The Queensland Courts have recently released guidelines for judicial officers and updated guidelines for non-lawyers, including self-represented litigants, in relation to the use of generative artificial intelligence (generative AI) in litigation.
The Supreme and District Courts of Queensland and the Planning and Environment Court have also recently issued practice directions in relation to the accuracy of references in submissions and the use of generative AI in expert evidence in criminal proceedings.[1]
Uses of AI
Generative AI is being increasingly used in litigation to streamline processes and make preparation more efficient. Generative AI such as ChatGPT and Microsoft Copilot can be used, for example, to:
- generate lists of authorities;
- assist with undertaking disclosure, particularly when dealing with large volumes of documents;
- edit or format documents; and
- identify relevant cases and legislation.
AI hallucinations
One of the main concerns with generative AI in litigation is that generative AI has been known to produce or fabricate case authorities, legislative references and academic resources, also called ‘AI hallucinations’. This year alone, there have been a number of decisions by the Queensland Supreme Court and the Queensland Civil and Administrative Tribunal (QCAT) in which self-represented litigants have been warned against relying on AI hallucinations.[2] In the QCAT decision of LJY v Occupational Therapy Board of Australia [2025] QCAT 96, Deputy President Judge Dann warned that litigants before the Tribunal should understand that including non-existent information in submissions or other material filed in the Tribunal:
- ‘weakens their arguments’;
- ‘raises issues about whether their submission can be considered as accurate and reliable’;
- ‘may cause the Tribunal to be less trusting of other submissions they make’;
- ‘wastes the time for Tribunal members in checking and addressing these hallucinations’; and
- ‘causes a significant waste of public resources’.[3]
Referring to false cases generated by AI in court documents without checking their veracity can also, of course, carry serious consequences for legal practitioners. Conduct of that nature can amount to breaches of the overarching ethical and professional obligations owed by solicitors and barristers to the Courts and to their clients, and the specific duty to not deceive or knowingly or recklessly mislead the Court. In August 2025, a Victorian solicitor lost his ability to practise as a principal lawyer, handle trust money and operate his own practice for submitting fictional AI-generated cases to the Federal Circuit and Family Court of Australia.[4] Although this is the first case where an Australian lawyer has been sanctioned for misusing AI in Court, more instances of lawyers relying on AI hallucinations could come to light.
Recent practice directions and guidelines in Queensland
Practice directions regarding accuracy of references in submissions
To address AI hallucinations and the broader risks of using generative AI as a tool in the conduct of litigation, such as delay and wasted costs, the Chief Justice of the Supreme Court of Queensland and the Chief Judge of the District Court of Queensland have recently issued Supreme Court Practice Direction 5 of 2025, District Court Practice Direction 12 of 2025 and Planning and Environment Court Practice Direction 7 of 2025 (together, Practice Directions). The Practice Directions are in identical terms.
Most notably, the Practice Directions require that, for written submissions to the Supreme, District or Planning and Environment Courts, a person or persons who take responsibility for the contents of the document (the responsible person(s)) must be identified by name at the end of the document.
For written submissions, it is not sufficient for a firm of solicitors on the record to be named; an individual legal practitioner (whether solicitor or barrister) must be named.
For oral submissions, the responsible person is the person making the oral submissions, and that person also becomes a responsible person in relation to any written submissions.
The Practice Directions provide that in circumstances where the responsible person is a solicitor or barrister, the placement of their name on written submissions, or the advancement of oral submissions, informs the Court that the responsible person has performed the following obligations:
- verified the accuracy and relevance of any references to legislation, authorities or other sources; and
- ensured that the document or oral submissions are expressed in terms which reflect their judgment as to the proper discharge of their professional and ethical obligations.
The Practice Directions also specify that self-represented litigants must endeavour to ensure the accuracy of references to any legislation, authorities or other sources referred to in any document prepared by them and relied upon in Court, and in any oral submissions made by them.
If a responsible person gives written or oral submissions to the Court which contain reference to non-existent cases, legislation or other material, the Practice Directions make clear that:
- responsible persons who are legal practitioners may be referred to the Legal Services Commissioner for investigation and/or be required to show cause why a costs order should not be made against them personally; and
- in all cases, this may result in an adjournment of the hearing and potentially an adverse costs order against the party who relied on the document.
In releasing these Practice Directions, Queensland follows New South Wales in its publication of similar, albeit stricter, practice notes relating to the use of generative AI by at the beginning of the year. A summary of those practice notes published by our firm is accessible here.
Practice direction regarding expert evidence in criminal proceedings
Supreme Court of Queensland Practice Direction 14 of 2024 Expert Evidence in Criminal Proceedings (other than sentences) has also recently been amended to address the increasing use of generative AI in the context of expert evidence in criminal proceedings. This practice direction applies to any expert report upon which the prosecution or accused intends to rely at the trial of the accused (including any pre-trial hearing) in the Supreme Court of Queensland, or which the prosecution agrees to obtain on request by the accused or a direction by the Court.
