DOI: 10.12688/f1000research.186904.1 ISSN: 2046-1402

A Risk-Based Artificial Intelligence Governance Framework for Higher Education Institutions in Low-Resource Settings: Evidence from Uganda

Businge Phelix Mbabazi, Vicent Mabirizi, Robert Tumusiime, Jones Muragira
Show PDF Cite
The rapid uptake of Artificial Intelligence (AI) in East African universities promises major advances in teaching, learning, research and administration, but introduces serious ethical, equity and regulatory risks in resource-limited settings. This paper showcases Kabale University’s Artificial Intelligence Policy and Academic & Administrative Guidelines, developed and being piloted as of 2025, as a replicable governance model for low- and middle-income higher education institutions. Co-designed through participatory workshops involving students, faculty, administrators, local community leaders and the Directorate of ICT Services, the policy aligns with Uganda’s Data Protection and Privacy Act 2019 (2025 enforcement guidelines), UNESCO’s Recommendation on the Ethics of AI (2021, updated 2025 toolkit), OECD AI Principles (2024), the EU Artificial Intelligence Act (Regulation (EU) 2024/1689), and the African Union Continental AI Strategy (2024–2030). The framework introduces a four-tier risk classification (Prohibited, High-Risk, Limited-Risk, Minimal-Risk), establishes a multidisciplinary AI Ethics Committee with student representation, mandates pre-deployment risk assessments for high-risk applications (e.g., automated grading, admissions shortlisting), and enforces human-in-the-loop oversight for critical decisions. In academic contexts, it actively encourages responsible AI use in presentations, sit-down tests, projects, conference papers and publications to strengthen skills development, while strictly prohibiting misuse that undermines academic integrity and capping AI-generated content at 25% of any submission unless expressly authorised and disclosed. Best-practice adoption at Kabale University has already yielded stronger ethical compliance, reduced bias in deployed tools, and innovative educational applications notably AI-supported peer review workflows and personalized tutoring without disadvantaging rural, low-resource students. The paper details the inclusive design process, emerging operational lessons, and concrete recommendations for scaling responsible AI governance across the East African Community, ensuring innovation coexists with academic integrity, data privacy and inclusivity in constrained environments.

More from our Archive