Withdraw AI Policy, Malatsi Told, as Fake Citations Row Grows

Why It Matters

The episode underscores the dangers of relying on generative AI for policy drafting, eroding public trust and potentially delaying critical AI governance reforms in a fast‑growing sector.

Key Takeaways

• Draft AI policy contained six fabricated academic citations
• Minister Solly Malatsi withdrew the 86‑page draft after scandal
• Opposition demanded a full review without AI tools
• Policy proposed seven new AI regulatory bodies
• Incident highlights South Africa’s tech‑literacy gap in governance

Pulse Analysis

The withdrawal of South Africa’s draft AI policy illustrates a cautionary tale for governments worldwide that are racing to codify artificial‑intelligence rules. While the draft aimed to position the country at the forefront of AI governance with a National AI Commission, Ethics Board, Insurance Superfund and Safety Institute, the inclusion of fabricated references revealed a lack of rigorous editorial oversight. In an era where AI tools can generate plausible but false citations, ministries must embed verification steps and human expertise to safeguard credibility.

Political fallout has been swift. Members of the opposition, including ANC MP Khusela Diko, called for the policy’s complete retraction and a review free from AI assistance. Even within the governing Democratic Alliance, senior figures like Dean Macpherson defended the minister, while former spokesperson Phumzile van Damme warned that the scandal could become a global exemplar of governmental tech illiteracy. The incident dovetails with broader concerns about AI‑driven misinformation, echoing recent controversies such as Deloitte’s AI‑assisted report error in Australia.

Beyond the immediate embarrassment, the episode may delay South Africa’s strategic AI agenda. Investors like Stafford Masie have warned that over‑regulation without proper infrastructure funding could stifle the nation’s participation in the global AI economy. A revised policy, vetted without reliance on generative AI, will need to balance robust oversight with flexibility to attract compute investment. The episode serves as a reminder that effective AI governance requires both technical competence and transparent, accountable policymaking.