AI-based cybersecurity anomaly detection for OT systems in the EU Resilience Framework, an illustrative simulation-based methodological study

Hunorfi, Péter (2026) AI-based cybersecurity anomaly detection for OT systems in the EU Resilience Framework, an illustrative simulation-based methodological study. BELÜGYI SZEMLE / ACADEMIC JOURNAL OF INTERNAL AFFAIRS: A BELÜGYMINISZTÉRIUM SZAKMAI TUDOMÁNYOS FOLYÓIRATA (2010-), 74 (10). pp. 2809-2827. ISSN 2062-9494

Preview

Text 08_Hunorfi_AI_Cybersecurity.pdf - Published Version Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (416kB) | Preview

Abstract

Aim: To examine the possible role and constraints of Artificial Intelligence in anomaly detection for Operational Technology considering the EU critical entity resilience framework and concentrating on the European Commission’s 2026 guidance.Methodology: A focused reading of the Critical Entities Resilience Directive, CER guidance, NIS2 and the EU’s AI Act is combined with an illustrative synthetic experiment. Three established detectors and a fixed-weight combination are evaluated on 7,481 aligned test timestamps.Findings: The combined detector achieved an F1 score of 0.700 but missed many attack-labelled observations and incorrectly flagged some normal observations as anomalous. These outcomes illustrate detection limits and do not demonstrate improved service continuity, effective oversight by qualified personnel or regulatory compliance.Value: This study links an accessible technical example with practical questions about integrated resilience measures, alert handling, accountability of responsible personnel and documentation. Its contribution is a bounded applied interpretation, not a new detection algorithm.

Actions (login required)

View Item