Trust and Blame in Self-driving Cars Following a Successful Cyber Attack.

Even as our ability to counter cyber attacks improves, it is inevitable that threat actors may compromise a system through either exploited vulnerabilities and/or user error. Aside from material losses, cyber attacks also undermine trust. Self-Driving Cars (SDCs) are expected to revolutionize the automotive industry and high levels of human trust in such safety-critical systems is crucial if they are to succeed. Should adverse experiences occur, SDCs will be particularly vulnerable to the loss of trust. This paper presents findings from an initial experiment which is part of an ongoing study exploring how fully autonomous Level 5 SDCs would be blamed and trusted in the event of a cyber attack. To do this a future thinking-based methodology was used. Participants were presented with a series of randomly ordered hypothetical news headlines about SDC-cyber incidents. After reading each headline, they were required to rate their trust and assign blame. Twenty different hypothetical SDC-cyber incidents were created and manipulated between participants through the use of cyber security specific terminology (e.g. hackers) and non-specific cyber security terminology. This was manipulated to investigate whether the wording – i.e. being explicitly or overtly cyber (versus non explicitly or covert) of a reported incident affected trust and blame. Overall trust ratings in SDC technology in the context of a cyber incident were low across both conditions which has the potential to impact uptake and adoption. Whilst there was no significant overall difference in trust between the overtly and covertly cyber conditions, indications for further lines of inquiry were evident – including differences between some of the scenarios. In terms of blame, attribution was varied and context dependent but across both conditions the SDC company was blamed the most for the cyber incidents.