South Africa’s government spent April producing AI policy documents that were, themselves, products of unverified AI. The irony is almost too tidy.
Here is what happened. Communications Minister Solly Malatsi, a DA man, published an 86-page draft National AI Policy earlier this month. It was Cabinet-approved, ambitious, and full of academic citations. News24 went through the bibliography. At least 6 of the 67 cited academic papers did not exist. The journals were real. The articles were fabricated. Malatsi pulled the policy on April 27, calling it an “unacceptable lapse” and promising “consequence management” for those responsible.
Before the dust had settled on that one, Home Affairs Minister Leon Schreiber, also DA, discovered the same problem in his department’s newly revised White Paper on Citizenship, Immigration and Refugee Protection. Except worse. News24 found that 102 out of 148 references in the document were not verifiable or traceable. An initial internal review suggests the fake citations were not woven into the main text but appended to the reference list after the document was drafted, which raises a different and more troubling set of questions about exactly who did what and when.
Schreiber suspended two senior officials: a Chief Director in the unit responsible for the White Paper, and a Director involved in the drafting process. He also appointed two independent law firms to manage the disciplinary process and, in a move that suggests someone has developed a sudden enthusiasm for thoroughness, ordered a review of every policy document produced by the department since November 30, 2022. That date is not random. It is the day OpenAI released ChatGPT.
The DA’s response to both incidents has been to mandate AI verification across all its ministerial departments, with Schreiber pledging to raise the issue at the next Cabinet meeting. The parliamentary portfolio committee chair offered a more pointed suggestion to Malatsi’s team: “skip using ChatGPT this time” when redrafting.
What makes this particularly uncomfortable for the DA is that the party has been the most vocal proponent of using technology to fix South Africa’s creaking public service. Schreiber built his Home Affairs reputation on digitisation and AI as corruption-busters. Malatsi’s entire portfolio is digital transformation. The argument was always that tech-forward governance is better governance. The counterargument is now exhibit A in two separate Cabinet-approved documents.
The technology itself is not the villain here. AI tools producing plausible-sounding fake citations is a known and documented problem, the kind of thing covered in roughly every technology publication since 2023. The officials involved either did not know this, or knew and did not check. Both explanations are bad.
South Africa is not alone. Governments from Brazil to Australia have had similar incidents. But the timing here is spectacular: the country’s draft framework for governing AI had to be scrapped because it used AI irresponsibly. You genuinely could not script it.
The ANC, sensing opportunity, has called for Malatsi to appear before parliament. Malatsi has said he will cooperate.
Meanwhile, somewhere in a government office, someone is presumably redrafting the AI policy. Hopefully by hand.
[Source: News24; MyBroadband; TechCentral; The Register; Daily Maverick; GroundUp]
