Governance of AI and Agentic Systems: Challenges and Methodologies

Управление ИИ и агентными системами: проблемы и методологии
Milena Kumurdjieva, Lyubka Doukovska, Nehla Ghouaiel, Ahmad Dhanani
2025-11-06

AI governanceAI lifecycleagentic AI systemsnormative governanceproactive risk management
The accelerating advancement of artificial intelligence (AI), coupled with the rise of agentic systems, is reshaping the landscape of opportunity and risk for both organizations and society at large. While AI governance frameworks have evolved to address ethical, legal, and operational challenges, existing models often fall short in managing the autonomy, adaptability, and emergent behaviors of agentic AI. This paper examines the limitations of current governance approaches and introduces comprehensive frameworks tailored to both traditional and agentic AI systems. Our proposed models integrate principles of transparency, accountability, security, and resilience across the AI lifecycle, emphasizing proactive risk management, stakeholder engagement, and normative governance. By aligning technical safeguards with ethical imperatives, the frameworks aim to foster trust, mitigate systemic risks, and ensure responsible innovation.
1
Existing AI governance models are insufficient for agentic systems because they inadequately address autonomy, adaptability, and emergent behaviors.
2
The frameworks align technical safeguards with ethical and normative governance to foster trust, mitigate systemic risks, and support responsible innovation.
3
The paper introduces governance frameworks tailored to both traditional and agentic AI systems across the full AI lifecycle.
4
The proposed frameworks integrate transparency, accountability, security, and resilience with proactive risk management and stakeholder engagement.

traditional and agentic artificial intelligence systems

governance frameworks addressing the autonomy, adaptability, emergent behaviors, ethical, legal, operational, security, and resilience risks of AI systems across their lifecycle

Publication Details
Publication Date
2025-11-06
Journal
International Conference on Big Data, Knowledge and Control Systems Engineering (BdKCSE)
Publisher
Institute of Electrical and Electronics Engineers
ISSN
Cited by
7
Access Type
Author Information
Authors
Milena Kumurdjieva
Lyubka Doukovska
Nehla Ghouaiel
Ahmad Dhanani
Explore further
Open the scid.ai AI chat with a ready-made request: it will find papers on a similar topic and help build a literature review.
Find similar papers in the chat →
Make a presentation
100%