Full Breakdown
South Africa Withdraws Draft AI Policy After AI-Generated Fabricated Citations Discovered
5/1/2026, 8:52:57 PM
Draft AI Policy and Its Intended Framework
The Department of Communications and Digital Technologies (DCDT) published a Draft National Artificial Intelligence Policy in early April 2026, proposing a National AI Commission, an AI Ethics Board, an AI Regulatory Authority and tax incentives for private-sector AI projects. The draft also outlined tax breaks, grants and subsidies to stimulate private-sector AI infrastructure development. A review found at least six of the 67 citations were AI-generated hallucinations—references that appeared legitimate but linked to non-existent papers—undermining the draft’s scholarly foundation.
Timeline of the Withdrawal
10 April 2026: Draft released for comment (deadline 10 June).
24 April: News24 flagged six fake references.
27 April: Minister Solly Malatsi withdrew the draft on X and LinkedIn.
Journal editors confirmed the citations were fictitious.
Data on Fabricated Citations
The draft contained six fabricated citations (? 9 % of 67). One source said 12 % of references in a 150-page version were false. The document spanned 150 pages, according to one source. A *Nature* study reported 2.5 % of 2025 papers had hallucinated citations, over 110 000 papers. Over 400 experts demanded human-in-the-loop safeguards. The DCDT spent about R14 million on the draft.
Government Response
Minister Malatsi said the lapse was not a mere technical issue and that it compromised the integrity and credibility of the draft policy. He announced that those responsible would face consequence management and that the department would treat the matter with gravity. The DCDT confirmed internal verification of the fabricated references.
Civil-Society Criticism
Civil-society groups and analysts condemned the unverified use of generative AI. An open letter signed by over 400 experts called for mandatory human-in-the-loop safeguards. Some civil-society groups called for the resignation of the department’s director-general. Commentators warned a 14-month delay in AI regulation and loss of digital-sovereignty confidence.
Conflicting Reports & Gaps
One source cites six of 67 citations (? 9 %); another reports 12 % of references in a 150-page draft are false. The exact drafting stage where AI was used is not disclosed.
Implications for AI Governance
The incident shows a policy meant to regulate AI can be compromised by AI-generated misinformation, raising concerns about document reliability, oversight and leadership ambitions and could affect foreign investment.
Prospects for Revision
The DCDT will audit all references, introduce verification protocols and revise the policy, but no publication date is set. Stakeholders expect the update to address technical and governance gaps.
Verbatim Quotes
- “The most plausible explanation is that AI-generated citations were included without proper verification. This should not have happened,” — Solly Malatsi, Minister of Communications and Digital Technologies
- “It’s a major embarrassment.” — Solly Malatsi, LinkedIn statement
- “South Africans deserve better.” — Solly Malatsi, LinkedIn statement
- “In fact, this unacceptable lapse proves why vigilant human oversight over the use of artificial intelligence is critical. It’s a lesson we take with humility. I want to reassure the country that we are treating this matter with the gravity it deserves. There will be consequence management for those responsible for drafting and quality assurance.” — Solly Malatsi, X post