Self-represented litigants have been warned against relying on AI hallucinations
Paragraph 16(l) of the practice direction now explicitly requires that where the expert has used generative AI to assist in the formulation or expression of the opinion(s) contained in his or her report, the report must:
- specify the name of the generative AI program used, and how;
- disclose (as an annexure to the report) a complete record of the inputs (prompts) used for the generative AI program to formulate or express the relevant opinion(s), including any source material, default values and/or variable sets;
- disclose (as an annexure to the report) a complete record of the outputs delivered by the generative AI program in the formulation or expression of the relevant opinion(s);
- specify if the way in which the generative AI program was used is regulated or addressed by any relevant code of practice that binds the expert and, if so, how that code of practice was adhered to by the expert in the formulation or expression of the relevant opinion(s); and
- identify any possible biases or other known limitations that might affect the accuracy or reliability of the opinion(s) formulated or expressed through use of the generative AI program.
Court Guidelines on AI use
The Queensland Courts have also released:
- a guideline on the use of generative AI to assist judicial officers;[5] and
- an updated guideline on the use of generative AI by non-lawyers (including self-represented litigants, McKenzie friends and lay advocates) who represent themselves or others.[6]
These guidelines apply to civil and criminal proceedings in all Queensland Courts and Tribunals, including the Supreme Court, District Court, Planning and Environment Court, Magistrates Court, Land Court, Childrens Court, Industrial Court, Queensland Industrial Relations Commission and QCAT.
In summary, both guidelines have the following key takeaways:
- the guidelines highlight the importance of not sharing private, confidential, suppressed or privileged information with AI platforms, as this carries the risk of waiving privilege, breaching confidentiality or breaching suppression orders;
- advocates have a personal responsibility to ensure the accuracy of the information that is relied on and provided to the Court or Tribunal. The guidelines for judicial officers state that the accuracy of any information they have been provided by a generative AI chatbot must be checked before it is used or relied upon;
- there is the possibility for gaps, errors and biases in AI responses, if the datasets upon which AI platforms have been trained contain those gaps, errors, biases;
- caution should be taken to avoid copyright infringement or plagiarism when uploading text to AI platforms, or when relying on summaries generated by AI platforms; and
- maintain security by finding out and following general cyber-safety best practices. Some specific steps to be taken by judicial officers to maintain security are set out in the guidelines for judicial officers.
The guidelines for judicial officers also:
- make it clear that generative AI can be used by judicial officers and their staff, but that the judicial officer is personally responsible for material produced in his or her name. The judicial officer must ensure that any use of AI tools by the judicial officer or his or her staff is consistent with the core judicial values of open justice, accountability, impartiality and equality before the law, procedural fairness, access to justice and efficiency. The guidelines also state that AI tools should not be used for decision-making or for developing or preparing reasons; and
- contain some more detailed guidance on how judicial officers should be alert to the use of generative AI by legal practitioners, non-lawyers and experts.
What can we expect in the future?
As identified in the Practice Directions, generative AI is rapidly developing, and the approach of the Courts to regulating the responsible use of generative AI must be reviewed regularly. We expect that this will include further or updated practice directions on generative AI, and more frequently updated guidelines for its responsible use.
It remains to be seen whether a more uniform approach to the use of generative AI in Australian Courts and Tribunals will be adopted nationally. At this stage, this seems unlikely in light of the individual practices of each Court and Tribunal. However, it may be something the Courts consider as generative AI becomes more commonly used in litigation and in our own lives.
[1] Supreme Court of Queensland Practice Direction 5 of 2025 – Accuracy of References in Submissions, District Court of Queensland Practice Direction 12 of 2025 – Accuracy of References in Submissions, Planning and Environment Court of Queensland Practice Direction 7 of 2025 – Accuracy of References in Submissions and amendments to Supreme Court of Queensland Practice Direction 14 of 2024 – Expert Evidence in Criminal Proceedings.
[2] See, for example, Khoury v Kooij [2025] QSC 217, LJY v Occupational Therapy Board of Australia [2025] QCAT 96 and Chief Executive, Department of Justice v Wise and Wise Real Estate Pty Ltd & Anor [2025] QCAT 222.
[3] At [26].
[4] See the Victorian Legal Services Board and Commissioner’s statement on this decision at https://lsbc.vic.gov.au/news-updates/news/statement-mr-dayal-matter.
[5] https://www.courts.qld.gov.au/__data/assets/pdf_file/0009/879714/the-use-of-generative-ai-guidelines-for-judicial-officers.pdf.
[6] https://www.courts.qld.gov.au/__data/assets/pdf_file/0012/798375/Artificial-Intelligence_Guidelines-for-Non-Lawyers.pdf, updated on 15 September 2025 from an earlier version published on 13 May 2024.