The AI Act: A realpolitik compromise and the need to look forward

First-generation technology regulation typically attempts to strike the right balance between rights protection and innovation. This tension is evident in the EU AI Act and in the way the risk management, the core element of any technology regulation, is framed. This chapter outlines the rationale behind the compromise solution adopted by the EU legislator to reconcile the protection of fundamental rights with the expected benefits of AI. It also discusses the decision to depart from a more holistic approach centred on the societal acceptability of AI, in terms of alignment with the values of the communities in which AI solutions are to be implemented. The chapter highlights the weakness of a primarily risk-based approach that does not place at the heart of the regulation the definition of key principles specifically tailored to the AI context and aimed at underpinning its development. Against this background, the role of fundamental rights in guiding the development of a human-centred AI in line with EU values is crucial. However, the implementation of the fundamental rights impact assessment in the AI Act is still underway. A more coherent framework is needed, combining the different assessments outlined in the AI Act, as well as a better definition of the scope and relevant criteria for the assessment. Finally, an appropriate model should be developed and made available AI providers and deployers, adopting a lean assessment design and combining expert-based evaluation and stakeholder/rightsholder participation.