FCFCOA Sanctions Lawyer: GenAI Hallucinations Lead to Personal Costs

• Australia's FCFCOA sanctioned a lawyer with a personal costs order and public reprimand for improper use of generative AI.
• The lawyer's submissions in *Dey v Minister for Immigration and Citizenship* contained 11 AI-generated 'hallucinations' or fabricated legal authorities.
• The court ordered the lawyer to personally pay $9,600 in costs, emphasizing the need for lawyers to verify AI-generated content.
• The FCFCOA highlighted its inherent supervisory jurisdiction and the unacceptability of courts having to fact-check AI-generated legal research.
• This ruling follows the FCFCOA's recent release of an AI use practice direction and other judicial warnings about AI chatbots.

This ruling establishes a clear precedent for the personal liability of lawyers who fail to verify AI-generated legal research, leading to sanctions like personal costs orders and public reprimands.

Australia's Federal Circuit and Family Court (FCFCOA) has issued a personal costs order and a public reprimand against a lawyer for the improper use of generative artificial intelligence (genAI) in judicial review proceedings. The ruling came in the case of *Dey v Minister for Immigration and Citizenship (No 2) [2026] FedCFamC2G 1799*, where the court denied the applicant's request for judicial review. The Minister, as the first respondent, subsequently sought costs, including against the lawyer involved.

The court identified the lawyer's reliance on genAI in preparing both the initial costs submission and the primary submission for the judicial review application, both filed on March 1, 2026. These documents contained citations that were either entirely fabricated by the AI, known as 'hallucinations,' or did not support the legal propositions for which they were cited. After the court requested clarification, the lawyer admitted to the improper use of genAI, acknowledging that six authorities in the judicial review submissions and five in the first costs submissions were AI-generated fictions.

The lawyer accepted that his inappropriate use of genAI, particularly in the submissions filed on March 1, 2026, and later withdrawn on May 5, 2026, led to the Minister incurring additional costs related to the costs application. While expressing remorse and offering an apology, which the court found genuine, the lawyer sought to limit his personal liability for these costs to $1,500. However, the FCFCOA ordered the applicant to pay the Minister's costs fixed at $6,700 and, significantly, directed the lawyer personally to pay the Minister's costs fixed at $9,600.

The FCFCOA underscored its inherent supervisory jurisdiction over legal practitioners and the imperative of upholding the interests of justice in its decision. Despite finding no reason to doubt the lawyer's genuine remorse and deeming a repeat offense unlikely, the court determined that a personal costs order was clearly warranted. This ruling establishes a clear precedent for the personal liability of lawyers who fail to verify AI-generated legal research, leading to sanctions like personal costs orders and public reprimands.

The court emphasized its expectation that lawyers ensure the authenticity of all cited authorities, rather than relying on genAI programs that produce 'hallucinations.' It highlighted that lawyers must fulfill their professional duties and obligations when incorporating genAI into their work. The FCFCOA further noted that it was the court itself that initially detected the genAI use in the submissions and had to expend its own resources to fact-check the fabricated citations, a situation it deemed unacceptable.

This incident is not isolated within the FCFCOA's recent engagement with AI. On May 29, 2026, FCFCOA Chief Justice William Alstergren released a practice direction on AI use, developed by the court's AI committee following extensive internal and external consultation. Furthermore, in a separate unlawful termination claim decided on July 2, 2026, a Division 2 judge described the case as a 'cautionary tale about the dangers of befriending Artificial Intelligence (AI)-powered chatbots who masquerade as legal advisors,' signaling a broader concern within the court regarding unverified AI-assisted legal work.

The FCFCOA's decision to impose a personal costs order and a public reprimand, by naming the lawyer in its reasons and orders, sends a strong message regarding Australian lawyer AI misconduct penalty. This outcome directly addresses the critical need for lawyer professional duty AI verification, particularly when utilizing genAI for legal research. The court's actions highlight that the responsibility for the accuracy and veracity of legal submissions ultimately rests with the legal practitioner, regardless of the tools employed in their preparation.

This case serves as a significant warning to the legal profession about the risks associated with uncritical reliance on generative AI. It reinforces that while AI tools can assist in legal research, they do not absolve lawyers of their fundamental obligation to verify all information, especially legal authorities. The FCFCOA's proactive stance, including its new AI practice direction and judicial comments, indicates a growing focus on the ethical and professional implications of AI integration in legal services. Lawyers and compliance officers must implement robust verification protocols for all AI-assisted work to mitigate professional misconduct risks and ensure the accuracy of legal submissions.

This ruling establishes a clear precedent for the personal liability of lawyers who fail to verify AI-generated legal research, leading to sanctions like personal costs orders and public reprimands. Lawyers and compliance officers must implement robust verification protocols for all AI-assisted work to mitigate professional misconduct risks and ensure the accuracy of legal submissions.

Source: Original reporting via Australasian Lawyer