AI Policy

​

​This policy sets out how my practice uses artificial intelligence tools, where AI is not permitted, and how I maintain accountability to my clients, the Court, and my professional obligations under the Lawyers and Conveyancers Act 2006 and the Conduct and Client Care Rules.

​

1. What We Use AI For

AI tools may be used, by approved personnel only, for the following tasks:

• Initial legal research and issue-spotting** — using AI to identify relevant lines of enquiry, statutory provisions, or case themes as a starting point, always followed by independent verification through primary legal databases (e.g. Westlaw NZ, LexisNexis NZ, NZLII).

• Drafting assistance — generating first-draft correspondence, memoranda, chronologies, or skeletons for submissions and affidavits, which are then substantively reviewed and rewritten by myself.

• Summarising large volumes of disclosed documents** — e.g. bank records, medical notes, or financial statements, discovery-heavy relationship property or trust disputes, to help locate material for human review, not to reach conclusions.

• Administrative and practice-management tasks— file notes, meeting summaries, scheduling correspondence, formatting, and proofreading.

• Plain-language explanations for clients — drafting simplified explanations of legal concepts or processes, for my review before sending, particularly useful given to clients who have not been involved in litigation before.

 

The AI tools that I use are reviewed periodically for data security, confidentiality safeguards, and jurisdiction of data storage. I will not put in any client-identifiable information into any AI tools that have not been approved and confirmed to meet our confidentiality and data-retention requirements.

​

 2. What Stays Human

The following remains my exclusive responsibility as your barristerm and are never delegated to or generated by AI without full independent verification:

• Legal advice and strategic judgment — advice to clients on prospects, strategy, settlement positions, or the exercise of discretion requires professional judgement informed by the specific facts, the client's circumstances, and case law that has been personally checked.Assessment of capacity and vulnerability— in PPPR matters particularly, judgments about a person's capacity, welfare, and best interests must be made by the lawyer (and appropriate medical/professional evidence), never inferred or summarised by AI in a way that substitutes for that judgement.

• Case law and citation verification— every authority cited in any document filed with the Court or relied upon in advice is independently checked against the primary source. AI-generated citations are never relied upon without this check, given the well-documented risk of fabricated or hallucionated authorities.

• Client communication involving sensitive or contested matters — advice on family dynamics, contested estates, allegations of undue influence or incapacity, and similarly sensitive matters is delivered personally, not via AI-drafted material sent without review.

• Final review and sign-off of all filed documents — pleadings, affidavits, submissions, and any document filed with the Court or sent to opposing counsel is reviewed and approved by me. I take full professional responsibility for its content.

• Evidential weight and credibility assessment— assessing witness credibility, weighing conflicting evidence, and forming views on disputed facts remains a human function.

​

3. How We Stay Accountable

• Named oversight — I am responsible for maintaining this policy, approving tools, and reviewing it at least annually or when new tools are adopted.

• Confidentiality and privilege— no client-identifiable or privileged information is entered into an AI tool unless that tool has been confirmed not to use inputs for model training and to meet our data security standards. Where in doubt, information is anonymised or withheld.

• Disclosure to clients— clients are told, in engagement terms or on request, that AI tools may be used to assist with drafting, research, or administrative tasks, and that all substantive advice and filed documents are reviewed by a lawyer.

• Verification requirement — every AI-assisted output is checked against primary sources before use. No AI output is treated as authoritative on its own.

• Record-keeping — where AI has materially contributed to a document (e.g. a first draft of submissions), a log of this is kept for internal quality-assurance purposes.

• When something goes wrong — if an AI-related error is identified (e.g. an inaccurate citation, a factual error, or a confidentiality breach), I will:

  • ​Correct the error immediately and assesses whether it has reached the client, opposing party, or the Court;

  •  Notify the client and, where the error has been filed with or relied upon by the Court, take prompt steps to correct the record;

  • Consider whether the error triggers any professional obligations, including under the Conduct and Client Care Rules, and seek guidance, including from the New Zealand Law Society, if appropriate.

• Ongoing review — this policy will be revisited as AI tools, professional guidance from the Law Society, and Court practice notes evolve.

​