By Jehan-Philippe Wood and Johnathan Edgelow
Courts and clients have not just acknowledged legal practitioners’ use of Artificial Intelligence (AI); they are demanding that they make more use of AI, resulting in an increase in risk factors for solicitors and barristers.
As recent cases show, overconfidence, and failure to supervise, verify AI generated outputs and apply professional judgement when using AI, are key risk factors for lawyers. Australian courts and regulators are prepared to sanction lawyers for misuse of AI, and professional negligence claims by clients present a significant risk.
A range of AI technology exists for use by legal practitioners, of varying sophistication, suitability and cost. There are simple (low cost) frameworks that practitioners can adopt to use AI tools effectively, safely and responsibly.
This alert addresses the following issues:
- Use of AI by Australian legal practitioners
- Legal and regulatory frameworks
- Key risk factors for legal practitioners
- Using AI safely and responsibly
AI use by Australian legal practitioners
The legal profession continues with rapid adoption of generative AI. A February 2026 LexisNexis study found that 69% of lawyers were using or planning to use generative AI, up from the year before, while Clio’s 2026 Legal Insights Report suggested adoption of AI among Australian legal professionals may be as high as 98%.
Confidence in AI is also increasing. Smaller firms appear more likely to use publicly available tools such as ChatGPT, with 35% of respondents reporting reliance on public AI platforms and 42% investing in more bespoke and domain specific AI tools.
AI appears to be most commonly used for document review, legal research, summarisation, drafting briefs and memoranda, correspondence, due diligence, and understanding legal concepts.
Despite this growth, adoption of legal-specific AI tools in Australia remains relatively low compared to the rest of the world. LEAP Legal Software’s Profitability in Law: Global Report 2026 found that only 16% of Australian legal professionals use legal-specific AI daily or as part of their core workflows, compared with 49% globally.
The impetus to adopt AI tools to improve efficiency, lower costs and innovate is coming from clients and courts. Justice Michael Lee of the Federal Court recently ordered parties in the McDonalds Breaks Class Action to confer about the suitability of using AI to prepare for mediation and trial (Rogers v McDonald’s Australia Ltd (AI-use) [2026] FCA 1264).
Legal and regulatory frameworks
Federal and State Courts, and administrative tribunals, such as the Administrative Review Tribunal and Fair Work Commission, have issued practice directions and guidance to govern lawyers’ use of AI in legal proceedings, potentially exposing lawyers who do not comply to sanctions.
Courts have sanctioned lawyers for misusing AI (see ‘Key risk factors’ below).
Professional and ethical standards apply to solicitors’ use of AI and present legal, ethical and professional risks for solicitors. The same applies to barristers.
Key areas of concern under these standards include:
- maintaining client confidentiality and privilege;
- delivering legal services competently and diligently;
- discharging duties to the court (e.g. not to mislead); and
- charging costs that are fair, reasonable and proportionate.
Key risk factors for legal practitioners
The risks associated with AI use are (or should be) well known but are nonetheless catching out legal practitioners.
They include:
- false or inaccurate outputs including ‘hallucinations’;
- unwanted bias;
- breach of confidentiality;
- overreliance on AI;
- inappropriate use of AI; and
- outsourcing critical thinking and legal reasoning to AI.
Even the most sophisticated Large Language Models can produce false outputs.
And as AI use grows, the number of adverse court decisions is growing. Recent cases against legal practitioners related to AI misuse have resulted in sanctions against legal practitioners (see ‘Case studies’ below), including:
- referrals to professional conduct authorities;
- adverse costs orders; and
- judicial warnings or reprimands.
Courts also have powers to sanction practitioners and their clients by striking out cases, initiating contempt proceedings and referring matters to the police. Further, the Legal Services and Complaints Committee of the Legal Practice Board of Western Australia has powers to investigate complaints against legal practitioners and take disciplinary action against them or prosecute serious professional misconduct before the State Administrative Tribunal.
The recent decision in Dey v Minister for Immigration and Citizenship (No 2) [2026] FedCFamC2G 1799 highlights the importance of carefully reviewing and independently verifying AI generated content before relying on it in legal practice. In that decision, the court ordered a lawyer to personally pay $9,000 in costs arising from their ‘unacceptable’ use of AI.
