AI for Law Firms Australia: Privacy, Privilege, Governance

A practical guide to AI for Australian law firms: privacy, confidentiality, privilege, court rules, use cases and a starter roadmap. General information…

AI for law firms in Australia comes down to one idea: you can use it, but your professional duties come with you into every prompt. Confidentiality, competence, honesty to the court and fair costs do not change because a tool wrote the first draft. Firms that get value from AI settle privacy, client confidentiality, legal professional privilege and governance first, and only then choose the tasks where AI helps.

That is the order of this guide: obligations first, then where AI fits, the risks and a starter roadmap. Many pages on this topic are lists of tools. We recommend none, and we are not a law firm.

This is general information, not legal advice. Nothing here is legal, ethical, privacy or professional-conduct advice for your firm. Rules differ between jurisdictions and courts, and they are changing. Everything we state about Australian guidance comes from official pages we read on 1 November 2026, listed under Sources. Your law society, regulator, insurer and own counsel decide what applies to you.

We work from Burleigh Heads on Queensland's Gold Coast, remotely, with no city offices, and we have been building software and AI systems for Australian businesses since 2016.

The short version

Start with duties, not tools. Lawyers must keep meeting their professional obligations when they use AI.
Public AI tools are a confidentiality risk. Official guidance says confidential, sensitive or privileged client information cannot safely go into them.
Privacy law applies. The OAIC says the Privacy Act applies to all uses of AI involving personal information.
Courts have their own rules. Check the practice notes for every court you appear in.
Write it down. A short policy, an approved-tools list, supervision and records beat informal use.

Start with your obligations

The joint statement on AI in Australian legal practice (6 December 2024) says lawyers must comply with the Legal Profession Uniform Law and the conduct rules when they use AI. It highlights four areas:

Confidentiality. Lawyers cannot safely put confidential, sensitive or privileged client information into public AI tools, and a commercial tool's contract terms should be reviewed carefully to make sure information stays secure.
Independent advice. AI tools cannot reason, understand or advise, so lawyers must exercise their own judgement.
Honesty, competence and diligence. No tool built on current large language models is free of hallucinations: fluent, convincing but inaccurate output. Lawyers must be able to verify what their documents say.
Fair, reasonable and proportionate costs. Billing should reflect the work done, and AI should not needlessly add to a client's costs.

The Law Society of NSW's guide for solicitors (January 2026) adds: apply the same care you would to a paralegal, and use AI to supplement legal services, not substitute for them.

None of this is a ban. In its April 2026 submission to the Fair Work Commission, the Law Council of Australia said a broadly framed prohibition on AI use would be neither practical nor proportionate, and listed the risks it sees: breach of confidentiality and waiver of privilege, breach of the implied Harman undertaking, and copyright infringement.

Privacy: what the Privacy Act and the OAIC say

If your firm is covered by the Privacy Act 1988, the Australian Privacy Principles apply to AI as they do to any other handling of personal information. The OAIC's guidance on commercially available AI products makes these points:

Inputs and outputs both count. Privacy obligations apply to personal information put into an AI system and to output that contains personal information, including inaccurate or artificially generated information about an identifiable person, such as a hallucination.
Use and disclosure (APP 6). Personal information put into an AI system must be used or disclosed only for the primary purpose it was collected for, unless there is consent or another permitted basis. The OAIC notes a secondary AI-related purpose may be hard to justify as within reasonable expectations.
Collection (APP 3) and accuracy (APP 10). AI-generated or inferred personal information is a collection, and organisations must take reasonable steps to keep personal information accurate.
Public tools. As best practice, the OAIC recommends not entering personal information, particularly sensitive information, into publicly available generative AI tools.
Ongoing diligence and transparency. The OAIC describes regular review, training and monitoring, privacy by design including a Privacy Impact Assessment, updated privacy policies, and clear identification of any public-facing AI tool such as a chatbot.

One date for the diary. The OAIC announced on 30 September 2026 that from 10 December 2026, APP entities must add certain information to their privacy policy where a computer program makes, or does something substantially and directly related to making, a decision that could reasonably be expected to significantly affect an individual's rights or interests, using their personal information. Whether any tool in your firm meets that test is a question for your privacy adviser, and you cannot answer it without an inventory of what you use.

Confidentiality and legal professional privilege

Privacy, confidentiality and privilege overlap but are not the same thing. The official guidance takes each seriously:

The Law Society of NSW guide says entering client confidential information into a public generative AI system is akin to putting it in the public domain, which is likely a breach of confidentiality, and that clients may lose privilege.
The joint checklist from the Queensland Law Society and six other law societies (February 2026) is more careful: it says the impact of AI on privileged communications is not clear, that AI prompts may not be privileged, and that anonymising information helps but is not a complete answer. It also warns that free or trial services may not provide clear terms, and that the trade-off for a free service is often wide data access.
Material obtained under compulsion is a category of its own. NSW Supreme Court Practice Note SC Gen 23 says suppressed material, material subject to the implied Harman undertaking, and subpoenaed material must not be entered into a Gen AI program unless the practitioner is satisfied of specified conditions, including that it stays in a controlled environment and is not used to train the model.
The Federal Court's practice note warns that entering information into even a ringfenced or confidential tool may breach the obligations attached to that information.

