The Advocate | Issue 373 | May 2026

Recently in the media we have seen a number of cases where self-represented litigants have advanced their claims with the assistance of AI. 

For example, in December 2025, it was reported that a woman in the United Kingdom was the first person in the UK to win her legal battle using an AI law firm. The woman was a healthcare worker who had been placed on a performance improvement plan. She was unable to afford a solicitor to assist her, so she engaged with Grapple Law (a fully automated law firm with no actual lawyers working on its cases). It was reported that the woman resolved her case and was paid £30,000 by her employer. 

We have also seen in decisions from the Courts, that there has been increased use of AI, particularly by self-represented litigants, in preparing for and advancing their claims. 

This raises the question, are the free and frank communications had with AI chatbots privileged, or could they be required to be disclosed by a Court?

Legal privilege is an important concept because it encourages free and frank communication between solicitor and client and provides certainty and trust that those communications will be confidential. 

This allows clients to manage risks without exposing sensitive information, and lawyers can work effectively knowing that their advice and communications cannot be disclosed. It is a fundamental aspect of our legal system in terms of upholding the rule of law.

This issue has been considered in the United States. In US v Heppner , No. 1:25-CR-00503 (S.D.N.Y.), the District Court for the Southern District of New York held that documents a defendant generated using AI were not covered by attorney-client privilege or “ work produc t” doctrines (i.e. where documents are prepared for the purpose of seeking legal advice). 

Turning to New Zealand, the issue is yet to be considered by the Courts.

Under New Zealand law, legal advice privilege applies where a person requests or obtains professional legal services from a legal adviser. “Legal adviser” means a lawyer, a registered patent attorney, or an overseas practitioner” (s 51 of the Evidence Act 2006). 

Although a person may request or obtain (arguably) professional legal services from AI, the key difference is that it is not obtained from a legal adviser. Therefore, legal advice privilege cannot apply. 

In the Employment jurisdiction, there is a limited extension to privilege. In that it can extend to communications with non-legal advisors (advocates), where the person is a party to proceedings in the Authority/Court. Specifically, clause 3(1) of Schedule 2 of the Employment Relations Act 2000 provides that: 

“Where any party to any matter before the Authority is represented by a person other than a barrister or solicitor, any communications between that party and that person in relation to those proceedings are as privileged as they would have been if that person had been a barrister or solicitor.”

This extension is primarily focused on litigation privilege, whereby privilege attaches only when litigation is contemplated or pending, and not to all legal advice given in the absence of such proceedings. The effect is that unless the communications are to/from a legal advisor (as defined above), privilege does not extend to general advice sought or given outside of current, pending or reasonably contemplated litigation. 

We know that the Authority/Court are prepared to require disclosure of communications between lay advocates and HR consultants, where it does not relate to litigation that is current, pending or reasonably contemplated (Faitala v Pacific Island Business Development Trust [2024] NZERA 34). Whether or not privilege could apply to AI communications is untested in the Employment jurisdiction. 

The argument may come down to whether AI could be considered a “person” in terms of clause 3(1) of Schedule 2 of the Employment Relations Act 2000. We asked CoPilot, and it told us that “it is not a person – it’s a tool made by people”, the key difference is “even those AI can seem human when it talks or writes, it does not think or feel in the human sense, it does not have opinions or desires of its own, and it simply generates responses based on patterns it has learned”. Under New Zealand law, the term “person” is generally taken to mean “a natural (human) person”, so this argument may not go far. 

However, there is a further extension to privilege in the litigation context, whereby it can extend to documents/materials prepared for the dominant purpose of preparing for a proceeding (s 56 of the Evidence Act 2006). There is potentially a stronger argument that preparatory materials privilege applies to chats with AI, where those chats and anything created by the AI as a result of them, were for the purpose of preparing for a proceeding. 

In summary, there is an argument that in some circumstances conversations with AI chatbots could be required to be disclosed by the Authority/Court where they were not made for the purpose of preparing for a proceeding. Conversely, all communications with lawyers, including the team at MGZ, are privileged.