South Africa AI policy: official resigns over fake citations scandal

South Africa's communications minister, Solly Malatsi, confirmed on 15 October 2026 that one of two department officials suspended over fabricated academic references in the country's draft national AI policy has resigned. Disciplinary proceedings against the second official, who remains employed by the department, are still ongoing.

The government pulled the draft in May 2026 after it emerged that the document cited academic works that do not exist. According to Malatsi, the officials failed to disclose that AI tools had been used to prepare material for the policy, which allowed hallucinated citations to slip through unnoticed. He said the department's director-general acted after he ordered that those responsible be held to account.

"Having such a major stain on an important policy is something that cannot just be allowed to slide," Malatsi said, describing the withdrawal as necessary despite the setback it caused.

A revised timetable

The minister said he is confident Cabinet will finalise the replacement policy before the end of March 2027, with public consultation expected to start in April. Speaking at the CollectiveXchange 2026 digital skills summit in Johannesburg, he admitted the process was running very late because of its false start. He denied, however, that a new deadline had been set, saying an earlier reference to January 2027 and the end of the 2026/27 financial year were not separate targets.

The new draft is being prepared by an independent review panel led by Professor Benjamin Rosman of the Machine Intelligence and Neural Discovery Institute at the University of the Witwatersrand. Malatsi said consultation would begin with input from government departments, followed by civil society, residents, the private sector and other stakeholders.

Why it matters

The episode embarrassed a government that has sought to present South Africa as a continental leader in AI governance, and it raised wider questions about oversight when generative AI is used in drafting public policy. Malatsi argued that much of the withdrawn draft remained sound, including protections for citizens' rights and dignity online, and said the final framework must give people a way to seek redress when AI is used unethically.