So the question to ask about a tool is not only "is it accurate?" but "where does the information go, who can see it and what can it be used for?" Whether a particular use affects privilege in a particular matter is a legal question for your ethics partner, law society or counsel, not for this guide.

What courts expect

Court rules are moving quickly, so treat this as a pointer to the primary documents, not a summary to rely on.

Federal Court of Australia. The Use of Generative Artificial Intelligence Practice Note (GPN-AI) was issued on 16 April 2026 and applies to everyone who appears before or files documents with the Court. It says, among other things, that users should have a basic understanding of the limits and risks of generative AI; that the responsible person should confirm cited authorities exist and support the proposition, and that facts and chronologies are accurate; that affidavits and witness statements must reflect the witness's own knowledge, with disclosure required in specified circumstances such as where AI was used to summarise or analyse material; and that misuse can lead to consequences including adverse costs orders.

Supreme Court of New South Wales. Practice Note SC Gen 23 (issued 28 January 2025, commenced 3 February 2025) says Gen AI must not be used to generate the content of affidavits, witness statements or character references, which must carry a disclosure to that effect. Citations and references in submissions must be verified, and not solely by an AI tool. Expert reports need leave. It also lists uses it does not prohibit, subject to its confidentiality rules, such as chronologies, indexes and witness lists.

Other courts. The joint checklist lists directions from the Supreme Courts of Queensland, Victoria, South Australia and Western Australia, and says the Northern Territory had none, as at February 2026. The Law Council's AI portal compiles constituent-body guidance and warns that some of it may be out of date, so check the current rules for each court and tribunal you appear in.

Governance that fits a law firm

Governance turns the obligations above into rules staff can follow. The regulators' joint statement suggests lawyers consider:

Clear, risk-based policies on which AI tools the practice uses, who can use them, for what purposes and with what information, how staff are supervised, and how AI-assisted documents are verified.
Limiting AI to lower-risk, easier-to-verify tasks, such as drafting a polite email or suggesting how to structure an argument, and prohibiting it for higher-risk tasks such as translating advice or analysing an unfamiliar legal concept.
Transparency with clients, including recording and disclosing when and how AI was used and how it is reflected in costs, if the client asks.

The joint checklist reduces this to two questions: can you satisfy your confidentiality obligations, and is the output, with your guardrails and human oversight, good enough to put your name to? It describes supervision in two layers: someone personally checking AI output, and the firm's usual supervision making sure those checks happen. Its resource list includes a Queensland Law Society template AI use policy and client warning.

For a national reference point, the Voluntary AI Safety Standard from the Department of Industry, Science and Resources sets out 10 voluntary guardrails, several of which map neatly onto a practice: accountability (1), testing and monitoring (4), meaningful human oversight (5), telling people about AI-enabled interactions (6) and keeping records (9). The Department's Guidance for AI Adoption (21 October 2025) sets out 6 essential practices. Both are voluntary. For the board-level view see AI Governance Australia, and for a formal management system see ISO 42001 in Australia.

Where AI fits: use cases and the questions to ask

The Law Society of NSW guide lists uses lawyers are making of generative AI: client or matter intake, summarising documents, drafting correspondence, legal research, transcribing meetings, drafting and reviewing contracts, and analysing case law. Our table sorts some of them by the questions the official guidance raises. It is a thinking aid, not a ruling that any use is permitted or prohibited.

Use	Questions the guidance raises	Sensible starting posture
Routine correspondence	Is client information entered? Is the tool approved? Is a lawyer reviewing it?	Low risk with approved tools and review
Meeting notes and transcription	Whose personal information is recorded? Where is it stored and processed?	Do the privacy assessment first
Summaries and chronologies	Is the material confidential, suppressed, subpoenaed or Harman-protected?	Only in a controlled environment, after checking the rules
Legal research	Hallucinated authorities. Can you check the sources?	Verify every authority independently
Client intake and chatbots	Privacy notices, accuracy, clear identification as AI	Design in privacy and disclosure up front
Affidavits, witness statements, expert reports	Direct court restrictions and disclosure rules	Check the court's rules first

The regulators' pattern is to begin where information is not confidential and errors are cheap to catch, and move up as your controls mature.

