Silver lining in South Africa’s national AI policy failure
While the inclusion of AI-hallucinated research in the draft South Africa National AI Policy caused embarrassment for the communications department and minister Solly Malatsi, there is a silver lining.
According to Adams & Adams partner Darren Olivier, Malatsi’s rapid withdrawal of the policy was the right move, as it demonstrated awareness of the seriousness of the issue and prevented further damage.
The Department of Communications and Digital Technologies published the policy for public comment in early April 2026.
However, the document was found to include six citations likely fabricated by large language models (LLMs), prompting harsh backlash against the department and Malatsi.
Malatsi withdrew the policy in response to the findings and the backlash, apologised publicly, and has committed to restoring the policy’s integrity before it is republished.
“Credit where credit is due,” Olivier said. “The decision to withdraw the draft policy quickly was the right one.”
“It showed awareness of the problem and a willingness to act before the issue grew even larger. In a governance context, speed matters.”
He said the longer a flawed document remained in circulation, the harder it became to rebuild confidence and trust.
Olivier said most organisations would hesitate in similar situations, hoping that the issue would fade or go unnoticed.
“Experience suggests that this strategy rarely works. Withdrawal, by contrast, signals respect for the process and for the public that relies on it,” he said.
“While the situation is clearly embarrassing, the response deserves recognition. It protected the integrity of the policy process and created space to fix the problem properly.”
Olivier said the issue provided a valuable lesson for every organisation in South Africa, as businesses were now experimenting with AI.
“Every professional is under pressure to work faster and deliver more,” he said. “Every team is learning, often in real time, how to integrate new tools into existing workflows.”
“Reputation is fragile and takes years to build. AI can damage it quickly, and once trust is shaken, recovery can take considerable time. It may never fully happen.”
Next draft matters more than the first
According to Olivier, the incident afforded South Africa the opportunity to rebuild the policy on stronger foundations.
“If the next draft reflects careful oversight, disciplined verification, and a clear understanding of local realities, this uncomfortable moment will have served a constructive purpose,” he said.
“The demand for clarity on AI governance is only going to grow. Businesses want guidance. Regulators want consistency, and citizens want reassurance that new technologies are managed responsibly.”
Olivier said the next version of the policy did not need to be perfect, but it must be credible. That will require three crucial elements.
Firstly, the policy should be rooted in local realities. “Borrowing ideas from other jurisdictions is sensible, but copying them wholesale rarely works,” he said.
“Our legal framework, infrastructure constraints, and economic priorities are different from those of Europe, the United States, or Asia.”
Secondly, Olivier said a credible process would show clear human oversight, adding that people would trust systems more easily if they knew who stood behind them.
“Thirdly, and perhaps most importantly, the policy should demonstrate verification. Sources should be checked carefully. Evidence should be traceable,” he said.
“Assumptions should be tested against real-world conditions. These steps may feel routine, yet they are the foundation on which public trust is built.”
AI can destroy reputations
Olivier said that South Africa has reached an important moment, as the body designed to create trust around AI became a case study in governance.
According to Olivier, governance was not about banning the technology. At the very least, it’s about knowing when the AI was used, who reviewed the output, and how the final decision was reached.
He explained that AI use was sensible for many organisations, including the government, and could provide various benefits.
“AI is very good at producing structure, summarising large bodies of information, and as a sounding board,” he said.
“In practice, the best results usually come from collaboration. AI accelerates the work. Humans apply judgment.”
Olivier said his working rule was to never outsource human thinking. Humans must provide judgment as the work takes shape.
“Then, before anything leaves the building, let humans verify every reference, every citation, and every assumption,” he said.
He stated that this final check is where credibility was protected and, coincidentally, where many organisations would underestimate risks.
Loading ...