Assigning Liability in the Context of Modern Artificial Intelligence

There is ambiguity surrounding the legal implications arising from or connected to the use of artificial intelligence ('AI'). The central objective of this thesis is to respond to the urgent need to address how legal liability should be assigned in instances where AI systems form, act or execute decisions. In addressing this question, a new hierarchical account of legal personhood is first proposed as a method for legally characterising AI. This account is then used to offer a nuanced and flexible three-phase approach to liability; one that emphasises the need to consider AI in context and resolve liability situationally.