Read enough legal-tech commentary and a pattern emerges. Australia's AI adoption lag gets framed, almost without exception, as an education problem. More CPD, cleaner demonstrations, better explanations of what a large language model does, and the gap will close. It is a tidy hypothesis, and the legal-tech community has been testing it for two years.
This week, Lawyers Weekly published a piece built around new LEAP data and InfoTrack's read on the Australian market. The numbers are familiar by now: sixteen percent of Australian firms use legal-specific AI on a daily basis, against a global figure of 49 percent. InfoTrack's Chief Product Officer, Ajay Kumar, pointed to education and change management as the levers. He also said something that tends to get overlooked: global enterprise AI tools are not always built to reflect the way Australian lawyers and conveyancers work on local matters.
We think that second observation is where the analysis should start, not where it should end.
The distrust number is worth sitting with. Australia is not a market where lawyers are unfamiliar with technology, resistant to efficiency, or disconnected from global practice. The country's large commercial firms run sophisticated operations. In-house teams face the same cost and speed pressures as their counterparts in London or New York. And yet, on legal AI specifically, Australian practitioners distrust the tools more than any other region on earth.
One explanation is that they have tried the tools and found them wanting. A lawyer who puts a question about Australian consumer law to a US-trained AI, checks the output, and finds it citing inapplicable US doctrine, or hallucinating a provision that doesn't exist in Australian legislation, learns something important about the tool's calibration. A second experience like that, and distrust is a professional judgment grounded in evidence. Framing this as a knowledge deficit inverts the problem. Lawyers who distrust tools that produce unreliable outputs on Australian law are exercising exactly the critical evaluation that responsible AI adoption requires. More training on a tool that doesn't fit will not make the tool fit.
That distrust also manifests in a specific way that adoption statistics don't fully capture. Many lawyers run a shadow workflow: use the AI output as rough orientation, then rebuild the research from primary sources they trust. The AI was used, technically, but it added a step rather than removing one, because the verification burden was transferred to the practitioner rather than resolved by the tool. That pattern generates extra work; it is not AI adoption in any meaningful sense.
What "fit" means for Australian legal work is worth spelling out, because it is more specific than the general observation about "local matters" might suggest.
Australian law runs on a distinct court hierarchy. The High Court's authority over constitutional and common law questions, the interplay between the Federal Court and state Supreme Courts, the procedural rules specific to each, these are not variations on a global theme. They are the thing itself. A research tool that doesn't understand which authority binds which court, or how a 2019 Full Federal Court decision sits relative to a 2022 state Court of Appeal decision in the same area, produces answers that are plausible-sounding and jurisdictionally wrong.
The binding versus persuasive distinction is a specific failure mode within that. A tool that surfaces a UK Supreme Court decision, a New Zealand Court of Appeal authority, or a Privy Council case on a statutory interpretation question without flagging its status in Australian courts creates a particular kind of problem: the output looks authoritative, the case is real, and the error is invisible to anyone who doesn't already know the answer. A practitioner relying on it to build an argument, rather than checking it, would not know what she had missed until the other side pointed it out.
Add the legislative patchwork. Australian consumer, employment, property, and financial-services law is a mosaic of federal and state instruments that interact in ways no global-enterprise AI trained primarily on US or UK materials is calibrated to handle. The National Consumer Credit Protection Act, the Fair Work Act, the various property and conveyancing Acts across jurisdictions, each carries its own legislative history, its own body of interpretive case law, and its own reform trajectory. Reform trajectories compound the problem: Australian consumer credit regulation and employment law have both moved significantly in recent years, and a model calibrated at a fixed point, drawing on a corpus that hasn't incorporated recent amending legislation and the interpretive decisions that follow it, produces answers that were once correct and are no longer. Getting answers that draw on Australian primary sources, and that show which source supports which proposition, is a precondition for trusting the output.
Conveyancing sharpens the point. The process varies meaningfully between states and territories, a fact Australian practitioners live with daily and that global tools handle badly or not at all. Kumar's observation that these tools are not always built to reflect how Australian lawyers and conveyancers work on local matters is precise. For a practitioner who needs to trust the output on a specific transaction in a specific jurisdiction, "not always" is not a reassuring qualifier.
The education-and-change-management framing is not entirely wrong. There are genuine knowledge gaps about AI capabilities, and they matter. A lawyer who doesn't understand what a retrieval-augmented generation system does differently from a base language model will have trouble evaluating which tool is appropriate for which task. That understanding is worth building.
The error is treating education as the primary lever when the tool itself is the mismatch. You can run as many CPD sessions as you like on the strengths and limitations of AI research tools. If the tool's corpus is global, its reasoning calibrated to US or UK legal culture, and its citations unverifiable, practitioners will distrust it, and they will be right to. The Australian adoption gap will not close through better-informed lawyers using a tool that was not built for Australian law. It will close when the tools are built for it.
This is the argument for an Australian-first approach, and it is a product argument. Habeas was built specifically for Australian legal research: a search engine that scans over 300,000 Australian cases and pieces of legislation in seconds, with results grounded in a closed dataset of legitimate Australian legal sources, so they are verifiable and traceable rather than plausible-sounding confections drawn from somewhere else in the common-law world. The traceable citation is the answer to the precise failure mode that generated Australia's distrust figures in the first place.
The research time savings follow directly from that trust. Foundational research that used to take a full morning can now be completed in minutes, because a practitioner can trust the output enough to build on it, rather than spending the morning verifying it against sources she has to locate herself.
Lisa, a General Counsel at a Series A startup, put it this way: "It feels like having a law firm in your pocket. Not something you blindly bet the house on, but a powerful first-line legal intelligence tool." A tool grounded in Australian primary sources, with traceable citations, gives the practitioner something worth bringing judgment to bear on. That is a different proposition from a generic AI that requires the practitioner to audit its work before she can begin hers.
Kumar is right that education has a role. Lawyers who understand what these tools can and cannot do will make better decisions about where to use them and where not to. The sequencing matters, though. Education about AI should accompany tools that work for Australian practice, not precede them as preparation for tools that don't.
The LEAP data is telling us something. A region that leads the world in AI distrust, despite having the legal capability and commercial pressure to adopt, is a region where the tools have not yet earned trust. Closing that gap requires building tools that deserve it.
That is where we are focused.
Habeas is a legal AI research platform built for Australian law. You can book a demo at habeas.ai.
If you want to try for yourself or get in contact, book a demo with us here. We also offer the capacity for self-serve individuals to sign up, and subscribe or register a free trial at app.habeas.ai.
The legal research in this article was conducted and every citation verified using Habeas, the Australian legal AI research platform.
Hero image: Macourt Media on Pexels
