South Africa is adopting a decentralized and flexible approach to artificial intelligence governance, prioritizing coordination over a single centralized regulator. The draft AI policy, published for public comment in April 2026, is designed to distribute oversight across existing institutions such as financial, health, and data protection regulators instead of creating a new super-agency. The framework is built on a risk-based system that categorizes AI applications into four levels—unacceptable, high, limited, and minimal risk—where stricter controls apply only to systems with greater potential harm. High-risk uses like hiring, lending, and healthcare will require audits, impact assessments, and human oversight, while lower-risk tools will face lighter regulation. A National AI Coordination Office and advisory council will help guide implementation, but enforcement will remain with sector regulators, raising concerns about uneven capacity and inconsistent oversight. The policy also emphasizes building local datasets and inclusive AI systems tailored to African contexts.
Read more here
https://techcabal.com/2026/04/09/south-africas-new-policy-is-betting-on-decentralised-ai-control
