South Africa Withdraws AI Policy Over Fake AI-Generated References

South Africa has withdrawn its first draft national AI policy after it was found to contain fake or AI-generated references that were not properly verified. The government said the error undermined the document’s credibility and blamed poor oversight in using AI tools. The draft had proposed major plans for AI regulation, ethics, and innovation but will now be revised, with no date yet set for a new version.

South Africa has withdrawn its first draft national Artificial Intelligence policy after serious concerns emerged that parts of the document contained fake or non-existent academic and policy references, believed to have been generated by artificial intelligence tools without proper human verification.
The draft policy, which had only recently been released for public consultation, was intended to serve as a foundational framework for how South Africa plans to develop, regulate, and benefit from artificial intelligence technology. It was designed to position the country as a continental leader in AI innovation while also setting guidelines to manage the ethical, social, and economic risks associated with the rapid expansion of AI systems.
However, controversy erupted when it was discovered that several of the citations listed in the policy’s reference section could not be verified. Some sources appeared to be entirely fictitious, raising immediate questions about the drafting process and the level of oversight applied before the document was made public. Officials later acknowledged that AI-generated content may have been included without adequate fact-checking.
The Minister of Communications and Digital Technologies, Solly Malatsi, confirmed the withdrawal and described the situation as a serious lapse in judgment. He stated that the most plausible explanation was that artificial intelligence tools had been used in preparing the document, but the outputs were not properly reviewed or validated by human experts. According to him, this failure undermined both the credibility and integrity of the draft policy.
Malatsi further emphasized that the incident should not be treated as a minor technical mistake, but rather as a significant governance issue. He noted that it demonstrates the risks of relying on AI systems without strict human oversight, particularly in official government documentation that is meant to guide national strategy.
Despite the setback, the draft policy had outlined ambitious plans for the country’s AI future. These included the creation of new institutions such as a National AI Commission to coordinate strategy, an AI Ethics Board to oversee responsible use, and an AI Regulatory Authority to enforce compliance and standards. The policy also proposed economic incentives, including tax breaks, grants, and subsidies, aimed at encouraging private-sector investment and collaboration in AI development.
The goal of these proposals was to strengthen South Africa’s digital economy and ensure that the country remains competitive in the global technology landscape while also addressing concerns around job displacement, data protection, and ethical use of AI.
Following the incident, Malatsi stated that there would be consequences for those involved in the drafting process, although he did not provide specific details. He also did not give a timeline for when a corrected version of the AI policy would be reissued, but indicated that lessons had been learned regarding oversight and accountability.
The withdrawal has sparked broader discussions about the role of artificial intelligence in policymaking and the need for stronger safeguards to ensure that AI tools are used responsibly, especially in sensitive government work.