Liability for damages caused by the use of AI: objective, subjective, or hybrid liability?

Abstract

The outbreak of artificial intelligence in everyday and professional life has created unprecedented challenges for law, particularly in terms of damage and liability. The autonomy and learning capacity of AI systems make it difficult to apply traditional regimes of strict and fault-based liability since the identification of the responsible subject and the reconstruction of the causal chain are hampered by the intervention of multiple agents and the unpredictability of the outcomes.

In this context, the distinction between autonomous and supervised AI systems, as well as the degree of human intervention, are decisive in assessing the attribution of responsibility. A hybrid and flexible approach, which allows for intelligent transition between aspects of the objective and subjective liability models, seems appropriate in view of the multiple challenges and peculiarities of AI.