The case reinforces that lawyers remain personally responsible for the work they produce and cannot avoid professional or disciplinary consequences by attributing errors to AI tools.
Using AI safely and responsibly
If you have not done so already, take some time to familiarise yourself with what AI tools may be available to enhance your service delivery to your clients.
At the same time, remember that using AI is not risk free and it is advisable to put in place measures to safeguard yourself and your firm from AI exposures.
How legal practitioners use AI and the associated risks are evolving fast. While exploring new tools and ways of working, it is sensible to ask yourself on a regular basis whether you have taken the necessary steps to mitigate your AI related exposures.
Those steps include:
- establishing and implementing clear, risk-based policies, procedures and workflows around the use of AI;
- identifying authorised and unauthorised use of AI;
- minimising data and security breaches;
- training and educating staff on how to use AI tools safely and responsibly (including protecting inputs and validating outputs);
- correcting AI generated mistakes and explaining how they occurred and will be avoided;
- ensuring appropriate transparency about AI use and obtaining appropriate consents from stakeholders (including the court, clients and other practitioners); and
- appropriately charging for the costs of using AI.
Solicitors seeking practical guidance on minimising AI hallucinations and ensuring that AI-generated outputs are grounded in the evidence should review the Fair Work Commission’s set-up prompt in Using AI to help you with your case for guidance.
Case studies
Two recent case studies illustrate the potential consequences of AI generated mistakes for solicitors and barristers.
| Case study 1 |
| What happened? A junior solicitor in a reputable firm used AI to carry out legal research to prepare a block transfer application under the UK Insolvency Rules. The application (to be determined on the papers) sought a court order releasing the liquidator from liability. The judge wrote to the firm querying whether the court had the power to grant the release. The junior solicitor used AI to prepare a written submission to the effect that Rule 12.37(e) gave the court the necessary power. The judge checked the Insolvency Rules and confirmed that Rule 12.37(e) did not give the court that power. The judge asked the firm to explain. The junior solicitor used AI to draft a further letter to the court explaining that they did not mean to quote a specific provision of the Insolvency Rules, but the Insolvency Rules collectively gave the court the power. Sensing something was awry, the judge summoned the firm to explain the error. The supervising partner and senior associate, and the firm’s General Counsel and Chair of the Innovation and Change Board, tendered witness statements explaining that the junior solicitor had used AI to make the application and draft the correspondence and had not been properly supervised, and the firm was in the process of rolling out an AI pilot and putting safeguards in place to manage the risk of future errors. What was the key failing of the AI? The AI was “wrong or…extremely misleading” about the content of the Insolvency Rules. It did not provide the proper text or effect of the rules, even when asked directly for it. What were the key failings of the junior solicitor? 1. They should have been aware of the dangers of using AI. 2. They should have consulted an authoritative source for the wording of the Insolvency Rules. 3. In asking the AI to write the letters to the court, they “outsourced the thinking process to the AI”. 4. They ignored the AI’s warnings to consult the actual text of the Insolvency Rules and check that the explanation to the court was accurate. 5. They allowed an untrue statement to be made to the court as part of an ex post facto attempt to explain away earlier mistakes, prompting a concern that they had tried to cover up their earlier mistakes (in this case, the judge concluded they were negligent rather than dishonest). 6. They should have been candid with the senior solicitors about the court’s correspondence and their use of AI. What were the key failings of the senior solicitors (partner and senior associate)? 1. They should have been aware of the dangers of using AI. 2. They failed to supervise the junior solicitor properly. 3. They should have checked the junior solicitor’s work. 4. The senior associate made assumptions about what the junior solicitor had done – instead, she should have scrutinised their work and discussed it with them (they appear to have communicated via Teams only) and should have escalated the issues to the partner. 5. The junior solicitor was not candid with their supervisors – but the senior solicitors should still have checked their work. What were the consequences for the solicitors? 1. The solicitors lost the matter (and the client) to another firm. 2. They had to pay the alternative solicitors’ costs and the costs of correcting their mistakes. 3. They were publicly admonished by the court. 4. They self-referred to the Solicitors Regulatory Authority (the UK equivalent of the Legal Practice Board of Western Australia). 5. The court considered it appropriate that the SRA investigate the conduct of the junior solicitor. Case citation: Cork v Smith [2026] EWHC 1199 (Ch) |
