Update on generative AI: disputes considerations and tips for businesses

2025 saw increasing scrutiny on AI’s role in legal disputes – particularly the use of AI in litigation and whether public-facing generative AI (Gen AI) systems should be considered “digital strangers” for privilege purposes.

The courts are also seeing concrete examples of AI outputs containing hallucinated authorities or otherwise unreliable content being put before decision-makers. This has been spurred by the use of ChatGPT and other Gen AI systems, which are creating evidentiary and procedural risk for the courts.

While recent decisions have shed light on how these questions may be answered in other jurisdictions, there is no case law or regulatory framework in New Zealand that directs how Gen AI systems should be treated for the purpose of legal professional privilege. Our view is that a court is likely to apply existing analogies from developing case law in similar jurisdictions to determine whether legal advice generated by, or with the assistance of, Gen AI can be privileged.

The following best practice considerations are relevant for responsible use of Gen AI in the legal context, and the preservation of legal privilege:

1. Public-facing Gen AI systems should not be used to generate, or assist with generating, legal advice. Recent decisions suggest that documents produced using public-facing Gen AI models may not attract privilege because they are not kept confidential, even if input by a lawyer (in-house or external). Privilege is less likely to attach to documents created by a non-lawyer using AI, as the documents are not communications between a client and a lawyer, but between a client and AI (United States v Heppner (S.D.N.Y., Feb. 2026)).

2. Where bespoke AI tools are used, use should be under the direction and supervision of a qualified lawyer. For example, enterprise Gen AI systems with contractual and system confidentiality protections in-built may be used to assist with the production of legal advice, provided such use is under appropriate legal supervision. Consider keeping a record of Gen AI use to demonstrate the degree of supervision and direction exercised in the production of legal advice.

3. Treat prompts and outputs as potentially discoverable. Treat all prompts provided to, and outputs generated by, Gen AI tools as material that could be discoverable in legal proceedings. While there is currently no judicial guidance in New Zealand on whether Gen AI chat logs constitute “documents” for the purposes of the High Court Rules 2016, the wide definition of “documents” is likely to capture these chat logs, subject to any valid claim of legal professional privilege.

4. Use AI-generated content with caution. Documents and information created by Gen AI should be used with caution. In a recent Supreme Court of New Zealand decision, the Court noted that reliance on false citations, including unverified outputs of AI applications, may in serious cases amount to obstruction of justice or contempt of court. In that case, a litigant in person had used Gen AI to draft submissions in a child custody matter, which included hallucinations and erroneous references to cases.

5. Stay abreast of regulatory and professional guidance. AI is moving at a pace that regulation, governance and businesses cannot match. It is important that in-house lawyers, as regulated individuals, stay abreast of regulatory and legal developments in this area. In our December 2025 Litigation & Regulatory Quarterly Brief we outlined the New Zealand Government’s latest AI strategy and Public Service AI Framework. In addition, the New Zealand Law Society has published a guidance note for lawyers on generative AI.

For more information, please refer to the comprehensive guide prepared by our UK team, on the current status of AI and legal privilege.