AI note takers for Australian law firms: what the law societies, insurers and courts say

What each body has said

The Queensland Law Society article on note takers (July 2026)

The Proctor article of 9 July 2026, written by David Bowles, is worth reading in full because it is about exactly this product category. It starts from the reach of these tools. A transcription or telephony tool hears the privileged advice, the confidential settlement figure and the admission the client should not have made. In the article's words, 'Nothing in a practice has broader reach into protected material than a service that listens in on every call.'

It then sets out what to examine in the provider's terms:

• What is done with the data. Is the audio or the transcript used to train models or to 'improve the service'? The article warns that 'We do not train on confidential data' may rest on the idea that stripping names removes confidentiality, and that a no-training promise means little beside 'We may share data with trusted partners'.
• Who it is shared with. Many providers run on someone else's AI model. The article says that is not a problem in itself. It comes down to the agreements between those companies.
• Where it is kept or processed. Storage in Australia avoids the cross-border accountability issue under s 16C of the Privacy Act 1988 (Cth). Storage offshore engages it.
• Security in concrete terms. A recognised certification such as ISO/IEC 27001 or SOC 2 reflects independent assessment. 'Hosted on AWS' does not, because the host only supplies the platform.
• The contract. The website FAQ is not the agreement. Where marketing and contract differ, the contract governs.

Its conclusion is permissive. 'There is nothing wrong with using AI to take the drudgery out of file notes and time recording.' The risk is treating the job as low stakes because it feels administrative. It also warns that a tool integrating with your practice management system does not mean anyone has done the due diligence.

The joint law society checklist (February 2026)

QLS launched the AI Selection and Use Checklist for Legal Practitioners on 23 February 2026, developed with the Law Institute of Victoria and the law societies of South Australia, Western Australia, the ACT, the Northern Territory and Tasmania. It is aimed at small and mid-sized firms buying a tool someone else built. It asks two main questions. Can you meet your confidentiality obligations to clients? And will the work product be good enough to put your name to?

Five points in it bear directly on a note taker:

• Privacy impact assessment. The checklist says one must be conducted before deploying the system. It describes that as a statutory obligation for entities regulated by the Privacy Act and a professional duty for others.
• Free and trial tiers. Free or trial services may not have clear terms, and the trade-off for a free service is often wide data access. Check which version of the terms applies to your subscription tier.
• Offshore processing. A footnote says it is easier to avoid the issue of international transfers by choosing a vendor that processes all data in Australia. Otherwise you assess the destination under Australian Privacy Principle 8 and r 9 of the Australian Solicitors' Conduct Rules (ASCR).
• Client consent. 'In most cases clients should be informed of AI data sharing and given access to any relevant documentation.' Annexure B gives sample disclosure and opt-out wording for an engagement letter.
• Privilege. The checklist says the impact of using AI to process privileged communications 'is not clear', and that using such material to train an AI system is arguably incompatible with an intention to keep it confidential. We cover that in does an AI note taker put privilege at risk? and are attendance notes privileged?

What the insurers say

We did not find a Lawcover or LPLC publication about AI note takers specifically. What both have published applies to them.

Lawcover (NSW). The March 2026 edition of Lawcovernotes announced a Sample AI Policy for insured practices. Its commentary says the person approving AI tools should work from a risk assessment covering data privacy and confidentiality, accuracy and reliability of outputs, potential bias, regulatory compliance, professional liability, impact on clients, and the auditability of the tool and its processes. It also tells firms to consider warning clients that putting their own information or their communications with you into an AI tool may unwittingly waive privilege.

LPLC (Victoria). LPLC's article 'Managing the risks of AI in law practices' (last updated 10 June 2026) says an AI usage policy should define which tools are permitted and for what, require protection of client data, and 'mandate the vetting of third-party vendors'. It also warns about staff using unauthorised tools, which it calls shadow IT. For a note taker that means one approved product, switched on by the firm, with free personal accounts on staff phones and laptops ruled out in writing.

