South African Lawyers Face Consequences for Using Fake References Generated by ChatGPT
Lawyers arguing a case at the Johannesburg regional court have been reprimanded for relying on fake references generated by ChatGPT, an artificial intelligence language model. The judgment revealed that the names, citations, facts, and decisions presented by the lawyers were all fictitious. Additionally, the lawyers’ client was issued a punitive costs order.
The case in question involved a woman who sued her body corporate for defamation. The counsel for the body corporate trustees argued that a body corporate couldn’t be held liable for defamation. On the other hand, Michelle Parker, the plaintiff’s counsel, claimed that there were previous judgments addressing the issue; however, they had not been able to access them due to time constraints.
To give both parties adequate time to source the necessary information, Magistrate Arvin Chaitram postponed the case until late May. In the subsequent two months, the lawyers involved in the case attempted to locate the information referred to in ChatGPT’s generated references.
Unfortunately, it was discovered that while ChatGPT had provided actual case names and citations, they were unrelated to the ones mentioned by the lawyers. Moreover, the cases and citations retrieved through ChatGPT were entirely inappropriate for defamation suits involving body corporates and individuals.
Eventually, the lawyers admitted that they had sourced the judgments through the medium of ChatGPT. Magistrate Chaitram ruled that the lawyers did not intend to mislead the court; rather, they were simultaneously simply overzealous and careless. Consequently, no further action was taken against the lawyers besides the punitive costs order.
This incident is not unique to South Africa. In the United States, lawyers were fined $5,000 for submitting a court brief containing false case citations generated by ChatGPT. The lawyers in question, Steven Schwartz and Peter LoDuca, along with their firm, were accused of neglecting their responsibilities by submitting fabricated opinions, quotes, and citations. The brief was filed on behalf of Roberto Mata, who alleged injuries from a 2019 flight from El Salvador to New York. Mata’s case was dismissed because it was filed after the statute of limitations had expired.
US District Judge P. Kevin Castel also required the lawyers to send a transcript of the hearing and the opinion to each of the falsely identified judges mentioned by ChatGPT. However, Judge Castel refrained from demanding a compelled apology, stating that a sincere apology cannot be forced.
Instances like these serve as a reminder that while technology can be a helpful tool for legal research, it should always be blended with traditional independent reading and verification. The efficiency of modern technology should not overshadow the importance of thorough and diligent research.
Lawyers and legal professionals should exercise caution when utilizing language models like ChatGPT, ensuring that the information they retrieve from such tools is double-checked for accuracy and relevance. Upholding the integrity of the legal profession requires a strong commitment to providing accurate and reliable information to the courts.
While this incident may have caused embarrassment for the lawyers involved, it serves as a valuable lesson for legal professionals worldwide, emphasizing the need to adhere to rigorous research practices and exercise due diligence in presenting their arguments.
