AI policy collapse forces reset: Malatsi calls in expert panel [VIDEO]

The Department of Communications and Digital Technologies (DCDT) has delivered a blunt account of why South Africa’s Draft National Artificial Intelligence (AI) Policy was withdrawn and how credibility will be restored.

The Portfolio Committee on Communications and Digital Technologies received a briefing from the department on Tuesday regarding the circumstances surrounding the withdrawal of the AI National Policy document, which had been published for public comment.

Policy withdrawal

Communications Minister Solly Malatsi pulled the policy after AI hallucinations fabricated fictitious references, including academic citations, journals and authors that simply don’t exist.

Malatsi declared that only a full reset could protect the integrity of such a critical policy area.

The draft policy released for public comment on 10 April 2026 was ambitious, proposing new oversight bodies, including a National AI Commission and a Road Accident Fund‑style AI Insurance Superfund to compensate citizens harmed by AI.

The fallout has been serious. Two officials have been placed on precautionary suspension, with investigations nearing completion.

Fixing errors

Addressing the committee, Malatsi said a panel would be established to help with the policy.

“Beyond that, as we stated in the budget speech, it is critical to begin fixing the fundamental errors that arose from the overlap in oversight and the irresponsible use of generative AI. To do so, we sought the expertise of leading specialists in artificial intelligence, governance, and academia, and announced the establishment of the AI Advisory Panel.

“This panel will guide the review of the withdrawn document and authenticate all references to ensure credibility,” Malatsi said.

The National AI Expert Review Panel will be chaired by Professor Benjamin Rosman of Wits University, the panel includes leading figures in AI research, law, governance, and cybersecurity:

• Professor Vukosi Marivate,
• Professor Alison Gillwald,
• Ms Heather Irvine,
• Dr Tshepo Feela,
• Dr Jabu Mtsweni, and
• Advocate Lufuno Tshikalange.

“Their mandate is strictly advisory but far‑reaching: to assess the withdrawn draft, identify weaknesses, recommend credible sources to replace fictitious citations, and ensure alignment with constitutional principles, international best practice, and South Africa’s strategic priorities,” the department said.

Costs

Asked by MPs who would bear the cost of the panel, Malati said they would provide their services free of charge.

“The services of the members of the panel would be pro bono. The only cost associated, which the department would bear, would be on logistical support, which would include, if needs be, things like for transportation purposes, if they are in person, in-person meetings, or of that nature, and arrangements, if needs be a venue. So, they won’t be any remuneration directly going to these individuals.

“We are grateful that you know, given their expertise and also their roles, that they’ve agreed to play a part in this process at no cost to the department when it comes to remuneration,” Malatsi said.

Reviews

The department said the panel will issue thematic reviews, consolidate recommendations, and provide final advisory opinions before Cabinet sign‑off.

It will lead to stakeholder engagement, with broad public consultation planned once Cabinet clears the revised draft.

“Timelines are tight: establishment in May 2026, orientation in June, advisory review in July, consolidation in August, drafting through September, final panel review in October, and Cabinet approval by November-December. The target date for public comment is January 2027.

Interventions

Malatsi stressed that this intervention is important for three reasons.

“First, until the exposure of fictitious references, much of the substance of the draft policy had not been widely contested. It was intended as a foundation for external contributions to enrich the final policy.

“Second, an internal drafting team will continue to support the advisory panel, ensuring Parliament retains a prominent role in finalising the concept. Speed is essential to avoid a prolonged policy vacuum.

“Third, the terms of reference for the panel’s work have been finalised. At its first meeting in the coming weeks, the panel will engage with these terms, refine them, and roll out a project plan for the policy’s completion,” Malatsi said.

The Minister of communications and digital technologies, Mr Solly Malatsi, leads department’s presentation on the withdrawal of national AI policy. #CommitteesAtWork >JM pic.twitter.com/E6NeoU2LQc

– Economic-Cluster (@EconoCluster) May 26, 2026

Finalisation

Malatsi said the AI policy is expected to be finalised by the end of this financial year.

“Our aim is clear: to finalise and deliver a credible AI policy outcome by the end of this financial year. This process will necessarily involve Cabinet engagement and adherence to the prescribed timelines for publishing draft policies for public comment.

“Ultimately, the revised policy must be free of the avoidable faults that undermined the withdrawn draft, and it must stand as a credible, evidence‑based framework for South Africa’s digital future,” he said.

The department has imposed a Responsible AI Use Policy internally and is overhauling its drafting processes to ensure that “this type of occurrence does not happen again.”

Officials told MPs that consequence management is non‑negotiable if public trust is to be rebuilt.

Parliament

The department’s message to Parliament was clear: “The withdrawal was a necessary act of accountability”.

“The new process, anchored by expert oversight and disciplinary action, is designed to deliver a credible, evidence‑based AI policy that reflects South Africa’s digital priorities and restores confidence in the government’s ability to regulate emerging technologies,” it said.

Last month, AI expert and associate Professor at the Wits University School of Statistics and Actuarial Science, Professor Rendani Mbuvha, said the AI “blunder” underscored the irony of a human‑centred framework being undermined by AI hallucination and highlighted the urgent need to train policymakers to understand both the promise and the shortcomings of the technology.