Machine Learning will seriously change the automotive industry and its security
May 2020 by Kaspersky
The Experiences Per Mile Advisory Council, which unifies experts from the car, automotive and tech industry, has recently published a forecast on vehicle connectivity and the surrounding customer experience. According to the report, today 48% of all new cars globally include built-in connectivity, but by 2030 that figure will rise to 96%. Similarly, by 2030, 79% of vehicles shipped around the world will have an L2 autonomy or higher.
The report also says that customer expectations are shifting from just “smart technologies” to a connected experience, including vehicle maintenance. As such, 57% of European and 80% of North American respondents are interested in early detection of necessary maintenance and repairs; 80% of respondents were willing to share anonymous or personal connected car data to gain access to such capabilities. Big data allows automakers to predict the maintenance and repair needs of their vehicles, in turn enabling dealerships to be optimized and downtime to be minimized.
Alexander Moiseev, Chief Business Officer at Kaspersky comments:
“Predictive analysis based on big data allows manufacturers manage many car functions and related operations. For example, machine learning in the automotive industry enables predictive maintenance that influences the whole management process of stock and logistics of spare parts. Machine learning can also change the driving experience, for example, by adjusting engine electronics based on driving style or optimizing many other parameters. The system will be able to predict the need for adjustments rather than just react to current factors.
Though these technologies for connected cars are at the early stage of development, it is already important to think about possible security risks they can be exposed to. In particular, machine learning can be compromised by adversarial technics, such as poisoning of machine learning-algorithms - when spoiled data is used for algorithm training.
Another possible risk for data-driven systems is the safety of various types of customer data – from driving parameters to personal data. It is collected by a car manufacturer and transferred to third party providers of services, such as insurance or maintenance. This may result in a new security issue of private data exfiltration from a vehicle and associated elements of the ecosystem.
These risks should be considered while car connectivity goes to a new level, as the recent report by the Experiences Per Mile Advisory Council has revealed. Kaspersky is also focused on addressing cybersecurity needs for connected cars and develops dedicated protection solutions and expertise for the automotive industry.”