Towards an African Union Treaty on Artificial Intelligence: Reconciling African Values with Global Standards

In the wake of the European Union (EU)'s landmark Artificial Intelligence Act and similar normative developments in other regions, the regulation of artificial intelligence (AI) has become a global race for governance legitimacy and leadership. While many jurisdictions are swiftly advancing binding legislative instruments to manage the opportunities and risks of AI, AI governance across African states and regional institutions has, to date, tended to favour a more deliberative and consultative approach, reflecting diverse national priorities and capacities. The African Union (AU), through various soft law initiatives and continental strategies, has established foundational principles that prioritise inclusiveness, context sensitivity, and sovereignty in technological governance. Chief among these initiatives is the AU Continental Artificial Intelligence Strategy, which articulates a vision for AI governance that reflects African values – a collection of indigenous values, a communal ethos, and socio-cultural realities – while cautiously engaging with the technical universality promoted by global standards. The purpose of this paper is to outline an agenda for discussions on the imperative and feasibility of transitioning from soft law frameworks to an AU treaty on AI. It argues that the central challenge is not merely the codification of rules, but the normative reconciliation of African values with existing global standards. Global AI standards – often shaped by Eurocentric epistemologies and market logic – may inadequately reflect African ethical frameworks, historical contexts and development priorities. Yet the pursuit of regulatory autonomy must not lead to technological isolationism. Instead, the article argues for a hybrid normative approach that retains the technical interoperability necessary for international collaboration, while embedding the African philosophies of Ubuntu, solidarity and fairness within the legal architecture of AI governance.