AI Hallucinations Strike Again at South African Courts

Artificial intelligence (AI) is transforming the way businesses work. From drafting documents to conducting research, AI tools are helping professionals complete tasks faster than ever before.

However, a recent Labour Court judgment has highlighted one of the biggest risks associated with using AI for legal research: AI hallucinations.

The court halted disciplinary proceedings involving three employees of a certain local Municipality after finding prima facie evidence that the chairperson of the disciplinary hearing may have relied on AI-generated legal authorities that could not be verified. The ruling has raised important questions about the integrity of disciplinary proceedings and the responsibility of decision-makers who use AI-generated information.

What Happened?  

The matter arose during disciplinary proceedings against three municipal employees.

The employees challenged two rulings made by the chairperson of the disciplinary hearing, arguing that several legal authorities relied upon in those rulings either did not exist or had been cited inaccurately. Apart from one correctly cited case, they alleged that the authorities references could not be verified and did not support the legal principles for which they were cited. They further argued that the citations appeared to have been generated using artificial intelligence, raising concerns about whether the chairperson had independently applied his mind to the issues before him.

After considering the evidence, the Labour Court found that several of the cited judgments could not be verified.

Although the court did not conclude that AI had definitely been used, it found there was sufficient evidence to question the chairperson’s reasoning and decision-making process. As a result, the disciplinary proceedings were stayed pending the outcome of a review application.

Why the Court Took the Matter Seriously  

The court made it clear that the issue was not simply whether AI had been used.

Rather, the concern was whether the legal authorities relied upon were accurate and whether the chairperson had exercised independent judgement before issuing the rulings.

The municipality argued that the chairperson may simply have intended to refer to different legal authorities and that the legal principles contained in the rulings remained correct.

The court rejected this argument.

The judge noted that if the citations had indeed been generated by an AI assistant and accepted without proper verification, such conduct would be inconsistent with the standards expected by South African courts. Even if the errors were not caused by AI, the chairperson had been given an opportunity to explain the incorrect citations but failed to do so.

AI Hallucinations Are Becoming a Growing Concern  

The judgment is the latest in a growing number of cases highlighting the risks of relying on generative AI for legal research.

An AI hallucination occurs when an AI system produces information that appears convincing but is inaccurate or entirely fabricated. This can include non-existent court cases, incorrect legal citations or quotations that do not exist in the original judgments.

The Labour Court referred to several previous judgments that have already warned legal professionals about these risks.

Among them was the 2023 Gauteng High Court decision in Parker v Forsyth, where the court emphasised that legal practitioners must apply an independent and questioning mind rather than simply repeating unverified research produced by a chatbot. The judgment also referred to a recent English High Court decision warning that AI systems can generate seemingly plausible responses that contain fabricated authorities and quotations. In addition, the court cited a July 2026 judgment of the Supreme Court of India, which found that reliance on non-existent or hallucinated material undermines the integrity of judicial decision-making.

This is not the first time South African courts have dealt with AI-generated citations. In 2025, the Gauteng High Court ordered an investigation into legal practitioners after fictitious case citations generated by AI were submitted to the court. The judge stressed that anyone using AI for legal research has a professional duty to verify its accuracy before relying on it.

What Does This Mean for Employers?  

While most employers are unlikely to appear before the Labour Court, many businesses are already using AI to assist with HR administration, disciplinary documentation, employment contracts and workplace policies.

The recent judgment serves as an important reminder that AI should be viewed as a support tool rather than a substitute for professional judgement.

Whether information is prepared by AI or by a person, employers remain responsible for ensuring that disciplinary procedures are fair, decisions are properly reasoned and legal references are accurate.

As AI becomes more common in the workplace, businesses should ensure that any information generated by these tools is carefully reviewed before being relied upon for employment or legal decisions.

Key Takeaways for Employers  

The Labour Court judgment reinforces several important lessons:

• AI can improve productivity, but it is not infallible.

• Legal authorities and references should always be verified before being relied upon.

• Decision-makers must apply their own independent judgement rather than accepting AI-generated information at face value.

• Fair disciplinary procedures depend on accurate information and proper legal reasoning.

• Employers remain accountable for workplace decisions, regardless of the technology used to assist them.

Final Thoughts  

Artificial intelligence is becoming an increasingly valuable business tool, and its use is likely to continue growing across every industry.

However, the latest Labour Court judgment demonstrates that efficiency should never come at the expense of accuracy.

Whether drafting documents, conducting research or preparing disciplinary rulings, AI should assist human decision-making—not replace it. The responsibility for ensuring that information is accurate, reliable and legally sound will always remain with the people making the decisions.

Need Guidance on HR and Labour Compliance?  

Technology is changing the workplace, but compliance still depends on informed decision-making.

If you need assistance with disciplinary procedures, employment contracts, workplace policies or HR compliance, Chamlabour’s experienced consultants are here to help. Book your FREE HR & IR Compliance Health Check and ensure your business remains compliant in an evolving workplace.