AI and ethics · 6 November 2026
What should a law firm ask before client facts go into a tool?
Don't put confidential client information into a public chatbot. Don't treat the tool's output as your advice. You are still the person who checks the document. Those lines are from the regulator. Under them are the questions I'd want in writing from any vendor, including me.
What the VLSB+C actually said
In December 2024 the Victorian Legal Services Board and Commissioner put out a joint statement with the Law Society of NSW and the Legal Practice Board of WA. For intake and first drafts, this is the bit that matters.
- You can't safely put confidential, sensitive or privileged client information into public AI chatbots or copilots like ChatGPT. If you use a commercial tool with client information, read the contract so you know it stays secure.
- You can't treat the output as a substitute for your own assessment. If AI helps prepare the document, you have to be able to check what's in it, and you actually have to check it.
- Costs stay fair and proportionate. Don't bill time you didn't spend, and don't let the tool make the matter dearer than doing it the old way.
- Be straight with the client. Note when and how AI was used, and tell them if they ask, including how it shows up in the costs.
The LIV guidelines add a practical point: don't put confidential or privileged information into an open or commercial AI system without the client's consent, tell them you used AI, and check every output.
In August 2025 the Board varied a Victorian practitioner's certificate after a court found inaccurate AI-generated citations and summaries had been tendered. For a wills or debt practice, the lesson isn't the citations. The lawyer stayed responsible for the document.
What I'd ask the vendor
- Does any client fact go into a public model, or into the vendor's training set?
- Who can open the matter? Is the login lawyer-only, and what happens to the data if we stop?
- Is the first draft on our precedent, and does a lawyer approve it before a client or a debtor sees it?
- What do we tell the client, in the costs agreement or the engagement letter?
- Where does the card payment land? For work not yet done, that conversation is with your trust accountant. The longer note is taking payment upfront for a fixed-fee will.
On Firmbrief, the client answers on your website and pays your fee. You review the draft. Nothing goes out until you approve it. Where it's hosted, and the contract terms, are a call. I won't invent them on this page. The privacy page for this website is here. It says some processors are in the United States. It does not cover the intake we set up for a firm. That sits in the agreement with you.
The longer note on the December 2024 statement and the LIV guidelines is can Victorian lawyers use AI to draft wills and letters of demand?
Questions on this
Can a Victorian lawyer put client details into ChatGPT?
Not safely. The VLSB+C's December 2024 statement says lawyers cannot safely enter confidential, sensitive or privileged client information into public AI chatbots or copilots like ChatGPT.
Who is responsible if an AI-assisted draft is wrong?
The lawyer. The VLSB+C says you cannot rely on AI output as a substitute for your own assessment, and that a lawyer using AI to prepare documents must be able to verify what they contain. In 2025 the Board varied a Victorian practitioner's certificate after inaccurate AI-generated citations were tendered.
Do I have to tell the client I used AI?
The LIV's guidelines say to disclose the use of AI to the client. The VLSB+C statement says to be transparent, and to record and disclose when and how AI was used if the client asks, including how it shows up in costs.
Sources: VLSB+C, Statement on the use of AI in Australian legal practice (December 2024); LIV, Ethical and Responsible Use of Artificial Intelligence; VLSB+C, statement on the Mr Dayal matter (2025); LPLC, Supervising AI.