Bridging policy, practice, and pedagogy: a multi-level framework for Responsible AI adoption in higher education assessment (ML-RAAF)

The rapid integration of artificial intelligence (AI) into higher education assessment offers opportunities for efficiency, personalisation, and innovation, while raising concerns about academic integrity, fairness, transparency, and cultural appropriateness. This paper proposes the Multi-Level Responsible AI Adoption Framework (ML-RAAF), a conceptual model linking three levels of governance: macro (global and national policy), meso (institutional governance and readiness), and micro (educator and student adoption). Drawing on published policies and academic literature, including UNESCO and OECD guidelines and empirical studies, the framework brings together trust, transparency, ethical safeguards, and cultural adaptability. It offers higher education stakeholders a way to connect policy directives with institutional capacity and classroom practice, supporting responsible AI adoption that is globally informed and locally practical.