The hurdles of artificial intelligence wrongdoing

Abstract

The prompt progress of artificial intelligence technology has triggered novel intricacies within the legal domain. This thesis explores the hurdles that may arise from AI-associated harm in the context of civil liability. At the outset, it investigates the calls for bestowing AI entities with legal personhood, a step that entails considering them as legal actors. Afterwards, considering the current classification of these entities as mere things and tools, the study examines different liability regimes that could be relevant in instances where these systems inflict harm. It assesses the viability of these regimes in ensuring fair compensation to victims, encompassing both contractual and non-contractual contexts. In conclusion, given that AI possesses unique attributes, I contend that the existing legal frameworks may lead to unsatisfactory results and leave liability gaps. Addressing AI legal implications thus entails the adjustment of liability rules to the age of AI, while acknowledging that a one-regime-fits-all solution may not be the optimal approach to governing AI-induced harm.