From Technocracy to Democracy: Recalibrating AI Governance?

AI ethics has recently emerged as a dominant governance paradigm, increasingly implemented through institutionalised, specialised and expert-driven tools and mechanisms, thereby compelling us to revisit the question of democratic deficit. This ethics-oriented technocratic approach is exemplified by the EU AI Act (2024), which promotes soft-law tools as part of a layered, risk-based governance framework. This article critically examines whether that model can address the democratic deficit in AI governance, focusing on harmonised standards, voluntary codes of conduct and regulatory sandboxes. Drawing on democratic theory and governance scholarship, it explores the extent to which these instruments advance participation, deliberation and accountability. By assessing these mechanisms as key sites of governance, the article concludes that they prioritise flexibility, technical expertise and market integration over public contestation and participation, offering little response to the democratic deficit. It argues that AI ethics can address this deficit only when complemented by institutional reforms that embed deliberation, broaden participation and ensure meaningful public accountability, including the involvement of non-expert citizens.