South Africa has withdrawn its first national artificial intelligence policy after the draft document was found to contain fictitious citations, likely generated by AI tools.
The policy was shelved in April 2026 following the discovery of the fabricated references. Minister Solly Malatsi publicly raised concerns about the authenticity of the draft, emphasizing the critical importance of human oversight in AI deployment. Those responsible for the discrepancies are now facing potential sanctions.
The incident highlights a growing risk in government and institutional use of AI writing tools — that AI-generated content can produce plausible-sounding but entirely fabricated sources, a phenomenon sometimes referred to as “hallucination.” The inclusion of such citations in an official national policy document suggests the draft may not have undergone sufficient human review before circulation.
Minister Malatsi’s response underscores a broader concern: that AI tools, without adequate human supervision, can introduce errors that undermine the credibility and integrity of official documents. The withdrawal of the policy suggests South African authorities are treating the matter seriously, with accountability measures underway against those involved in producing the flawed draft.
The episode may prompt other governments developing AI policy frameworks to strengthen review processes and verification standards when AI-assisted drafting tools are used in official capacities.
Source: Tech-Economic Times