AI is already in Australian law firms. Readiness isn't.

AI is already part of Australian legal practice. The organisational question is whether firms can see that use, set appropriate boundaries and prepare the information and people it depends on.

Ask most Australian law firm partners whether their firm is "using AI" and you'll get a cautious answer. Ask their staff, and the answer is often different.

The research published this year tells a consistent story: AI use in Australian legal practice is now common, it's mostly happening through general-purpose tools, and it's running well ahead of the policies, training and systems that would make it safe and genuinely useful.

What the Australian numbers say

The most rigorous local picture comes from the regulator itself. The Victorian Legal Services Board and Commissioner's 2025 Lawyer Census, answered by 1,887 practising Victorian lawyers, found:

• 37% of respondents use AI tools in their legal practice, rising to 46% among those working in law firms.
• 65% of users rely on general-purpose tools such as ChatGPT, Claude, Gemini and Copilot. Legal-specific AI tools are used far less.
• Only 42% said their workplace had AI guidelines.
• 60% had not read either the courts' AI guidelines or the regulators' statement on AI use, including close to half of those already using AI.

Industry surveys point the same way. LexisNexis's 2025–26 Australian Legal AI Survey of more than 1,000 legal professionals found 69% using or planning to use generative AI, with 39% naming inaccurate or fabricated outputs as their biggest concern. Its 2026 Asia-Pacific survey, which includes Australia, put it bluntly in its title: adoption is near universal; governance is not. Across the region, 94% of respondents use AI, but only 58% work under an AI policy.

Use is high. Integration is low.

There's an important distinction hiding in those numbers. Lawyers are using AI, but mostly beside their firm's systems rather than inside them.

LEAP's 2026 research, which included 219 respondents from Australia and New Zealand, found just 16% of Australian respondents use legal-specific AI daily or as part of core workflows, against 49% globally. Australian respondents also reported the highest rate of low or no trust in AI integration of any market surveyed (32%).

The same research points to why. Two in three Australian respondents (68%) work across three or more platforms, and one in three (33%) said their firm has no documented process for capturing knowledge when staff leave. AI can't draw on knowledge that only lives in someone's head, or find the right version of a document stored in four different places.

That's the gap we see most often: the appetite is there, the individual use is there, but the firm underneath isn't yet set up for AI to work across it.

What the regulators actually expect

If you're expecting Australian regulators to tell firms to stop, they aren't.

The Queensland Law Society's Guidance Statement No. 37 says plainly that QLS does not seek to discourage the use of AI tools, and does not see that use as incompatible with ethical duties. What it does expect is use within appropriate guardrails. Among other things, it says:

• firms should make clear that AI may only be used on legal work or firm systems with approval, and set out the conditions that apply to approved tools;
• partners and legal practitioner directors are responsible for the quality of work produced using AI;
• if a tool's user agreement doesn't address how your data is used and whether it trains the model, the tool should be avoided.

The Law Society of NSW's solicitor's guide, updated in January 2026, takes the same line: solicitors remain responsible for their advice, must exercise independent judgement, and should not rely on generative AI to verify sources that AI produced. The courts have added their own layer, including the NSW Supreme Court's Practice Note SC Gen 23 on generative AI in court documents.

The message across all of it is consistent: use AI, but know what's being used, by whom, on what, and under which rules.

What this means for your firm

Put the research and the regulatory guidance side by side and the risk isn't that your firm uses AI. It's that your firm uses AI without seeing it, without rules around it, and without the foundations that would let it do more than individual drafting shortcuts.

Three questions are worth asking at your next partners' meeting:

1. Do we know what AI tools our people are using, and for what? In the Victorian census, most users were reaching for general-purpose tools.
2. Do we have written rules, and has everyone read them? Fewer than half of Victorian respondents had workplace guidelines at all.
3. Could an AI tool find the right version of our documents, and our people's know-how, today? If the honest answer is no, that's where the real work starts.

None of those questions is a reason to wait. Each one is something a firm can begin this month.