Artificial Intelligence (AI) is a powerful technology that will play a central role in mobility of the future. The breakthroughs achieved by AI systems in automated driving are contrasted by their higher complexity and opacity, potentially leading to safety and cybersecurity risks. The growing digitalisation of vehicle systems increases the potential cybersecurity attack surface and can lead to stronger impacts if automation mechanisms are compromised. Biases in and lack of generalisation of AI systems and data may cause vehicles to malfunction in natural conditions or as a result of a cyber-attack, putting the lives of passengers and road users at risk. Vehicle testing procedures need to take into account the specificities of AI to ensure that safety and cybersecurity risks are properly addressed. Addressing by design the AI safety and cybersecurity challenges is key to securing the many benefits that automated driving can bring to society.
Number of pages