NextFin News - South Africa’s ambitions to lead the continent’s digital revolution suffered a humiliating setback this week as the government was forced to withdraw its flagship National Artificial Intelligence Policy after it was discovered to be riddled with AI-generated "hallucinations." The scandal has left two senior cabinet members, Communications Minister Solly Malatsi and Higher Education Minister Nobuhle Nkabane, facing intense scrutiny over the institutional integrity of their departments.
The crisis erupted when investigative reports by News24 and Daily Maverick revealed that at least six of the 67 academic citations in the draft policy—published for public comment on April 10, 2026—were entirely fictitious. These "phantom" sources included fake articles attributed to real journals, a classic symptom of Large Language Models (LLMs) being used to draft technical documents without human verification. Minister Malatsi officially withdrew the document on April 26, admitting that the lapse had "compromised the integrity and credibility" of the state’s regulatory efforts.
The irony of a policy designed to regulate AI being undermined by the very technology it sought to govern has not been lost on the South African tech sector. Beyond the embarrassment, the incident exposes a profound capacity gap within the Department of Communications and Digital Technologies (DCDT). If the department responsible for setting the guardrails for AI cannot distinguish between factual research and algorithmic fiction in its own white papers, its ability to oversee complex private-sector deployments is now under a cloud of doubt.
The fallout extends to the Ministry of Higher Education, as the draft policy relied heavily on academic frameworks that were supposed to be vetted by state-funded research bodies. Critics argue that the reliance on generative AI for policy drafting suggests a "shortcut culture" within the civil service. While the use of AI in administrative tasks is increasingly common globally, the failure to perform basic fact-checking on a document that passed through Cabinet approval indicates a systemic breakdown in oversight.
From a market perspective, the withdrawal creates a regulatory vacuum. South Africa had positioned itself to be Africa’s primary hub for AI innovation, aiming to attract venture capital and multinational tech investment by providing a clear legal framework. Instead, the delay leaves businesses in a state of uncertainty. Legal experts at Fasken noted that this "hallucinatory irony" might deter international firms who require stable and intellectually rigorous regulatory environments before committing long-term capital to the region.
However, some industry observers offer a more cautious, if not slightly optimistic, counter-perspective. They suggest that this public failure could serve as a necessary "stress test" for South African institutions. By withdrawing the document immediately rather than defending the errors, Malatsi may have preserved some long-term credibility, signaling that the administration is at least aware of the risks posed by unverified AI outputs. This incident may force a more rigorous, human-centric approach to the next iteration of the policy, potentially resulting in a more robust framework than the one originally proposed.
The immediate consequence remains a stalled digital agenda. As other emerging markets like Kenya and Nigeria accelerate their own AI governance strategies, South Africa now finds itself back at the starting line, tasked with rebuilding trust in its technical expertise. The DCDT has initiated internal inquiries to determine how the hallucinated text bypassed multiple levels of senior review, but the damage to the nation's "digital leader" brand will take significantly longer to repair.
Explore more exclusive insights at nextfin.ai.
