Research and analysis of international standards and regulatory requirements for artificial intelligence se-curity, development of a security model for Ukraine

Abstract

The article presents a comprehensive analysis of international approaches to the regulation of artificial intelligence (AI) security and the potential for their application within the Ukrainian context. It examines contemporary cybersecurity challenges in the era of AI development, including threats of data breaches, attacks involving generative models, and the misuse of deepfake technologies. Key international standards are analyzed, such as the EU AI Act, the NIST AI Risk Management Framework, ISO/IEC 23894, as well as regulatory approaches implemented in the United States and the United Arab Emirates. Special emphasis is placed on the importance of a multi-level AI risk management system that takes into account technical, ethical, legal, and social aspects. Particular attention is given to the study of Ukraine’s strategic documents aimed at developing a national AI policy aligned with European requirements. The concept of gradual implementation of ethical, legal, and technical norms into AI regulation is highlighted, through mechanisms such as certification, the establishment of regulatory sandboxes, and public consultations. The proposed author's model for AI security in Ukraine is based on the principles of multilateral interaction among the state, businesses, the academic community, civil society, and international partners, involving the active engagement of all stakeholders in the process of policy and standards development. The model envisions a risk-oriented approach to the development and deployment of AI systems, the introduction of a public ethical AI registry, the participation of citizens in audit and monitoring processes for high-risk systems, and the promotion of the responsible use of emerging technologies in socially significant sectors. The article emphasizes that Ukraine has a unique opportunity to offer the world its own approach to the secure and ethical use of AI, which is flexible, open, and adaptive, based on the principles of trust, responsibility, digital resilience, and respect for human rights.