South Africa’s Minister of Communications and Digital Technologies, Solly Malatsi, has withdrawn the Draft National Artificial Intelligence Policy after fictitious sources were discovered in the document’s reference list.

The minister announced the decision in a statement shared on his X account on Sunday.

He said the lapse constituted a serious breach that compromised the integrity and credibility of the proposed policy framework.

Malatsi said internal checks were initiated after concerns were raised about the references cited in the draft, confirming that several of the listed sources were not authentic.

He added that the issue was more than a technical oversight, describing it as a failure by the Department of Communications and Digital Technologies to uphold expected standards.

“This failure is not a mere technical issue but has compromised the integrity and credibility of the draft policy. As such, I am withdrawing the Draft National Artificial Intelligence Policy,” he said.

He said South Africans deserved better from an institution responsible for shaping the country’s digital policy landscape.

The minister added that the most probable explanation was that AI-generated citations were included in the draft without proper human verification.

“The most plausible explanation is that AI-generated citations were included without proper verification. This should not have happened,” Malatsi stated.

“In fact, this unacceptable lapse proves why vigilant human oversight over the use of artificial intelligence is critical. It’s a lesson we take with humility,” the Minister added.

He said the incident underscores the dangers of using AI tools without adequate oversight, particularly in public sector policymaking.

Malatsi said the government was taking the matter seriously and that appropriate accountability measures would be implemented.

“There will be consequence management for officials responsible for drafting the policy and carrying out quality assurance checks,”he said.

The withdrawal followed an earlier report by South African publication News24, which alleged that the AI policy contained at least six fictitious references in its bibliography.

Experts cited in the report said the fake sources bore the hallmarks of AI hallucinations, a phenomenon in which AI systems generate plausible-sounding but non-existent citations.

The revelations sparked criticism of the government’s drafting process and raised questions about the credibility of the proposed policy.

The withdrawal means South Africa may have to revisit parts of its consultation process on the national AI policy, potentially delaying efforts to put in place a formal framework to regulate and guide artificial intelligence adoption.

South Africa is among the few African countries to have developed an AI policy, even as the technology continues to spread rapidly across the continent.