“A massive embarrassment” - Minister withdraws South Africa’s Draft National AI Policy due to fictitious sources

The inclusion of various fictitious sources in the reference list of South Africa’s pioneering Draft National AI Policy “compromised the integrity and credibility” of the policy and was a failure that went beyond mere technical issues, said Communications and Digital Technologies Minister Solly Malatsi as he yesterday officially withdrew the Policy, which had been published on 10 April for public feedback. The submissions period was originally to run until 10 June 2026. .

“South Africans deserve better,” said Minister Malatsi in an official statement yesterday. “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 … this should not have happened.”

The most plausible explanation, said Malatsi, was that AI-generated citations were included without proper verification, which “should not have happened”. The unacceptable lapse - which the Minister this morning on SABC News called “a massive embarrassment” for his own department and the South African government - also underlined why “vigilant human oversight” was critical in the use of AI. 

What was meant to be a breakthrough moment for South Africa, in terms of national policy, had now been tainted by the massive failure, admitted Minister Malatsi. 

The events of this weekend, culminating in the embarrassing withdrawal of a pioneering Policy, is a reminder of why a belief in AI has to come with responsibility, says Naomi Thompson, founder of the Legal Innovation Summit and a leading voice and strategist working at the intersection of business, law, data, and technology.

“Minister Solly Malatsi’s decision to withdraw it was the right call; that's accountability in action,” says Thompson, who is a self-confessed ‘big believer’ in AI and the value it adds in business and law. “The cost of skipping proper verification isn’t just an embarrassing mistake. It causes a loss of credibility and trust. I’ll keep saying this: technology is only as good as the people using it responsibly.”

Writing on LinkedIn today, Lucien Pierce, Head of Telecommunications, Media, and Technology at Phukubje Pierce Masithela Attorneys, said that while the Policy’s hallucinated references were disappointing, South Africa should resist ‘throwing the baby out with the bathwater’ as overall he considered the Policy well-thought through, with a practical approach tailored to South Africa’s own circumstances, rather than “a slavish copy of another country’s AI policy or framework.” 

Pierce was concerned by the likely delays created by yesterday’s withdrawal of the Draft National AI Policy, anticipating it may be 6-12 months before the public sees the next, fine-tuned version, and even longer before legislation or regulation occurs. 

“If South Africa waits until the next version is published, we are likely to miss the AI bus,” writes Pierce. “While the Draft AI Policy is being fine-tuned, South Africa urgently needs to see sector specific regulators stepping in… Financial services, communications, data protection and competition regulators, working with the private sector, must step in and issue directives and guidelines.”

The Policy’s withdrawal followed a News24 report on Saturday, which found several sources cited in the Draft AI Policy did not exist. The Policy was intended to signal a bold step toward fostering a sustainable, inclusive AI ecosystem. By aligning AI advancements with the Constitution and socio-economic goals, it aimed to create a resilient digital economy prioritising equity, innovation, and national competitiveness.

“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,” said Malatsi on SABC News earlier today. “We fell short of that, and as the government, particularly as the executive authority, we have to take responsibility for the lack of robust oversight. We’ll make sure that there’s accountability internally in the department, and that it gets subjected to much more rigorous oversight.”