South Africa’s AI Policy Under Fire for Fake Citations

A draft artificial intelligence policy released by South Africa’s Department of Communications and Digital Technologies (DCDT) has been withdrawn after critics discovered fabricated citations within the document.

The policy, intended to decentralize AI oversight with different agencies monitoring various risk levels, included references to articles that either didn’t exist or were not published in reputable academic journals. News24 reported that at least six citations appear to be completely fabricated, potentially generated by AI itself—a particularly embarrassing discovery for a document designed to govern responsible technology use.

Minister of Communications and Digital Technologies Solly Malatsi responded swiftly, ordering an investigation into the matter and confirming the accuracy of the criticism. He attributed the errors to “AI-generated citations that were included without proper verification,” adding that those responsible would face consequences.

The incident highlights both the growing importance of AI governance frameworks in Africa—where countries like Nigeria are exploring similar approaches—and the risks of relying on automated systems for critical policy development.