Artificial Intelligence Regulation in Wartime Governance: Adapting EU AI Act Approaches to Ukraine

The article examines the challenges of regulating artificial intelligence in public governance under martial law and substantiates the need to adapt risk-based approaches embedded in the EU Artificial Intelligence Act (EU AI Act), adopted by the European Union, to the specificities of Ukraine’s wartime context. The study aims to assess the limitations of directly implementing the EU AI Act under emergency governance and to develop an adaptive regulatory approach focused on digital resilience, security, and human rights protection. The methodology relies on an interdisciplinary framework and includes legal and regulatory analysis of the EU AI Act provisions, comparative analysis of international approaches to AI regulation, institutional and structural-functional analysis of wartime governance, and conceptual modelling. The findings indicate that the static risk-based logic of the EU AI Act is insufficient for wartime public governance marked by emergency powers, accelerated decision-making, and a growing role of algorithmic systems. Key regulatory constraints are identified, supporting a shift toward a flexible, context-sensitive regulatory approach. A model of flexible AI regulation for wartime public governance is proposed, grounded in proportionality, temporality, and reinforced human oversight. Scientific novelty lies in a conceptual approach to AI regulation under martial law that extends the EU AI Act framework by integrating wartime conditions, an adaptive regulatory loop, and a system of mandatory risk-limiting safeguards. The findings can inform the design of national AI policy and legal instruments for Ukraine during martial law and throughout post-war recovery. and legal instruments for Ukraine during martial law and throughout post-war recovery.