The German cyber security authority (BSI) and ZF Friedrichshafen AG will investigate how the safety of artificial intelligence systems in cars can be tested. The results will be used to develop test criteria for AI-based systems. The aim is to develop future car safety audits and advance international standardisation efforts in this area.
AI-based systems, such as those that monitor driving paths or detect pedestrians, can be vulnerable to a wide range of attacks that can be dangerous to drivers and passengers as well as pedestrians and people in other vehicles. Cybercriminals may, for example, attempt to make the AI ‘blind’ to other objects in the driving lane by manipulating computer vision (the image recognition system) to misclassify pedestrians. Such a vulnerability poses a serious threat, especially for autonomous vehicles that could soon be on the roads.
BSI said that its aim is to develop guidelines based on the knowledge gained from the research. The authority plans continuing the project to implement the guidelines developed.