Open Weights, Hidden Dependencies: Governing Open-Weight Al in Africa

Digital Policy Hub Working Paper

July 21, 2026

Open-weight artificial intelligence foundation models are systems whose trained parameters are publicly released for download and local deployment. They are increasingly capable and widely available, and represent a lower-cost pathway to technological sovereignty for governments globally. As open-weight models become more accessible, the barrier to adoption is no longer technical readiness. The absence of operational governance frameworks — which specify accountability structures, pre-deployment requirements and institutional responsibility for system behaviour — remains the primary obstacle to defensible adoption. African public institutions face this challenge under conditions of uneven institutional capacity and significant infrastructure gaps. Tiered policy recommendations addressing national procurement, update governance and regional coordination can enable safe open-weight adoption without waiting for comprehensive governance infrastructure that may remain out of reach.

About the Author

Amal Hussein is a Digital Policy Hub master’s fellow and an M.Sc. candidate in systems science and engineering with a concentration in interdisciplinary artificial intelligence at the University of Ottawa.