“I am embarrassed. This should not have happened” - Solly Malatsi on AI policy scandal

Key topics:Minister apologises for fake sources in Draft AI PolicyDraft AI Policy withdrawn; accountability and review promisedScandal highlights AI hallucinations and need for human oversight.Sign up for your early morning brew of the BizNews Insider to keep you up to speed with the content that matters. The newsletter will land in your inbox every morning on weekdays. Register here.Support South Africa's bastion of independent journalism, offering balanced insights on investments, business, and the political economy, by joining BizNews Premium. Register here.If you prefer WhatsApp for updates, sign up to the BizNews channel here..By Myles Illidge.Communications minister Solly Malatsi has publicly apologised for his department’s publication of the Draft National AI Policy, which included fabricated sources in its reference list.In an interview with SABC News, the minister described the situation as a massive embarrassment not only to his department but also to the national government.“This is an apology to South Africa. It should not have happened. This is a massive embarrassment to the department and the government,” he said.“It just exposes the irony inasmuch as we were in the process of developing a policy that will also have a package of guidelines around the responsible use of AI,” Malatsi said..Read more:.Stafford Masie: Open letter to Malatsi on GNU's draft AI policy.“We fell foul of that. We fell short of that, and you know, as government, particularly as the executive authority, we have to take responsibility for the lack of robust oversight.”He added that what was meant to be a breakthrough moment for South Africa, in terms of policy, is now tainted by the massive failure.“We’ll make sure that there’s accountability internally in the department, and we will make sure that it gets subjected to much more rigorous oversight,” Malatsi said.He said his department would aim to restore the policy’s integrity before it is re-published for public comment.The minister said that the error showcased the need for much more vigilant human intervention in the AI space.“As a tool, it’s very useful. It can’t be a substitute for human responsibility,” Malatsi said.The department’s use of AI-hallucinated research papers and journals in the Draft National AI Policy was exposed in a News24 report, which found that several of the sources cited didn’t exist.Malatsi said that he “interfaced” with the policy and missed the fabrications, adding that he had been humbled as the minister responsible for the portfolio.“I interfaced with the policy a lot of times in terms of reviewing it and subjecting it to scrutiny internally. We missed this,” he said.Malatsi published a statement on Sunday, 26 April 2026, withdrawing the Draft National AI Policy, just over two weeks after it was published for public comment.An unacceptable lapse.In a statement announcing the withdrawal, Malatsi said the error was more than a technical issue; it was a failure that compromised the policy’s integrity and credibility.“I am withdrawing the Draft National Artificial Intelligence Policy. South Africa deserves better,” the minister said.“The Department of Communications and Digital Technologies did not deliver on the standard that is acceptable for an institution entrusted with the role to lead South Africa’s digital policy environment.”He acknowledged that the most likely explanation was that AI-generated references were included in the document without proper verification.“This unacceptable lapse proves why vigilant human oversight over the use of artificial intelligence is critical,” Malatsi said.“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.”The minister published the Draft South African National Artificial Intelligence Policy for public comment in the Government Gazette on 10 April 2026.“It was approved by Cabinet on 25 March 2026 (combined with the Special Sitting of Cabinet on 1 April 2026),” the notice stated.Therefore, President Cyril Ramaphosa and other members of his Cabinet gave the draft policy the thumbs up before its publication.Fake sources in the Draft South Africa National AI Policy.On Friday, 24 April 2026, Article One wrote to Malatsi about the irregularities, and the following day, News24 published an article revealing the inclusion of fabricated research.Activist Zackie Achmat published part of Article One’s letter to Malatsi, which raised concerns about the use of fake sources.“Initial study of the policy has revealed that a number of sources cited in the policy do not in fact exist. I refer to the following six sources,” it said. The fictitious sources are as follows:Source 5 — Babatunde, O., & Mnguni, P. (2023). “Challenges and Opportunities in Regulating AI: Perspectives from South Africa.” AI Policy Journal, 2(3), 143-156.Source 9 — Burman, A., & Sewpersadh, K. (2022). “Legal Frameworks for AI in South Africa: Balancing Innovation and Accountability.” South African Journal of Philosophy, 41(2), 207-217.Source 10 — Cavaliere, F., McGregor, R., & Hersh, M. (2022). “Artificial Intelligence and Ethics in Emerging Economies: The Case of South Africa.” AI & Society, 37(4), 565-583.Source 19 — Etale, A. E., & Naidoo, P. (2021). “Accountability and Transparency in African AI Policies: Lessons from South Africa.” African Journal of Public Affairs, 9(1), 39-52.Source 22 — Fourie, I., & Botha, A. (2021). “AI Ethics in South Africa: Implications for Trust and Fairness in AI.” African Journal of Science, Technology, Innovation and Development, 13(5), 677-688.Source 56 — Smith, M., & Mahomed, R. (2021). “The Impact of AI on Social Justice in South Africa.” Journal of Ethics and Social Philosophy, 18(3), 313-329.Article One said it was highly concerning to find fake sources cited in any National Policy in South Africa, adding that their inclusion suggests the use of a large language model (LLM) in drafting the policy..Read more:.AI or human? Why proof of personhood may soon be essential online.“LLMs are known to hallucinate information when generating text,” it said.“The inclusion of these potentially hallucinated sources in the policy further underscores the need for human oversight of AI technologies.”.This article was first published by MyBroadband and is republished with permission.

“I am embarrassed. This should not have happened” - Solly Malatsi on AI policy scandal

Key topics:

Related Stories