Every week, another African government signs an MOU to adopt this AI framework or that responsible AI charter — documents authored in Brussels, Washington, or Silicon Valley boardrooms far removed from the lived realities of Lagos, Nairobi, or Kinshasa.
I do not dispute the urgency of AI governance. I dispute who should be doing the governing, and on whose terms.
The Colonial Epistemology Problem
When the EU AI Act classifies systems by risk tiers that were designed with European social contexts in mind, it implicitly embeds assumptions about state power, individual rights, and the role of technology that may not translate — and in some cases actively conflict — with how African societies organise themselves.
Consider biometric identification systems. In a European rights framework, mass biometric deployment by the state is high-risk, approaching prohibited. Yet in Nigeria, Ghana, and Kenya, national ID systems anchored to biometrics have been transformative for financial inclusion, reducing the unbanked population by tens of millions. Applying the EU framework verbatim would chill exactly the applications generating the most social benefit on this continent.