Artificial intelligence (AI) has revolutionised how people engage with the law and obtain legal advice. Increasingly, clients are using AI systems to help them frame their requests for legal advice. However, it is important to be mindful of the risks associated with the use of AI in the court process.
Legal professional privilege
Legal professional privilege (LPP) is a right protecting confidential communications between a lawyer and a client from compulsory disclosure, such as to a court or regulatory body or to the opposing party. The protection afforded by LPP encourages full and frank disclosure by a client to a lawyer and facilitate the provision of informed legal advice.
There are two main forms of LPP:
- Advice privilege — which attaches to communications for the dominant purpose of the giving or receiving of legal advice or the provision of legal services; and
- Litigation privilege — relating to communications for the dominant purpose of being used in actual or reasonably anticipated litigation.
The ability to claim LPP is not absolute and may be waived voluntarily or inadvertently in certain circumstances, including (but not limited to) where confidentiality of a communication or document is not maintained.
Critically, where LPP has been waived in whole or in part, solicitor-client communications may lose this protection and third-party disclosure may become mandatory, such as in a litigation proceeding or regulatory investigation.
Inadvertent waiver of LPP — the use of AI
The advent of AI, particularly publicly available AI systems (such as OpenAI’s ChatGPT, Claude, Harvey and others), has presented a novel way in which LPP may be waived.
When confidential information is input into a publicly available AI system, the information is accessible to third parties who may retain and analyse prompts, and by extension, it is placed in the public domain. Where third parties can access information, the essential characteristic of confidentiality (which is the cornerstone of LPP) is lost. As a result, requests for legal advice, internal discussions about your case and confidential documents may need to be produced.
Examples of client use of AI systems
We have observed that AI systems have been used by clients to:
- Provide legal advice.
- Consider legal advice already provided by solicitors such as in letters or emails.
- Summarise information and prepare ‘case briefs’ for solicitors.
- Transcribe discussions including with solicitors during online meetings using AI transcription tools. AI transcription tools are often built into teleconferencing services such as Zoom and Teams and are capable of transcribing online meetings and providing a written transcript. The risk is that data provided to publicly available AI transcription tools may be stored and analysed by third parties, thereby waiving LPP.
United States v Heppner
In United States v Heppner1, decided on 10 February 2026, the US District Court for the Southern District of New York considered whether confidentiality was abandoned and LPP waived due to a client’s use of a generative AI system in criminal proceedings.
The defendant, Bradley Heppner, uploaded documents to Claude and prompted it to outline possible arguments that may be raised in opposition to the relevant charges against him.
In summary, the court found that communications between Bradley Heppner and Claude were not privileged at the time they took place for the following reasons:
- The communications were not between a client and a lawyer, given that Claude is not a lawyer.
- The material input to Claude was not confidential in view of the terms of Claude’s privacy policy which provided that, among other things, data may be disclosed “to third parties in connection with claims, disputes or litigation”. Therefore, Bradley Heppner could not have had any reasonable expectation of confidentiality.
- The court rejected the argument that LPP was capable of being created retrospectively by Bradley Heppner sharing Claude’s outputs with his solicitors.
Recent developments in Australia
Australian courts have not delivered a landmark decision concerning the use of AI and waiver of LPP in the same way as the court did in United States v Heppner. However, decisions such as United States v Heppner would likely be persuasive and at the very least considered, should a similar case arise in Australia. It is foreseeable that a decision of this kind will be handed down in Australia in the near future.
Various Australian jurisdictions2 have responded to the widespread use of AI in recently released Practice Notes intended to guide litigants as to the use of AI systems. While courts have embraced the use of AI to enhance access to justice and improve efficiency, they make it clear that AI systems should be treated with caution and seek to make users mindful of potential adverse consequences flowing from the same.
For example, the Federal Court of Australia, by way of its recently released Use of Generative Artificial Intelligence Practice Note, warns that AI systems may generate incorrect or misleading information, fictitious cases and incorrect case citations.
Turning to the risk of inadvertent waiver of LPP, the Federal Court Practice Note expressly states that:
“If information is provided to a generally accessible Generative AI tool (such as a standard Generative AI tool), it may become available to other people. Users may not know where that information is stored, how it is used, or who will have access to it”.
The Federal Court has highlighted the possibility of inadvertent waiver of LPP through the improper use of AI systems and for practical purposes it has foreshadowed that the court’s reasoning in United States v Heppner relating to the waiver of LPP may be applied in Australian jurisdictions.
Our recommendations
- Do not upload or prompt AI models using privileged or confidential information. In particular, do not upload legal advice to AI models or you risk waiving LPP.
- Refrain from using publicly available AI models because any confidentiality is lost upon doing so. Closed or purpose-built generative AI models should be preferred as these do not place inputs into the public domain and are less likely to be subject to contract terms allowing third parties to store or analyse inputs.
- Seek professional legal advice if you are unsure about your obligations when using AI or how AI may impact your legal rights and entitlements.
- Disclose the use of AI to your solicitor, particularly if generative AI has been used in the preparation of court documents, which must be disclosed to the court.
Contact us
Rigby Cooke Lawyers can assist with navigating these emerging issues and mitigating the risks associated with the use of AI in legal matters. To discuss how AI may affect your legal rights, obligations or ability to maintain legal professional privilege, please contact a member of our Litigation & Dispute Resolution group.
References
1. United States of America v Heppner 25 Cr 503 (JSR), United States District Court for the Southern District of New York, 17 February 2026.
2. Including the Federal Court of Australia (Use of Generative Artificial Intelligence Practice Note (GPN-AI)); Supreme Court of Victoria (Practice Note SC Gen 25 The Use of Artificial Intelligence by Court Users); County Court of Victoria (PNCCV 1-2026); and Victorian Civil & Administrative Tribunal (VCAT) (Use of Generative Artificial Intelligence Practice Note (PNVCAT 11).
| Disclaimer: This publication contains comments of a general nature only and is provided as an information service. It is not intended to be relied upon, nor is it a substitute for specific professional advice. No responsibility can be accepted by Rigby Cooke Lawyers or the authors for loss occasioned to any person doing anything as a result of any material in this publication.
Liability limited by a scheme approved under Professional Standards Legislation.© 2026 Rigby Cooke Lawyers |