AI in the Law: The Good, the Bad and the Ugly

The Good: AI’s efficiencies

Almost every Australian lawyer uses some form of AI, and for good reason. AI tools produce real efficiencies, freeing up time previously spent on repetitive and low-value tasks and enabling solicitors to focus on the analysis and judgement that clients come to legal practitioners for.

But what happens when these tools are used by solicitors who don’t exercise adequate oversight? This article sets out how solicitors are expected to use AI in their legal practicI can not find thee to ensure they comply with their obligations under the Legal Profession Uniform Law Application Act 2014 (Uniform Law) and the Australian Solicitors’ Conduct Rules 2015 (ASCR).

The Bad: risks the regulators have identified

In December 2024, the Law Society of NSW, the Legal Practice Board of Western Australia and the Victorian Legal Services Board + Commissioner issued a joint Statement on the Use of Artificial Intelligence in Australian Legal Practice (Statement).¹ According to the Statement, professional obligations of legal practitioners apply in full where AI is used in legal practice. Risks of AI use were identified with respect to four existing solicitor obligations:

1. Confidentiality (ASCR r 9.1): solicitors cannot safely enter confidential, sensitive or privileged client information into public AI tools. When commercial tools are used, their contractual terms must be reviewed to confirm information will be kept secure.
2. Independent advice (ASCR r 4.1.4): AI tools “cannot reason, understand, or advise”. Solicitors must exercise their own analysis and judgement.
3. Competence and diligence (ASCR rr 4.1.2 – 4.1.3): no tool built on the current large language models is without hallucinations. Solicitors must be able to verify AI-prepared content personally to make sure it is accurate.
4. Fair costs (Uniform Law ss 172–173; ASCR r 12.2): billing must reflect the legal work done by law practice staff. AI must not unnecessarily increase costs, for example through time spent verifying or correcting AI outputs.

Solicitors are also expected to adopt risk-based AI policies, supervise junior and support staff, limit AI to lower-risk tasks that are easier to verify, and be open with clients about how AI is used.

The Statement is in response to the increasing misuse of AI that courts are tackling. A clear example of such misconduct is Dayal [2024] FedCFamC2F 1166.²

The Dayal matter was heard in the Federal Circuit and Family Court and related to the conduct of a Victorian solicitor in a family law matter where he had tendered authorities that neither the judge nor her associates could find. The solicitor admitted that the list had been prepared by an AI based legal software and he had not verified the output. Her Honour stressed that the duty of legal practitioners not to mislead the court and the duty of candour fully apply when AI is used. On 19 August 2025, the solicitor’s practising certificate was varied by the Victorian Legal Services Board + Commissioner. This resulted in the solicitor no longer being entitled to practise as a principal or handle trust money and the requirement that he undertake two years of supervised practice.³ This regulatory action was the first sanction of its kind in Australia.

The Ugly: how to comply

There are several things that solicitors can implement in their daily practice to comply with their obligations under the ASCR and the expectations set out in the Statement.

First, be upfront with how AI may be used and how that is reflected in costs. Prioritise client confidentiality by using enterprise-grade tools for client information and check the provider’s terms on data retention and use before uploading anything. As part of our transparency efforts, our engagement terms include a dedicated AI clause confirming that all work is checked by a qualified practitioner.

Accordingly, solicitors should treat anything AI produces as a first draft prepared by the most junior person in the room. This means checking every authority, legislative reference and quote to confirm it says what it is cited for. Supervision needs to be built into the system.

The tool must also be matched to the task. AI should be used for lower-risk, easily identifiable work and not for translating advice, analysing unfamiliar legal concepts or decision making. Staff should accordingly be trained on what AI does well and what it doesn’t. Additionally, putting a written policy in place can identify which tools are approved, who may use then, for what purposes and with what information and how AI generated content is checked for accuracy before finalisation, as discussed above.

Finally, check the forum’s rules. Courts and tribunals now have their own AI practice notes, so solicitors should check the rules of the relevant forum before filing anything.

AI changes the tools, not who is accountable. Checking AI output is a personal obligation that cannot be delegated so ensure you are observing your obligations as a practitioner.

‍

^{1. Law Society of New South Wales, Legal Practice Board of Western Australia and Victorian Legal Services Board + Commissioner, Statement on the Use of Artificial Intelligence in Australian Legal Practice (December 2024).}

^{2. Dayal [2024] FedCFamC2F 1166.}

^{3. Victorian Legal Services Board + Commissioner, Statement on Mr Dayal matter (19 August 2025).}