Agentic Artificial Intelligence and ethical sovereignty: A framework for global AI governance

Abstract

The growing adoption of artificial intelligence (AI) in critical domains such as healthcare, cybersecurity, and governance expose the limits of current regulatory models, which treat AI as a passive tool rather than a moral participant. Existing frameworks, including the EU AI Act, OECD AI Principles, and AU AI Strategy, emphasize compliance and risk control but overlook AI’s emerging autonomy and ethical agency. This paper proposes the Agentic Artificial Intelligence Framework (AAIF), a model that repositions AI as a moral co-governor within socio-technical systems. Grounded in Ubuntu ethics and global governance theory, the AAIF integrates three moral dimensions: autonomy, accountability, and adaptability to support ethically resilient and culturally responsive governance. Developed through the Design Science Research methodology, the framework embeds ethics within institutional and policy design. By harmonizing legal, cultural, and ethical imperatives, the AAIF advances a pluralistic approach to AI governance that strengthens ethical sovereignty, particularly in Global South contexts. Future research will focus on simulation-based and policy-level validation of the framework in applied domains such as healthcare cybersecurity.