Home Affairs suspends two officials over AI use linked to revised White Paper on Citizenship and Immigration

2026-04-30 · Source: Welcome to the Artificial Intelligence Incident Database · Field: Government & Public Sector — Public Policy & Governance, Regulatory & Compliance, Digital Government & E-Government · Depth: Fundamental Awareness, quick

Summary

The South African Department of Home Affairs (DHA) has suspended two senior officials, a Chief Director and a Director, following the discovery of apparent Artificial Intelligence (AI) "hallucinations" in the reference list of its recently Cabinet-approved Revised White Paper on Citizenship, Immigration and Refugee Protection. The suspensions are precautionary, and the DHA has appointed two independent law firms to manage disciplinary actions and review all policy documents produced since November 30, 2022, when Large Language Models became publicly available. The department will also implement new AI checks and declarations in its internal approval processes. While the problematic reference list has been withdrawn, the DHA confirms the White Paper's main body accurately reflects government policy and was developed through extensive collaboration. The DHA apologized for the oversight, viewing this incident as an opportunity to modernize its operations.

Key takeaway

For government agencies or policy-making bodies integrating AI tools, you must establish robust verification protocols for all AI-generated content, especially references or factual claims. This incident highlights the critical risk of "hallucinations" undermining document credibility and necessitating costly reviews. Implement mandatory AI checks and declarations in your approval workflows to prevent similar public embarrassments and ensure policy integrity. Proactively define acceptable AI use to avoid severe operational disruptions and maintain public trust.

Key insights

Unverified AI-generated content in official documents can lead to significant organizational repercussions and policy reviews.

Principles

• AI outputs require rigorous human verification.
• Public sector AI adoption demands new governance.
• Uncited references undermine document credibility.

Method

The DHA initiated precautionary suspensions, engaged independent legal reviews of past documents, and plans to implement AI checks and declarations.

In practice

• Implement mandatory AI content verification steps.
• Review existing policy documents for AI-generated content.
• Develop internal guidelines for AI tool usage.

Topics

• AI Hallucinations
• Public Sector AI
• Policy Document Integrity
• Government Accountability
• Large Language Models
• Digital Governance

Best for: CTO, VP of Engineering/Data, Director of AI/ML, Policy Maker, Legal Professional, Consultant

Related on AIssential

• South Africa yanks AI policy after AI-assisted drafting invents citations — AI-generated content, especially citations, requires rigorous human verification to maintain integrity.
• Understanding AI Hallucinations: Why AI Sometimes Gets Things Wrong — AI hallucinations are fabricated outputs from models predicting patterns, not truth, requiring user vigilance.
• Researchers who use hallucinated references to face arXiv ban — Preprint servers are implementing strict policies to combat unchecked generative AI use and maintain scientific integrity.
• Hallucinated citations are polluting the scientific literature. What can be done? — AI-generated hallucinated citations are rapidly increasing, polluting academic literature with fabricated and inaccurate references.
• Every AI output is a hallucination — Large Language Models inherently produce fluent, confident outputs that are often wrong yet indistinguishable from truth, posing significant reliability issues.
• Sullivan & Cromwell apologises to US court for filing errors caused by AI hallucinations - Bar and Bench — AI hallucinations pose significant risks in legal contexts, requiring strict adherence to verification protocols.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Welcome to the Artificial Intelligence Incident Database.