The Irony of AI in South Africa's Policy Landscape

South Africa’s higher education sector has long prided itself on rigorous research standards upheld by bodies such as the Department of Higher Education and Training (DHET), the Council on Higher Education (CHE), and the National Research Foundation (NRF). Yet in April 2026, a national policy document intended to guide the responsible use of artificial intelligence became the centre of an embarrassing scandal involving AI-generated fictitious citations.

Background to the Draft National AI Policy

The Department of Communications and Digital Technologies (DCDT) published the Draft South Africa National Artificial Intelligence (AI) Policy for public comment on 10 April 2026. The 86-page document, approved by Cabinet, aimed to position AI as a driver of inclusive economic growth, capacity development, and human rights protection across sectors including education and research.

Discovery of Fabricated References

Investigative reporting by News24 revealed that at least six of the 67 academic citations in the bibliography were fictitious. Some referenced non-existent journals, while others credited real journals with articles that had never been published. Experts attributed the errors to AI hallucinations — instances where generative AI tools produce plausible but entirely invented information when data is insufficient.

Official Withdrawal and Ministerial Response

On 26 April 2026, Minister Solly Malatsi announced the withdrawal of the draft. In a public statement, he acknowledged that the inclusion of unverified, AI-generated citations compromised the document’s integrity. The minister emphasised that the lapse was not a minor technical issue but a serious failure of human oversight during the drafting process.

Photo by Filiz Elaerts on Unsplash

Implications for South African Universities and Research Integrity

The episode has sent ripples through the higher education community. Institutions such as the University of the Witwatersrand (Wits) and the University of Johannesburg (UJ) have highlighted the risks of unverified AI use in academic and policy work. The NRF, which funds much of the country’s research output, has faced renewed scrutiny over citation practices in grant applications and publications.

CHE and DHET guidelines on research quality now face pressure to incorporate explicit protocols for AI-assisted writing. Academics worry that similar hallucinations could infiltrate peer-reviewed outputs or student theses, undermining South Africa’s standing in global university rankings.

Stakeholder Perspectives from Academia and Government

Parliament’s Portfolio Committee on Communications and Digital Technologies called for accountability, with some members demanding suspensions and independent reviews. Academic commentators at Wits noted that the scandal exposed a broader gap in digital literacy training across government departments and universities alike.

Student organisations and early-career researchers have expressed concern that eroded trust in official AI guidance could slow adoption of beneficial tools in teaching and research at institutions such as Stellenbosch University and the University of Cape Town.

Lessons for Responsible AI Integration in Higher Education

The retraction underscores the need for mandatory human verification of all AI-generated content in policy and scholarly work. Universities are being urged to update their research integrity policies to include AI disclosure requirements and training modules on detecting hallucinations.

Professional development programmes offered through bodies like Universities South Africa (USAf) are expected to expand, focusing on ethical AI use in grant writing, literature reviews, and curriculum development.

Future Outlook for National AI Policy and Research Governance

Minister Malatsi indicated that a revised draft would be prepared with stricter verification processes. Higher education stakeholders anticipate closer collaboration between DCDT, DHET, and the NRF to ensure the next iteration supports rather than undermines research excellence.

Long-term, the episode may accelerate the development of national standards for AI in scholarly publishing, potentially influencing CHE accreditation criteria and NRF funding conditions.

Photo by Swiss Educational College on Unsplash

Actionable Steps for Academics and Administrators

University leaders are advised to audit current AI tools in use, implement citation verification workflows, and foster cross-departmental dialogues on responsible innovation. PhD candidates and postdoctoral researchers should prioritise transparent disclosure of AI assistance in their work to maintain credibility with international collaborators.