South Africa’s Department of Communications and Digital Technologies was forced to withdraw its own draft National Artificial Intelligence Policy just weeks after publishing it, following the discovery that the document contained fake, AI-hallucinated citations in its reference list.
The policy, published for public comment on April 10, 2026, was meant to lay the groundwork for how South Africa would responsibly regulate AI adoption across both public and private sectors — with an explicit goal of eventually evolving into binding national law.
Minister Solly Malatsi withdrew the policy on April 26, just over two weeks after its release, after the fabricated references came to light — a particularly embarrassing irony given the document was meant to demonstrate responsible AI governance. The incident became a notable case study in the risks governments face when using AI tools to help draft policy without adequate human verification, especially for documents meant to set the very standards for trustworthy AI use.
The episode adds a cautionary note to South Africa’s broader AI ambitions, even as the country continues positioning itself as a regional leader in AI policy and infrastructure development across the continent.
