Exploring the risk governance mechanisms under the forthcoming EU Artificial Intelligence Act

Artificial intelligence (AI) promises great opportunities but also considerable risks for government and society.

How to properly govern this fast-moving and complex technology is a significant challenge, which in the European Union will be addressed by the forthcoming AI Act.

While the ethical principles and legislative proposals that have been advanced by the Act have gained much attention, little is known about its supporting harmonised standards or how its formal and informal governance mechanisms will relate.

This project responds to this knowledge gap by developing sociological theory about the AI Act´s risk governance model.

Drawing on social scientific literature on AI, risk, expertise, standards and governance, it will investigate three issues: i) the major perspectives that shape the AI Act, ii) how its laws and standards interact, and iii) to what extent it meets the challenge of ensuring trust(worthiness) in AI.

The empirical material will consist of official and supporting documents, and interviews with three target groups: the AI High-Level Expert Group, law makers and stakeholders involved in the Act´s development, and relevant experts at the European standards organisations CEN and CENELEC.

Outputs will include five peer-reviewed research publications, and a report and workshop aimed at disseminating project findings to a broad audience. The research will take three years and will be hosted by the Department of Technology and Society, Lund University.