INTRODUCTION

South Africa presently has no existing laws governing the use of Artificial Intelligence (“AI”) although current, existing legislation merely touches on AI aspects. It is crucial to gain a comprehensive understanding of the role that AI plays in the South African legal sphere. It is of paramount importance to delve into AI mechanisms to examine the impact such processes have on legal practise.

CASE LAW

Northbound Processing (Pty) Ltd v South African Diamond and Precious Metals Regulator and Others:

The court in this case referred to the applicant’s heads of arguments containing non-existing case citations that were “hallucinated” by generative artificial intelligence¹. The counsel for the applicant admitted the fact that the citations constituted “hallucinations” and counsel based their defence on the fact that they were under tremendous time pressure and that they relied on an online subscription tool known as “Legal Genius” which is exclusively trained to South African legal judgements and legislation. The court referred the above issues to the Legal Practice Council for further investigation, and the registrar of the court was directed to bring this judgement to the attention of the Gauteng Provincial Office of the Legal Practice Council.

The case highlights the broader risk of the emergence of specialised AI systems tailored to specific use towards legal practice and will most likely threaten the critical thinking skills and independent, sound-mind reasoning that is required of legal practitioners.

Mavundla v Member of the Executive Council, Department of Co-operative Government and Traditional Affairs:

Reference is once again made to legal practitioners resorting to the use of AI because of being under a lot of pressure². This highlights the issue that in such a fast-paced profession AI is highly relied upon by many practitioners as it is an easy yet unreliable source. The attorneys in the above matter resorted to the use of ChatGPT for finding legal cases that did not exist. This case takes cognisance of the fact that AI is making legal practitioners lazy and unbothered, as in the above case it is stated that the plaintiff used ChatGPT to generate research and the plaintiff therefore lazily omitted to verify the research. Relying on AI tools when compiling legal research is in the judge’s view irresponsible and downright unprofessional. The registrar of the court sent a copy of the judgement in this case to the Legal Practice Council for further action, with regards to the attorney’s reliance on AI, and the disciplinary proceedings that are to follow.

In Mavundla, the court emphasised the trite duty of legal practitioners not to mislead the court, whether through negligence or intent. This includes the duty to present an honest account of the law, which means not presenting fictitious cases.

The reasoning in Mavundla, is consistent with the judgement of the English High Court in Ayinde v the President of the King’s Beach Division wherein the following observations were made with regards to the use of AI in legal research:

1. In the context of legal research, the risks associated with using artificial intelligence are widely known³. AI tools can produce apparent plausible responses; however, these plausible responses may be revealed to be erroneous. Sources that are of no existence may be cited, and they may appear to quote a passage from a genuine source, that does not in fact appear in that source.
2. Those who rely on AI to conduct legal research, have a professional duty to always check on the accuracy of the research and whether it makes reference to authoritative sources, prior to making use of it within their professional, legal work.
3. There are serious implications for the administration of justice and public confidence in the justice system if AI is misused. Practical and effective measures must be taken by professionals within the legal sphere. These measures are a guiding principle to ensure every individual complies with their ethical and professional obligations and their duties to the court, if they make use of AI at their disposal.
4. The court has inherent powers that ensure legal practitioners comply with their ethical duty to the court. In circumstances when these duties are not adhered to the court’s response is one of immediate action through disciplinary measures, such as imposing a costs order, imposing a wasted costs order, striking out a case, initiating contempt proceedings and referring the case to the police.

Parker v Forsyth:

The plaintiff’s attorneys made cite of specific case law in their arguments that the defendant’s attorneys were incapable of locating their source in law⁴. Once the plaintiff’s attorneys were questioned with regards to their use, it became evidently clear that these cases were generated through the means of ChatGPT. The court voiced grave concerns on the unsubstantiated dependence on AI. As punishment, punitive costs were accounted for in this regard, as the introduction of erroneous and fictitious case law was an attempt to defraud the court and an attempt to obstruct the justice system.

CONCLUSION

Strict timelines, being under a lot of pressure, as well as lazily omitting to verifying the search are all substantive reasons as to why there is such a rapid increase in legal practitioners relying on AI for the use of their professional work. The courts and the judicial system have in response to this imposed harsh penalties, strict costs orders as well as referral to the Legal Practise Council for disciplinary action. As much as measures are put in place to regulate the unethical, misuse of AI it is up to the legal practitioners themselves to ensure that they adhere to and comply with ethical and professional legal standards.