Withdrawal Of South Africa’s Draft National Artificial Intelligence Policy

In April 2026, the South African government released its Draft National Artificial Intelligence Policy (“Draft AI Policy”) for public consultation, marking an important milestone in the country’s efforts to develop a coordinated framework for artificial intelligence (“AI”) governance. Following the publication of the Draft AI Policy, concerns were raised regarding the authenticity of a number of academic sources cited in the document. Investigations confirmed that several cited articles had not been published in the journals referenced and that the works attributed to specific authors did not exist.

In response, the Minister of Communications and Digital Technologies announced the withdrawal of the Draft AI Policy on 26 April 2026. The Minister characterised the matter as a serious failure that went beyond a technical or editorial error, noting that the presence of fictitious sources undermined the credibility and integrity of the policy. The Department subsequently confirmed that the most likely explanation was that AI‑generated content had been incorporated into the policy without sufficient verification and quality assurance. The Minister further indicated that internal processes would be reviewed and that accountability measures would follow in respect of those responsible for drafting and oversight. At the time of withdrawal, no revised date was announced for the re‑publication of the policy.

While the withdrawal of the Draft AI Policy has delayed the formalisation of South Africa’s AI governance framework, the incident offers a number of important lessons from a governance and legal perspective.

First, it underscores the limitations and risks associated with the use of generative AI tools in policy development, particularly where outputs are relied upon without rigorous human review. The generation of plausible‑looking but inaccurate information, including fabricated references, is a known limitation of current AI systems.

Secondly, the decision to withdraw the policy reflects an emphasis on institutional credibility and administrative integrity. Notwithstanding prior Cabinet approval and the commencement of public consultation, the withdrawal signals that evidentiary accuracy and transparency are essential prerequisites for legitimate and effective regulation, especially in technologically complex and rapidly evolving fields such as AI.

Finally, the incident highlights the need to regulate AI responsibly while simultaneously integrating AI tools into internal administrative and drafting processes. The incident mirrors the very risks that modern AI governance frameworks seek to address, including the need for accountability, traceability and human oversight in AI‑assisted decision‑making.

The withdrawal of the Draft AI Policy should not be understood as a retreat from South Africa’s commitment to AI governance. Public statements indicate that the policy will be revised and reissued following corrective measures. When this occurs, it is expected that enhanced verification and quality‑control processes will form a central part of the revised approach. In the meantime, companies should consider how the Draft AI Policy would apply to them and begin to take proactive steps to comply with the policy.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.