 (no automation) to level 5 (full automation). For this article, we are assuming these driverless cars are L4 and L5 on the SAE scale, promising automation across the entire journey.

Why do driverless cars matter to actuaries?

At present, risk factors are calculated by actuaries using data on the driver, such as their age, gender, driver history, location, and years of experience. But with the promise of driverless cars coming into effect in the future, many of these factors become obsolete. Simultaneously, new risks could emerge that could complicate motor insurance pricing as we know it [1].

A significant problem moving forward revolves around an increase in cyber risk. Driverless cars are expected to rely heavily on AI and continuous connectivity, which could make hacking of cars more prevalent. Criminals could exploit weaknesses in the software of these vehicles, allowing them to take control of a vehicle and redirect it to a new location of their choosing. This can create some very scary scenarios, including a greater potential for terrorist attacks, and a new method for criminals to steal cars [2]. If a driverless car brand proved to be vulnerable to these types of attacks, insurers may view this as an unacceptably high risk level. Ultimately, this could lead to them becoming uninsurable unless these vulnerabilities are addressed.

Autonomous vehicles provide a large amount of real-time information to process, which may require actuaries to have to be more confident in handling more complex datasets than they are used to in the traditional motor insurance world. In the future, actuaries may be required to have skills in machine learning and predictive analytics to interpret who is liable and interpret patterns that could have never been known with traditional motor data. This can be explained in further depth using the 5 V’s of big data: Volume, velocity, variety, veracity and value. As discussed, the volume of data required to interpret is much greater than traditional motor insurance data, which makes both storing and processing this data more difficult. Unlike traditional motor insurance, driverless cars continuously generate data through sensors and cameras. This creates a problem for actuaries to filter this large amount of data to identify what is relevant to the claim from a pricing and risk assessment perspective. The variety in these types of claims is also enormous, as data can be collected in many different formats e.g GPS, cameras, LiDAR (spatial data) and speed/braking sensors, leading to potentially more advanced data management compared to traditional motor insurance. Veracity refers to the reliability of the data. This becomes very important with driverless cars, where insurers must ensure that the data collected from the car’s systems are trustworthy and credible before being used for pricing. This topic is also relevant to fraudulent claims, which is touched on further later in this article. Value asks the important question of whether the data we have collected is actually useful. Although a large volume of data is collected, a lot of this may not contribute to an improved risk prediction. Therefore, actuaries must be accurate in identifying which variables add predictive value without unnecessarily complicating their models. As a result, using machine learning, predictive analytics and data engineering may align closely to the engineering world, leading to increased collaboration with engineers due to risk factors becoming more technical than behavioural.

Impacts on insurance

According to Orsa (a website related to road safety management)[3], it is estimated that 94% of car crashes are due to driver error. Driverless cars would largely remove the risk for driver error, shifting the liability risk away from the owner and towards the manufacturer, as they are partially responsible for the machines system and whether it crashes. However, the Automated and Electric Vehicles Act 2018 states that if not properly insured, the owner is still liable for the car. Therefore, although the owner is much less likely to be responsible for any damage caused to the vehicle, it is unlikely that this would cause a material decrease in premiums sold by insurance companies.

So, we know that driverless cars will eliminate accidents caused by human error almost completely, but does this truly suggest that these vehicles are completely safe? Autonomous vehicles rely on sensors to detect obstacles, pedestrians, and any other factors on the road that may cause it to change speed or direction. Despite the need for these sensors to work perfectly, concerns have already risen as these vehicles have been known to experience issues in adverse scenarios, such as fog, heavy rain, snow or unexpected road layouts [4]. So, despite all the talk from these companies on the reduction of accidents, will this fall be as drastic as it’s predicted to be? And even if the fall is drastic, we can still expect an increase in the claims amount due to the cost of repairing these complex systems if they were to be in an accident.

We must also consider the topic of subrogation, as driverless cars could cause this to become very complicated. Subrogation is the legal process by which an insurer can pursue the party responsible for the loss of the covered policyholder and recover claim costs from them. Driverless cars are expected to greatly reduce minor accidents but not have such a large effect on the more severe accidents, which would still be expected to occur. If large accidents occur in the future, many questions are raised regarding subrogation. Which manufacturers should be pursued? The AI software? The navigation system? The car itself? This question becomes even more complicated when the car is only semi-autonomous, because what level of liability then lies with the driver compared to the car? [5] With the cameras and systems in place, this is likely to help with evidence in these cases and reduce fraudulent claims simultaneously. But are these cameras even truly reliable? Because recordings can still be overwritten after a certain time passes. Furthermore, with the rise of artificial intelligence, it is becoming increasingly possible to manipulate recorded evidence. This creates additional challenges when determining liability as the footage must be verified and have certainty that the footage had not been modified. Therefore, although cameras may improve the evidence available to insurers, advances in artificial intelligence mean that the concerns surrounding the accuracy and credibility of evidence are not completely eliminated. Driverless cars will ultimately create additional challenges to subrogation such as deciding which party to pursue for the losses, and preserving the evidence needed for that claim.

As discussed earlier, with driverless cars will come a shift in risk factors, and this will most likely lead to a change in how we build our pricing models. In the future, the price of premiums is much more likely to be dependent on technological factors, such as sensors, AI, navigation systems, and system updates. Overall, the new exposure measures will result in pricing models needing to be recalibrated and possibly more prudence required due to the higher severity of claims and the uncertainty of the various factors that come with driverless cars. With this, driverless cars are likely to create a new segment in the insurance industry, as insurers will need to underwrite cyber insurance, product liability, and software liability much more commonly than before. This could see the potential for motor insurance to move away from personal lines and towards commercial lines, making business to business relationships more regular. This may cause actuaries to steer more towards using experience rating over exposure rating in the future.

Overall, driverless technology is likely to cause less frequent claims that are higher in severity, and with the potential complexity of these claims they could take longer to settle once reported. This would affect the net claim costs, and these costs would need to be modelled into reserves.

What the future looks like

Moving forward, we should prepare to expand our actuarial toolkit to ensure that we can broaden our network and collaborate with a wider range of companies. No matter how the questions raised in this article are answered in the future, insurers and actuaries will have a key role to play in the world of driverless cars. Driverless cars may change the industry, but they can create significantly more opportunities for actuaries to innovate and shape the future of the motor insurance industry.