Australia: professional conduct rules and AI
By the ChamberOne team 5 minute read
Australian solicitors do not need a new rulebook to use AI responsibly. They already work under conduct rules that deal with competence, confidentiality, honesty to the court and supervision. The practical question is how those existing duties apply when a drafting or research tool is involved. This article sets out how to think about it. It is general information, not advice, and you should read the guidance from your own regulator.
Key points
- Existing duties of competence, confidentiality, candour and supervision apply to AI use.
- Regulation is by state and territory, so read the guidance from your own regulator.
- Verify every authority and quotation at source, and check each court's practice direction before filing.
- Adopt a short firm policy on approved tools, permitted data and checking.
Regulation is local, so start with your own regulator
Legal practice in Australia is regulated at state and territory level, with professional bodies and regulators in each jurisdiction. The details of rules, guidance and enforcement can differ depending on where you hold your practising certificate. A partner in one state may find that a colleague's guidance in another is helpful but not binding.
So the first step for any practitioner is simple. Find the current guidance on AI from the regulator or law society that governs you, read it in full, and note what it says. Guidance has been developing, and what applies today may be refined. Do not rely on a summary on a vendor's website, or on this article, as a substitute.
Competence and diligence
The principle that a lawyer delivers legal services competently and diligently does not change when a tool is used. If an AI system produces a draft or an answer, the lawyer who adopts it is responsible for it. The tool is not a defence if the output is wrong.
In practice this means understanding what a tool does well, what it does badly and how it can fail. The best-known failure is the confident but fabricated answer, such as an authority that does not exist or a quotation that was never said. Everything that goes to a client or a court needs checking against primary sources.
- Learn what the tool is for. Use it for first drafts, summaries and structure, not as a final authority.
- Verify every case, statute and quotation at source before it leaves the office.
- Keep a record of how you checked, especially for work that will be filed.
- Do not use a tool you do not understand on a matter where mistakes would be costly.
- Learn what the tool is for
- Verify every case, statute and quotation at source
- Keep a record of how you checked
- Treat the tool as a final authority
- Use a tool you do not understand where mistakes would be costly
- Send or file anything before it is verified
Confidentiality and privilege
Client information is confidential, and communications may be privileged. Putting client material into a third-party AI service raises questions that you should be able to answer before you do it. Where does the text go? Who can read it? Is it kept? Is it used to train a model? Under what terms?
Consumer-grade tools are the highest-risk route, because their terms may allow review and reuse of what you enter. A tool with business terms, a clear data processing agreement and a no-training commitment is a better starting point, though you still have to read the terms. Some clients, especially government and large corporate clients, set their own rules on AI use in their engagement terms. Check them first.
Think also about what is being sent. Removing names and identifying details, or using only the parts of a document that the task needs, reduces the exposure. It does not replace a proper review of the tool.
Candour to the court
Practitioners owe duties to the court, including not misleading it. Filing material that contains invented authorities or misstated law is a serious matter whether it came from a junior, a tool or the practitioner's own error. Courts in a number of places have issued their own practice notes or directions on AI use. These can include rules about disclosure of AI use or limits on the use of AI to prepare certain documents, and they differ between courts.
Before you file in any court or tribunal, check whether it has a practice direction on generative AI and read it. Do not assume a direction from one court applies to another.
Supervision and firm systems
Principals and supervising practitioners are responsible for the work of those they supervise, and for having appropriate systems in the practice. If juniors or support staff are using AI tools, the firm needs a policy. A workable policy is short, and covers the following points.
- 1
Which tools are approved, and which are not allowed for client work.
- 2
What categories of information may and may not be entered.
- 3
What checking is needed before AI-assisted work is used or sent out.
- 4
Who to ask when staff are unsure.
- 5
How and when clients are told that AI has been used, where that is appropriate.
- 6
A review date for the policy, since both the tools and the guidance move quickly.
Costs, clients and honesty
Fee disclosure and billing duties continue to apply. If AI reduces the time a task takes, think carefully about how that is reflected in the bill, and how time billed is described. Billing a client for hours that the work did not take is a problem however it is dressed up. Your regulator's guidance on costs applies here, and so do any costs agreements with the client.
Where AI use is significant to the way you deliver a matter, consider telling the client at the outset, and recording their instructions. Clients generally prefer a firm that is open about its methods to one that is vague.
Practical steps for the next month
- Find and read your regulator's current AI guidance, and any practice direction from the courts where you appear.
- Make a list of the AI tools anyone in the firm is using, official or not.
- Review the terms for each against the questions on training, retention and location.
- Adopt a one-page policy and brief everyone.
- Set a standard for checking: nothing is filed or sent without verification at source.
- Review the position every six months.
Treat the above as a prompt for your own thinking, not as a statement of what the rules in your jurisdiction require. When in doubt, ask your regulator's ethics service or an experienced colleague.
This article is general information about the law, legal practice and technology, current when it was published. It is not legal advice. Laws and guidance change: check the sources linked above and take advice on your own circumstances from a lawyer qualified where you are.
Read next
Generative AI guidance for lawyers: how six markets compare
How to find what regulators, courts and bar bodies have said in six markets, and the common themes across them.
5 minute readAustralia: court practice notes on generative AI and what they require
Courts in Australia have issued practice notes on generative AI. Read the one for your court, every time.
4 minute readEU AI Act: what it means for firms outside the EU with EU clients
How a firm outside the EU should think about the EU AI Act when it works for European clients.
5 minute read