THE RACE TO AI REGULATION: IS A HUMAN RIGHTS APPROACH FEASIBLE?

Abstract

Against a background of global competition to harness the potential of artificial intelligence (AI), there is also currently competition to secure a first mover advantage in the AI regulatory field, known as a ‘race to AI regulation’. In 2021, the European Commission released proposal for an AI Act, which is the first comprehensive attempt for legally binding regulation of AI. The draft AI Act puts forward a risk-based approach to the regulation of AI. The present dissertation critically evaluates the proposed risk-based approach and emphasises its shortcomings, presents a possible human rights approach to AI regulation and discusses whether that approach could be aviable alternative to the risk-based approach in the draft AI Act. It is argued that the risk-based approach in the draft AI Act is problematic as it creates dangerous loopholes and is based on a model of risk regulation that is not suitable for the AI context. The proposed human rights approach offers a viable alternative to the risk-based approach in the draft AI Act, but it does not provide a silver- bullet solution, so a hybrid approach to AI regulation is needed.