The risks, in plain terms

Hallucinated authorities. The Federal Court notes AI may produce fictitious cases, citations or quotes, and may confirm that wrong information is accurate when asked.
Confidentiality, privilege and compelled material. The mistakes that cannot be undone, because the information has already left your control.
Costs and supervision. AI should not inflate costs, and if checking is rushed or nobody owns it, the control fails.
Vendor terms and shadow use. The checklist asks where data is stored, whether integrations widen access and whether you can terminate and have client data deleted. And you cannot govern tools nobody has listed.

A starter roadmap for a small or mid-sized firm

We give no durations, because the right pace depends on your size, practice areas and how much AI is already in use. Run these in order.

Find out what is already happening. Ask every team which AI tools they use, including personal accounts, extensions and features inside software you already pay for, and mark anything touching client or personal information.
Set an interim rule now. One page: approved tools, what may never be entered (client-identifying, privileged, suppressed or compelled material), and who to ask.
Check your obligations. Ask your law society, ethics partner, counsel and professional indemnity insurer what applies, including court directions. These are questions for professionals, not for us.
Assess privacy and tool terms. Run a privacy impact assessment for each tool you keep and read the terms on data use, training, retention, location, access and termination.
Write a short AI policy covering approved tools, permitted and prohibited uses, supervision, verification, records, client communication and costs.
Pilot one low-risk use, with a defined good output, a named checker and a recorded check.
Train and keep records. Make checking someone's actual job, with time allocated to it.
Tell clients what you do, in engagement terms and matter updates, with your own advisers.
Review on a schedule. Terms, court directions and guidance change, so give someone ownership.

Where Adaptive Media helps

We are a technology and AI consultancy, not a law firm. We give no legal advice and recommend no AI vendors. We do the practical work around the decisions your own advisers make:

AI inventory and risk workshops, mapping tools, uses and where client and personal information flows.
Governance design: a plain-language policy, approved-tools process and oversight rules for your lawyers to review and adopt.
Technical controls and readiness: scoped data access, logging and review steps built into systems, and a sequenced plan, as in our guide to AI readiness assessment in Australia.
Part-time AI leadership: if nobody in the firm can own this, see What Is a Fractional CAIO?, What Is a Chief AI Officer? and Fractional CAIO Gold Coast.

We deliver remotely from Burleigh Heads, Queensland.

Frequently asked questions

Can Australian law firms use generative AI at all?

Nothing in the official pages we read bans it. The regulators' statement expects lawyers to keep meeting their professional obligations when they use AI, and the Law Council has said a broad prohibition would be neither practical nor proportionate.

Can we put client information into a public AI chatbot?

The regulators' statement says lawyers cannot safely enter confidential, sensitive or privileged client information into public AI chatbots or other public tools. The OAIC separately recommends not entering personal information, particularly sensitive information, into publicly available generative AI tools.

Does using AI waive legal professional privilege?

We cannot answer that for any matter, and it is a legal question. The Law Society of NSW says entering confidential client information into a public generative AI system is akin to putting it in the public domain, with a risk that clients lose privilege. The joint checklist says the effect on privileged communications is not clear. Ask your ethics partner, law society or counsel.

Does the Privacy Act apply to our AI use?

The OAIC says the Privacy Act applies to all uses of AI involving personal information. If your firm is covered by the Act, the Australian Privacy Principles apply to personal information you enter and to personal information in the output. Whether your firm is covered is a question for your privacy adviser.

Do we have to tell clients or the court that we used AI?

It depends on the client, matter and court. The regulators' statement recommends transparency with clients, including disclosing when and how AI was used, if they ask. The Federal Court's practice note sets out circumstances requiring disclosure, and the NSW Supreme Court's practice note has its own disclosure requirements. Read the rules for your court.

Can AI draft affidavits or expert reports?

Check the court's rules first. NSW Practice Note SC Gen 23 says Gen AI must not be used to generate affidavit, witness statement or character reference content, and that expert reports need leave. The Federal Court says affidavits and witness statements must reflect the person's own knowledge and requires disclosure in specified circumstances.

Is there an official standard our firm should follow?

For the profession, the primary guidance is from your law society, the regulators' statement and the courts' practice notes. More broadly, the Voluntary AI Safety Standard offers 10 voluntary guardrails and the Guidance for AI Adoption sets out 6 essential practices. Neither is a legal requirement. For a formal management system, see ISO 42001 in Australia.

Sources

Official and professional-body pages read on 1 November 2026:

Next step

List every AI tool your firm uses today, including the informal ones, and mark which touch client or personal information. That inventory is the starting point for your policy, your privacy assessment and your conversations with your law society and advisers. If you would like help turning it into a governance and readiness plan, book an AI governance consult and we will talk through your systems, your workflows and a sensible first step. We do not provide legal advice, and your own advisers remain the authority on your obligations.