South Africa had a plan. A big one. Then AI sabotaged it.
Earlier this year, the Department of Communications and Digital Technologies unveiled an ambitious 86-page Draft National AI Policy, a document designed to cement South Africa’s place as the continent’s AI governance leader. The policy was spearheaded by the Department of Communications and Digital Technologies and was expected to be fully implemented in the 2027/2028 financial year. The vision was sweeping: a National AI Commission, an Ethics Board, a Regulatory Authority, an Ombudsperson, and even an AI Insurance Superfund. It didn’t survive a bibliography check.
News24 discovered that at least 6 of the policy’s 67 academic citations were AI-generated hallucinations, fake articles placed inside real journals, credited to authors who had never written on the topics attributed to them. Communications Minister Solly Malatsi pulled the document immediately, calling it an “unacceptable lapse” and promising accountability for everyone involved in drafting and sign-off. The irony cuts deep. A policy built to regulate AI was ultimately undermined by AI itself.
READ ALSO: Yango’s $150M Africa Expansion Reshapes Ride-Hailing
What made this scandal worse was the timing. The fabricated citations appeared in a government document about artificial intelligence, written by the department responsible for the country’s digital technology strategy, during the exact period when the world’s most consequential AI governance debates are being fought in Brussels, Washington, and Beijing. South Africa didn’t just embarrass itself ,it stumbled onto the global stage at the worst possible moment.
What’s particularly striking is the policy’s original design philosophy. Rather than building an entirely new bureaucratic structure, South Africa’s framework was built around institutions already embedded within existing sectors. The Financial Sector Conduct Authority and the South African Reserve Bank would oversee financial AI systems, SAHPRA would handle AI in medical diagnostics, and the Information Regulator would retain its role as the primary enforcer of data privacy. The logic was sound: regulators closest to the problem are best placed to manage it. The vision wasn’t wrong. The process was.
So what happens now? South Africa has appointed an independent seven-member panel chaired by Wits University AI researcher Benjamin Rosman named to Time’s list of the 100 most influential thinkers in AI in 2025 to rebuild the draft. The revised policy is expected to go to Cabinet by November 2026, with a target publication date for public comment set for January 2027. The delay is costly, but the reset may ultimately produce something more credible and durable.
The hallucinated citations reveales two critical failures ,epistemic integrity and information integrity ,and the policy was not equipped to govern either, having itself demonstrated both. For every government tempted to let AI do the heavy lifting on complex documents, South Africa’s experience is a stark reminder. AI can help draft the future.
It cannot be trusted to fact-check it. And in the gap between those two realities, entire governance frameworks can collapse.
