South Africa’s debut AI policy collapsed within weeks after fabricated, AI-generated citations exposed a critical oversight failure.
South Africa has withdrawn its first ever draft national artificial intelligence (AI) policy less than three weeks after its publication, after investigators found that parts of the document contained fabricated, AI-generated academic references.
The Department of Communications and Digital Technologies released the Draft South Africa National Artificial Intelligence Policy for public comment on April 10, positioning the country as an emerging leader in AI innovation. However, just 16 days later, the document was pulled after inaccuracies were identified in its citations.
The country’s communications minister, Solly Malatsi, confirmed that at least six of the policy’s 67 references were “hallucinated” by AI, pointing either to non-existent academic journals or to articles that were never published in legitimate publications, raising serious concerns about oversight, credibility and governance.
“The most plausible explanation is that AI-generated citations were included without proper verification. This should not have happened,” Malatsi said, adding that the lapse had “compromised the integrity and credibility” of the policy.
The issue first came to light following an investigation by a South African broadcaster, which flagged discrepancies in the references. Editors of established journals, including the South African Journal of Philosophy, AI & Society, and the Journal of Ethics and Social Philosophy, confirmed that the cited articles did not exist, British newspaper, The Independent reported.
The now-withdrawn draft sought to establish a comprehensive AI governance framework, including proposals for a national AI commission, an ethics board and a regulatory authority. It also outlined tax incentives, grants and subsidies to boost private-sector investment in AI infrastructure, while addressing the ethical, social and economic challenges posed by the technology.
However, the discovery of fabricated citations, likely generated using tools such as ChatGPT or Google Gemini, has derailed those plans.
The minister acknowledged that the issue went beyond a technical error and reflected a failure of human oversight. “This unacceptable lapse proves why vigilant human oversight over the use of artificial intelligence is critical,” he said, adding that accountability measures would follow.
The AI citations not only invented sources but also falsely attributed research to credible journals and scholars, creating an illusion of authority. Analysts warn this mirrors broader risks associated with generative AI, including misinformation, deepfakes and the manipulation of digital identities.
The government is expected to revise the draft policy before reissuing it for public consultation.