Court practice notes

The court documents govern material used in proceedings. They do not regulate an internal file note. They matter in two situations: when a note later supports an affidavit, and when a call discusses material that came to you under compulsion.

• NSW Supreme Court, Practice Note SC Gen 23 (reissued 28 January 2025, commenced 3 February 2025). Paragraph 6 says Gen AI, for the practice note, does not include technology that merely provides transcription. A tool that writes a structured note does generate content, so assume the practice note applies to it. Paragraph 3 names Read AI, a meeting note taker, among its examples of Gen AI programs. Paragraph 9A bars entering suppressed, subpoenaed or Harman-undertaking material into a Gen AI program unless confidentiality conditions are met.
• Supreme Court of Victoria, Practice Note SC Gen 25 (issued and commenced 14 May 2026, replacing the May 2024 guidelines for litigants). Court users should not enter confidential or sensitive information into a public AI tool. For a closed AI tool they must satisfy themselves, by reviewing the contractual terms or settings, that the information stays in a secure environment (paras 6.8 and 6.9). Affidavits and witness statements must reflect the person's own knowledge and words (para 6.11).
• Federal Court of Australia, Use of Generative Artificial Intelligence Practice Note (GPN-AI) (16 April 2026). Paragraphs 4.13 to 4.15 warn that information given to a generally accessible Gen AI tool may become available to others, and that privileged or confidential information must not be entered in a way that breaches the obligations attached to it.

For evidence, the practical rule is the same in all three. An AI-drafted note can refresh your memory. The affidavit is still your own recollection in your own words. See file notes as evidence, and can lawyers use AI for file notes? for the conduct rules one by one.

The checklist to run on any AI note taker

These questions are drawn from the Proctor article and the joint checklist. The right-hand column is CallNote's answer to each, including the ones where we are not the strongest option.

Record the answers, date the document and file it with your AI policy. The Proctor article says to keep a record of the assessment, and the joint checklist says to document the decision and map review intervals. Our AI note taking policy template gives you somewhere to put it.

What a reviewed note looks like

Whatever tool drafts it, the note on the file should read like this after a solicitor has checked it. The client and matter are invented.

The alternatives, fairly

There are four kinds of tool a firm will look at. What follows is only what each vendor says on its own website, checked in October 2026.

• AI inside your practice management system. Smokeball's Archie is described on Smokeball's site as an AI assistant that works with a matter's full context across Word, Outlook and Smokeball, with add-on apps for jobs such as turning a recording into a transcript. Smokeball says Archie runs under zero-data-retention agreements and that Smokeball is certified to ISO 27001:2022. If you are a Smokeball firm, this is the better pick for work that lives in the matter.
• A legal phone system that records. VXT says it records calls, writes a file note after every call and meeting, and saves it to legal software including LEAP, Smokeball, Actionstep and Clio. If you want the phone, the recording and the practice management link from one vendor, VXT is the stronger fit. You take on recording consent in exchange. See recording law by state.
• Microsoft Teams with Copilot. Microsoft's support page says Copilot can be used in a meeting without recording or transcription when the organiser picks 'Only during the meeting'. No meeting transcript is created, though Copilot prompts and responses may still be retained under the organisation's Microsoft Purview retention policies. That suits a firm that wants no transcript kept, but it leaves no transcript to build a file note from.
• Wearable recorders. These capture in-person meetings. Consent and storage need their own analysis, covered in Plaud for lawyers in Australia.

CallNote sits in a fifth spot. It turns a transcript your phone or meeting system already produced into an attendance note in your firm's format. Transcripts arrive automatically from Aircall, Dialpad and Microsoft Teams, or you paste, upload or email one in. It is the wrong choice if you need a matter-level link into LEAP, Smokeball or Actionstep. More on how it works for law firms, and a wider comparison in best attendance note software for law firms.

Common questions