Key Takeaways
- Two DCDT officials were placed on precautionary suspension after fictitious references were found in the Draft South Africa AI Policy.
- Minister Solly Malatsi withdrew the policy and ordered an internal investigation led by Director‑General Nonkqubela Jordan‑Dyani.
- The Portfolio Committee on Communications and Digital Technologies questioned the minister and officials on the breach, the suspension process, and funding for a new advisory panel.
- Investigations traced the fabricated references to a Chilean document that had been translated from Spanish to English and used as a drafting source.
- A seven‑member National AI Expert Review Panel, chaired by Prof. Benjamin Rosman, has been appointed to strengthen the policy’s integrity and oversight mechanisms.
Background on the Draft AI Policy
The Department of Communications and Digital Technologies (DCDT) released a draft Artificial Intelligence (AI) Policy for public comment in mid‑2026, outlining an ambitious governance framework. The proposal called for the creation of a National AI Commission, an AI Ethics Board, an AI Regulatory Authority, an AI Ombudsperson, a National AI Safety Institute, and an AI Insurance Superfund intended to compensate individuals harmed by AI systems where liability is unclear. The draft was positioned as a cornerstone of South Africa’s strategy to harness AI while safeguarding public interests, and it invited stakeholder input before finalisation.
Discovery of Fabricated References
During the public consultation period, reviewers identified several entries in the policy’s reference list that did not correspond to any verifiable sources. These fictitious citations raised immediate concerns about the document’s scholarly rigor and the department’s internal vetting procedures. The discovery prompted Minister Solly Malatsi to halt the policy’s advancement, citing a compromise to its integrity that necessitated a thorough investigation before any further public engagement could proceed.
Ministerial Response and Withdrawal
Upon learning of the problematic references, Minister Malatsi requested that Director‑General Nonkqubela Jordan‑Dyani launch an internal inquiry. He emphasized that the withdrawal was a precautionary measure to preserve public trust and to allow the department to address the shortcomings without prejudice. Malatsi stressed that the action was not an admission of guilt but a procedural step aligned with natural justice, public service regulations, and labour law requirements for consequence management.
Parliamentary Portfolio Committee Scrutiny
The Portfolio Committee on Communications and Digital Technologies convened a special session to examine the fallout from the policy’s withdrawal. Committee members grilled both the minister and the director‑general on how the error occurred, the rationale behind placing officials on precautionary suspension, and the financial implications of establishing an external advisory panel. The hearing underscored the committee’s role in overseeing executive accountability and ensuring that remedial actions are transparent and fiscally responsible.
Precautionary Suspension of Officials
As part of the investigation, Director‑General Jordan‑Dyani placed two unnamed DCDT officials on precautionary suspension. She clarified that such suspension does not imply guilt but serves to protect the integrity of the inquiry by removing potential interference while safeguarding the officials’ rights under natural justice. The measure was intended to allow investigators to work unimpeded and to uphold procedural fairness prescribed by public service and labour regulations.
Internal Investigation Findings
Jordan‑Dyani reported that the investigation had progressed to its final stages and traced the origin of the fabricated references to a specific external document. She disclosed that officials had consulted a Chilean source, which had been translated from Spanish into English, and that this document contained the erroneous citations that subsequently appeared in the AI policy’s bibliography. The revelation pointed to a breakdown in source verification rather than deliberate misconduct, though the investigation remained ongoing to determine any lapses in due diligence.
Advisory Panel Formation
To rectify the oversight gaps that allowed the AI‑generated hallucinations to enter the policy, Minister Malatsi announced the creation of a seven‑member National AI Expert Review Panel. Chaired by Prof. Benjamin Rosman of the Wits Machine Intelligence and Neural Discovery Institute, the panel includes experts from academia, law, cybersecurity, and digital policy: Prof. Vukosi Marivate (UP), Alison Gildwald (Research ICT Africa), Heather Irvine (Bowmans), Dr Tshepo Feela (National Planning Commission), Dr Jabu Mtsweni (CSIR), and Advocate Lufuno Tshikalange (Tshikosi Attorneys). The panel’s mandate is to review the AI policy, strengthen authentication processes, and restore confidence in the department’s outputs.
Explanation of Source Document from Chile
Jordan‑Dyani elaborated that the Chilean document, which officials had used as a contextual reference, had undergone translation from Spanish to English before its incorporation into the draft policy. During this process, the erroneous references were inadvertently carried forward, ultimately appearing in the policy’s bibliography. She noted that while the use of external sources is standard practice, the incident highlighted the necessity for rigorous cross‑checking and validation, especially when relying on translated materials that may introduce transcription errors or misattributions.
Committee Chair’s Concerns and Closing Remarks
Committee chair Khusela Sangoni‑Diko expressed skepticism about the explanation, questioning whether reliance on a translated foreign document adequately accounted for the presence of fictitious sources. She acknowledged that time constraints limited further probing but warned that lingering uncertainties could erode legislative confidence in the department’s capacity to govern emerging technologies responsibly. Sangoni‑Diko concluded by urging the DCDT to expedite the investigation, provide clear updates, and ensure that remedial measures are both effective and transparent.
Implications and Next Steps
The episode underscores the challenges governments face in integrating AI‑assisted drafting tools while maintaining rigorous scholarly standards. Moving forward, the DCDT must implement robust verification protocols, potentially including AI‑usage disclosure requirements and mandatory source audits, to prevent recurrence. The newly constituted expert panel will play a pivotal role in reshaping the AI policy, and its recommendations will likely influence the final version slated for public comment. Stakeholders will watch closely to see how the department balances innovation with accountability in South Africa’s AI governance journey.