Senior Practitioner Referred to Regulator for Incorporating Fake AI Generated Decisions in Submissions

In Valu v Minister for Immigration and Multicultural Affairs (No 2) [2025] FedCFamC2G 95 (31 January 2025), Judge Skaros of the Federal Circuit and Family Court of Australia (Division 2) referred the legal advocate for the applicant to the relevant legal Regulator on account of such advocate including in their primary submissions fake case references. These arose from “hallucination” by the practitioner’s use of Generative AI in formulation of those submissions.  His Honour wrote (the practitioner’s name was anonymised to “the ALR”):

…

[28]  Counsel submitted that the ALR was at the end of his legal career and had experienced a range of health issues that deteriorated or affected his performance. Counsel implored the Court to distinguish Dayal [2024] FedCFamC2F 1166 (Dayal) on the basis that, in the present case, the fake cases were not referenced in the final hearing as the ALR had filed an amended application and outline of submissions. Counsel submitted that the ALR was profoundly embarrassed by the conduct and had taken steps to improve his knowledge of AI.

[29]  Counsel further submitted that the Practice Note issued by the Supreme Court of NSW has provided clarity on the subject of AI, and that people would not be assisted by another re-examination of the ALR’s case which would only serve to embarrass him, even if he is not personally identified.

[30]  The Minister submitted that the ALR had failed to exercise adequate care in allowing submissions which contained false citations and false quotes to be filed. The Minister referred to the matter of Dayal in which her Honour Judge A Humphreys directed the referral of the solicitor in that matter to the Victorian disciplinary authorities in similar circumstances. It was submitted that the same course should follow in the present case. The Minister submitted that the misuse of artificial intelligence in legal proceedings was a matter of current public interest. It was submitted that such misuse would likely be a matter of increasing concern in future, and there was a public interest in ensuring the OLSC is aware of the misuse of AI in cases as they arise.

[31]  In oral submissions, the Minister submitted that Dayal should not be distinguished and that the conduct of the ALR, in providing fake quotes from the Tribunal’s decision as well as fake case citations to the Court, went beyond the conduct of the solicitor in Dayal . It was submitted that such conduct would continue to occur and must be “nipped in the bud”.

[32]  The Court notes that in Dayal, the solicitor in that matter provided the Court with a list of fictional authorities and case summaries which had been generated using an AI tool within the LEAP Practice Management Software. Her Honour, citing the USA District Court case of Mata v Avianca Inc 678 F. supp. 3d 443 (S.D.N.Y. 2023), set out the potential harms that could result from the filing of submissions which contain false information. Her Honour referred to the duties of legal practitioners and considered that the solicitor in that matter had breached the professional standards expected of them. Her Honour accepted that the solicitor’s apology was genuine and that the conduct would unlikely be repeated, but nevertheless considered it in the public interest that the relevant regulatory authority (the Victorian Legal Services Board and Commissioner) was made aware of the professional conduct issues that arose in that matter given the increased use of AI tools by legal practitioners in litigation.

[33]  The circumstances of the present matter raise similar issues to those in Dayal.

[34]  The Court accepts that the ALR genuinely regrets his conduct. He acknowledged that the provision of false case citations and quotes alleged to be from the Tribunal’s decision was in breach of his duty to his client and to the Court. The Court accepts that the conduct will not be repeated and that the ALR has undertaken to further his knowledge and understanding of the risks of using generative AI tools.

[35]  The Court accepts that the ALR, as soon as becoming aware of the non-existent case citations and Tribunal quotes in the submissions, took steps to provide amended submissions. However, by the time this occurred, the Court and the Associates had already spent a considerable amount of time attempting to locate the cases and enquiring as to whether the correct Tribunal decision had been filed with the Court. The Minister had also prepared their submissions in reply and filed them. As a consequence of the ALR’s conduct, the matter could not conveniently proceed, and further timetabling was required to facilitate the proper filing of an amended application and submissions on behalf of the applicant. The ALR’s conduct created unnecessary additional work for the Court and the Minister. The ALR acknowledged the inconvenience to the Court (and the Minister) caused by his conduct and was apologetic.

[36]  The Court accepts that the ALR is deeply embarrassed by his conduct, and it is certainly not the Court’s intention to embarrass him any further by referring the conduct to the OLSC. The Court has anonymised the name of the ALR to protect his identity. If the Court refers the conduct to the OLSC, a preliminary assessment will be undertaken to ascertain whether any further action or investigation is warranted. In making its assessment, the OLSC will take into account the ALR’s experience as a legal practitioner, his advancing age, his significant health issues, and the regret and remorse he has expressed for his conduct. Consideration of these factors and whether they mitigate the ALR’s conduct will of course be a matter for the OLSC.

[37]  There is a strong public interest in referring this conduct to the regulatory authority in NSW given the increased use of generative AI tools by legal practitioners. The use of generative AI in legal proceedings is a live and evolving issue. While the Supreme Court of NSW has issued guidelines around the use of generative AI, other Courts, including this Court, are yet to develop their guidelines. The Court agrees with the Minister that the misuse of generative AI is likely to be of increasing concern and that there is a public interest in the OLSC being made aware of such conduct as it arises.

[38]  Having regard to all the circumstances, the Court considers it in the public interest to refer the ALR’s conduct to the OLSC and will so direct that the Principal Registrar of this Court to refer the matter accordingly, and to provide the OLSC with a copy of the Court’s reasons for the direction, a copy of the judgment in the substantive proceedings, the ALR’s affidavit, the written submissions of the parties and the amended applications field 3 December 2024.

(emphasis added)

The full decision may be found on this link.