| Case study 2 |
| What happened? The Full Court of the Federal Circuit and Family Court of Australia discovered that the appellant’s Summary of Argument and List of Authorities contained “non existent, inaccurate and misleading authorities”. The Court considered that the documents appeared to have been prepared with artificial intelligence and contained “AI-generated hallucinations”. After the issue had been identified, an amended Summary of Argument and List of Authorities was filed, accompanied by a letter that apologised for “significant errors in the citations” and the “oversight”. However, the Court remained concerned as the extent to which AI had been used, and the cause of all the errors, had not been fully explained. Ms G, the solicitor for the appellant, stated that she had not personally used AI and that the errors arose from her paralegal’s use of AI without her knowledge. Notwithstanding this, Ms G accepted responsibility for her paralegal’s conduct and for the documents filed on behalf of the appellant. Counsel for the appellant, being the junior and senior barristers who appeared in the appeal, denied personally using AI in the preparation of the documents. However, the Court found that both counsel had settled the Summary of Argument and List of Authorities and were responsible for the content of the documents filed for the appellant, including their inaccuracies. What was the key failing of the AI? The primary failure was that the AI generated content containing “inaccurate and misleading references to case law (hallucinations).” What were the key failings of the legal practitioners? 1. Failure to verify AI-generated authorities. 2. Failure to properly review and verify the content of documents submitted to the Court. 3. Failure to appreciate that use of AI (personally or by someone else) does not diminish professional responsibility. 4. Lack of transparency regarding the use of AI, particularly as, after the errors were corrected, the reason for the amendments was left “entirely unexplained”. 5. Failure to adequately explain ongoing errors. What were the consequences for the legal practitioners? 1. The solicitor was referred to the South Australian Legal Profession Conduct Commissioner. 2. Counsel were referred to the Victorian Legal Services Board and Commissioner. 3. By consent, the solicitor agreed to pay the respondent $10,000 for costs thrown away in correcting the AI-generated errors. 4. The Court published detailed reasons discussing the practitioners’ conduct and emphasised the professional and ethical obligations of lawyers who use AI. Case citation: Mertz & Mertz (No 3) [2025] FedCFamC1A 222 |
Key messages
Using AI appropriately has the potential to enhance your practice. However, misusing AI can lead to adverse consequences.
Familiarising yourself with the measures that will enable you to use AI safely and responsibly and regularly reviewing your use of AI will assist you in reaping the benefits of AI while mitigating the risks and consequences of AI mistakes.
Look out for details of a CPD seminar on this subject with more practical insights coming soon.
The authors wish to thank Bijoux Arguijo and Alex Winslade-Mullane for their contributions to this article.
Further reading
Federal Court of Australia, General Practice Note – Use of Generative Artificial Intelligence Practice Note (GPN-AI), 16 April 2026
Supreme Court of Western Australia, Guidelines for the use of generative AI, 20 November 2025 (now part of Practice Direction 9.21)
Administrative Review Tribunal (Generative AI) Practice Direction, 20 August 2026
Legal Profession Uniform Law and the Legal Profession Uniform Law Australian Solicitors’ Conduct Rules 2015
Legal Profession Uniform Conduct (Barristers) Rules 2015
Rogers v McDonald’s Australia Ltd (AI-use) [2026] FCA 1264
Cork v Smith [2026] EWHC 1199 (Ch)
Mertz & Mertz (No 3) [2025] FedCFamC1A 222
Dey v Minister for Immigration and Citizenship (No 2) [2026] FedCFamC2G 1799
Solicitors Regulation Authority Ltd v Kumar (Case No. 12884-2026)
Law Society of Western Australia website, Risk management for legal practices (which has questions for law practices to consider and links to other useful resources)
Fair Work Commission Practice Note: Using AI to help you with your case.
