Liability for artificial intelligence

This chapter analyses liability for artificial intelligence from a comparative perspective, outlining existing rules and principles in major jurisdictions. Three principal areas of discussion are highlighted: the arguments in favour and against liability, the standard for liability (reasonable person or reasonable AI) including relevant circumstances and possible safety measures, the preferable liability regime (fault or strict liability). Alternative regulatory measures are briefly discussed, as are specific regimes for special kinds of AI and areas of application. In addition, several key issues for future regulation are identified: the need to differentiate between sectors of application and the kind of activity (commercial or private); how to deal with open source and free software; compensating pure economic loss. Future EU rules are referenced where applicable.