The Pitfalls of AI in Legal Research: A South African Perspective

​​​​​​Artificial intelligence (AI) is revolutionising industries worldwide, and the legal field in South Africa is no exception. AI tools promise to enhance efficiency, enabling legal practitioners to draft documents, conduct research, and even predict case outcomes with unprecedented speed. However, as demonstrated in the recent South African cases of Mavunda v MEC: Department of Cooperative Governance and Traditional Affairs, KwaZulu-Natal (Mavunda) and Parker v Forsyth N.O. (Parker), the unchecked use of AI in legal practice can have serious consequences, including professional embarrassment, wasted resources, financial penalties and potentially adverse outcomes for clients.

SUMMARY OF mavunda judgment

In Mavunda, the applicant, Philani Godfrey Mavundla, sought to appeal against the court's findings and rulings contained in an order issued on 19 June 2024. Interestingly, the applicant filed his appeal just days after the order was granted, citing various grounds, even though the court had not yet provided the reasons for its judgment. These reasons were later delivered on 16 August 2024. Following this, a supplementary notice for leave to appeal was filed on 6 September 2024. The first respondent opposed the application for leave to appeal.

The court's primary task was to determine if Mavundla had reasonable prospects of success on appeal and if there were compelling reasons for the appeal to be heard. However, the court dismissed the application for leave to appeal, concluding that there were no reasonable prospects of success on appeal and no compelling reasons to justify granting leave, emphasising the stringent criteria for granting leave to appeal under the Superior Courts Act.

A key focus of the judgment denying leave to appeal was the applicant's apparent reliance on AI for its legal research. The supplementary notice of appeal contained numerous references to case authorities that were later found to be non-existent or incorrectly cited. This reliance on AI-generated research, without proper verification, caused unnecessary delays and costs which led to significant issues during the proceedings.

Judge Bezuidenhout criticised the applicant's legal representatives for failing to verify the accuracy of the AI-generated research. The judge stated "Relying on AI technologies when doing legal research is irresponsible and downright unprofessional." The judge emphasised the importance of human oversight in legal practice, noting that legal practitioners must ensure the accuracy and existence of cited authorities to avoid misleading the court.

The court further commented on the ethical responsibilities of legal practitioners by quoting from Chetty v Perumaul: "The legal practitioner’s duty to court requires that lawyers act with honesty, candour and competence... lawyers must not mislead the court and must be frank in their responses and disclosures to it."

The court ordered the applicant's attorneys to pay the costs incurred for the additional appearances, de bonis propriis (out of their own pockets).

Similarly, in Parker, attorneys shared a list of case authority with opposing counsel, which was later revealed to be fabricated by an AI tool. Although the error was unintentional, it misled opposing counsel and wasted the court’s time, leading to a punitive cost order.

These cases serve as cautionary tales, reminding legal practitioners of their ethical obligation to ensure accuracy and authenticity in their submissions. South African courts are increasingly aware of the role of AI in legal work but expect its outputs to be treated with the same level of scrutiny as traditional research methods.

Guidance for Legal Practitioners

To navigate the integration of AI responsibly, legal practitioners should adopt best practices in document production and the handling of evidence:

Verification Is Key

Always verify the output of AI tools against credible legal databases or primary sources. AI can generate plausible-sounding content, but it lacks the capacity to discern fact from fiction. Practitioners should cross-check any case law, statutory references, or legal principles provided by AI before including them in documents or arguments.

Document Production Standards

When using AI to assist in drafting pleadings, contracts, or other legal documents, apply the same standards of review as with manually prepared documents. Carefully proofread for factual accuracy, proper formatting, and logical consistency. AI should streamline your workflow, not replace your judgment.

Evidence Handling

AI can be a valuable tool for organising and analysing evidence, but it is not a substitute for due diligence. Practitioners must ensure that all evidence submitted is authentic, relevant, and admissible. If AI tools are used to assist in evidence processing, their role should be disclosed, and any results independently validated.

Ethical Use of Technology

Familiarise yourself with ethical guidelines and professional standards regarding the use of technology in legal practice. For instance, the use of AI should not undermine client confidentiality or compromise the quality of representation. Transparency about the extent of reliance on AI is crucial to maintaining trust with clients and the court.

the path forward

1. The use of AI in the provision of legal services is a rapidly growing trend across the African continent. As African nations continue to develop their technological infrastructure, the legal profession must adapt to these changes in order to preserve the integrity of judicial processes.

2. The introduction of AI presents both opportunities and challenges. On one hand, AI can help bridge gaps in legal resources, providing access to vast databases and analytical tools that can enhance legal research and case management. On the other hand, the reliance on AI without proper oversight can lead to errors and ethical dilemmas, as seen in the South African cases.

3. The integration of AI into the legal profession offers transformative potential, but it must be approached with caution. By combining AI's efficiencies with a strong framework of verification, ethics, and due diligence, legal practitioners can harness its capabilities while maintaining the integrity of the legal system. The cases of Mavunda and Parker remind us that innovation must be balanced with accountability.

4. The future of AI in South Africa and Africa's legal landscape is promising, but it requires a commitment to upholding the integrity and credibility of the legal system. By embracing AI responsibly, legal practitioners can leverage its benefits while safeguarding justice and fairness.

5. Used properly, AI technology can help deliver a more cost effective and often, more efficient legal solution to client, as long as all necessary checks are in place.