Artificial Intelligence (AI) is starting to reshape how legal disputes are prepared, presented and managed. Open AI tools such ChatGPT, Gemini and Claude AI (Open AI) are now commonly used by the general public to draft documents, summarise evidence and assist with legal analysis.

On the face of it, AI tools may offer clear efficiencies. However, recent developments highlight that its use can carry significant legal risk. Courts and Authorities in New Zealand and other jurisdictions have started signalling a clear message: AI can assist, but it does not replace professional judgement, and its issue can have serious consequences.

Loss of Legal Professional Privilege

Courts in other jurisdictions have raised a significant issue resulting from the misuse of AI tools: the potential loss of legal privilege and confidentiality.

Legal privilege and confidentiality protect communication between lawyer and client by allowing open, transparent and candid communication without needing to disclose to the other party. Privilege depends on confidential communications between a client and their lawyer. Privilege, as a broad statement, is lost if confidentiality is lost.

Open AI tools work by processing and retaining user inputs. The legal risk it poses is this: entering sensitive and confidential information into an Open AI tool which is publicly available to all users may amount to disclosure to a third party. Therefore, privilege and confidentiality are lost.

In United States v Heppner (2026), the defendant used Open AI to analyse legal strategy and prepare material for his lawyers. Here, the Court rejected privilege entirely because:

• the communications with Open AI are not communications with a lawyer;
• there was no reasonable expectation of confidentiality; and
• inputting information into Open AI amounted to disclosure to a third party.

In UK v Secretary of State for the Home Department (Hamid) [2026] UKUT 81, the Tribunal observed that uploading confidential material into an Open AI platform is equivalent to placing it in the public domain. This resulted in waiver of privilege and confidentiality.

Accuracy Remains the User’s Responsibility

In a New Zealand case - O’Driscoll v Rehab Co Mobile Limited - the Employment Relations Authority (Authority) has considered the risks associated with AI-generated material.

In O’Driscoll, a self-representing party relied on submissions generated by an AI tool which generated fake legal information. The Authority responded by awarding increased costs of $2,000.00 against the self-represented party. This reinforces the position where parties are responsible for the accuracy and reliability of anything filed in proceedings, especially those who are self-representing.

These risks are starting to have real consequences during a legal proceeding. In O’Driscoll, it delayed resolution and increased costs for both parties and their respective lawyers where time was not well spent trying to track down legal cases which did not exist.

As such, Courts and Authorities are starting to take a proactive approach by awarding increased costs, disregarding material or withdrawing it completely if it has not been verified

Key Takeaways

There is no doubt that AI tools can be used as a valuable tool in and out of the legal profession. However, the Courts and Authorities are clearly stating that AI tools are not changing the underlying legal principles. As both lawyers and clients should remember when using AI tools:

• Do not rely on AI as a source of legal authority; and
• Exercise with caution when handling confidential information.

Mortlock McCormack Law’s use of AI is restricted to secure AI tools only to ensure your data is protected and all staff are trained to review and verify AI results.

If you have any questions or concerns, please contact Ariana Kim, Solicitor ([email protected]/03 925 7956) or Jeff McCall, Principal ([email protected]/03 925 7951).