<div> Realising the Promise of Open-source AI for Inclusive and Equitable AI Outcomes in Africa.  </div>

Open-source artificial intelligence has emerged as a pathway toward more inclusive and equitable AI development, particularly for African contexts that have historically been excluded from the concentrated centres of global AI research and governance. Yet the relationship between openness and equity is neither automatic nor straightforward. This paper examines two interconnected questions: what is the role and promise of open-source AI in achieving inclusive and equitable AI, particularly for African contexts, and how can that promise be fulfilled while managing risks, with specific attention to the roles national and regional governments can play?

Drawing on a systematic desk-based review of academic literature, policy frameworks, technical reports, and African AI initiative documentation from 2019 to 2025, this paper applies TGov's five-pillar equity framework, spanning quality datasets, fair access and benefits, transparency, participatory development, and bias auditing, across the full open-source AI configuration spectrum. Three original contributions are made. First, a granular mapping of which openness configurations advance which equity dimensions, moving beyond the binary open versus proprietary framing that dominates existing literature. Second, a three-dimensional analytical framework that distinguishes access and capacity constraints, representation and performance gaps, and governance power asymmetries, preventing the systematic overestimation of what licensing changes alone can accomplish. Third, governance recommendations are calibrated specifically to African institutional realities rather than transplanted from Global North precedents.

Key findings indicate that open-source AI provides necessary but not sufficient conditions for inclusive and equitable outcomes. Benefits materialise only when openness is paired with complementary investments in infrastructure, skills development, sustainable institutional support, and community governance. Risks, including misuse, bias perpetuation, and dependency reproduction, are distributed unevenly, falling disproportionately on African populations with the least capacity to mitigate them. Government roles in infrastructure provision, risk-based regulation, capacity-building, and regional coordination are essential and currently underdeveloped across the continent. The paper concludes with actionable recommendations for African governments, regional bodies, development partners, and open-source communities.