Az EU AI Act hazai implementációja: Jogi, kormányzási és biztonsági kockázatok a mesterséges intelligencia megfelelésében = Implementation of the EU AI Act in Hungary: Legal, governance and security risks in Artificial Intelligence compliance

Herman, Zsuzsanna (2026) Az EU AI Act hazai implementációja: Jogi, kormányzási és biztonsági kockázatok a mesterséges intelligencia megfelelésében = Implementation of the EU AI Act in Hungary: Legal, governance and security risks in Artificial Intelligence compliance. SCIENTIA ET SECURITAS. ISSN 3057-9759 (In Press)

Preview

Text 112-article-10.1556-112.2026.00331.pdf - Published Version Available under License Creative Commons Attribution. Download (340kB) | Preview

Abstract

A tanulmány azt vizsgálja, miként illeszkedik az Európai Unió mesterséges intelligenciáról szóló rendelete (EU 2024/1689 – AI Act) és annak hazai végrehajtása, a 2025. évi LXXV. törvény a magyar jogi és intézményi környezetbe. A kutatás különös figyelmet fordít a kockázatalapú szabályozás logikájára, a trustworthy AI elvének érvényesülésére, valamint a human-in-the-loop követelmény gyakorlati megvalósítására. A vizsgálat kvalitatív jogelemzési és szabályozás-összehasonlító módszereket alkalmazott. Eredményei szerint Magyarország jelentős előrelépést tett a hatósági és infrastrukturális végrehajtás terén, azonban a szervezeti kultúra és az etikai érettség fejlettsége még nem éri el a kívánatos szintet. Ez a hiányosság biztonságpolitikai kockázatokat is hordoz, különösen a kritikus infrastruktúrák és a közszolgáltatások területén. A tanulmány javaslatokat fogalmaz meg az etikai irányelvek fejlesztésére, a szervezeti integritás erősítésére és az MI-compliance rendszerek mélyebb beágyazására a hazai kormányzási gyakorlatba. | This paper explores how Regulation (EU) 2024/1689 on artificial intelligence (AI Act) is being implemented in Hungary and how the European Union’s risk-based governance model is translated into the national legal and institutional landscape. It examines the intersection between legal compliance, organizational governance, and securityoriented risk management in the context of AI technologies. The central question concerns how the EU’s trustworthy AI paradigm—grounded in legality, transparency, accountability, and human oversight—can be effectively operationalized within Hungarian administrative and organizational practice. Methodologically, the study applies qualitative legal interpretation, comparative regulatory analysis, and document-based content examination. It draws on EU-level materials (AI Act, EDPB guidelines, OECD AI Principles), international standards (ISO/IEC 42001:2023), and national legal instruments such as Act LXXV of 2025 and Government Decree 344/2025 (X. 31.). This multi-layered approach allows for a comprehensive assessment of the emerging governance architecture surrounding AI regulation in Hungary. The findings show that Hungary has made notable progress in establishing the institutional infrastructure required for AI oversight—creating the Hungarian Artificial Intelligence Council, introducing a regulatory sandbox, and designating competent authorities. These measures align the national framework with the EU’s risk-based logic of classification, conformity assessment, documentation, and human control. Yet, the research also reveals that domestic compliance practices remain largely formalistic and enforcement-driven. While the regulatory framework is strong, the internal culture of ethical awareness, proactive risk management, and transparency within organizations is still at an early stage of development. A key contribution of this study is the identification of a structural imbalance between legal-institutional maturity and the ethical risk culture of organizations. This gap represents a latent security vulnerability: high-risk AI systems demand not only regulation and supervision but also continuous ethical reflection, built-in auditability, and ethicsby-design principles. Without strengthening human-in-the-loop mechanisms and clear accountability structures, the full implementation of the AI Act risks remaining procedural rather than substantive. The paper concludes that enhancing Hungary’s AI governance capacity requires a shift from formal compliance to value-based integrity governance. It recommends embedding AI compliance into organizational management systems, fostering interdisciplinary competence centers, supporting empirical research on AI governance, and cultivating a national AI culture that values transparency, accountability, and human-centric oversight. Strengthening these dimensions is essential not merely for regulatory alignment but for sustaining public trust, institutional integrity, and long-term societal security.

Actions (login required)

Edit Item