Beware of artificial intelligence and the potential waiver of legal professional privilege

Legal professional privilege protects confidential communications and documents between a solicitor and client made for the dominant purpose of providing legal advice or use in current or anticipated litigation. 

Where legal professional privilege applies, a party is not required to disclose the relevant communication or document. However, if a party acts in a way that is inconsistent with the privileged document or communication remaining confidential, for example, by disclosing legal opinion or advice, then privilege may be waived. To ensure that privilege is maintained, practitioners must handle privileged material prudently, especially when incorporating the use of AI into practice.

Most open generative AI tools operate by processing user-provided inputs on third-party servers to generate responses.  AI algorithms can assist with a huge gambit of legal tasks, from the preparation of research and correspondence to the review and interpretation of pertinent documents. 

To improve how practitioners deliver these services, and in order to better assist clients, some practitioners may be tempted (of course, with the client’s consent) to upload client documents, communications or advice for this summarisation, analysis or redrafting. 

In doing so, the practitioner may be disclosing confidential information, if this information is uploaded to a third-party provider that sits outside the lawyer-client relationship. Disclosure to a third-party server may waive legal professional privilege if it results in the dissemination of information, and the undermining of the confidentiality of the document or communication. 

The risk of waiver is compounded by the functionality of generative AI tools. 

Many publicly available AI tools, such as ChatGPT, contain terms of use that grant the platform broad rights over user inputs, including the ability to store, process and use the provided information to improve the AI model. If confidential client information or material is uploaded, it may be retained on third-party servers and potentially exposed through later investigations by other users, system vulnerabilities or data breaches. This loss of control over the information is fundamentally inconsistent with the maintenance of legal professional privilege.

This concern may be addressed by the use of legally focused closed data-set AI tools, rather than those publicly available. Paid, in-house or firm hosted tools often provide greater confidentiality, clearer terms and often draw only on their own database (rather than, for example, the entire world wide web). These tools also often provide a more secure basis for documents to be uploaded in a way that is not broadly accessible. 

Whether legal professional privilege has been waived turns on the terms and conditions of any AI platform being used, and the extent to which this tool is actually closed.  

It is crucial that practitioners exercise care to ensure they understand and are satisfised with the terms and conditions of any AI tool being used, including the rights that AI developer has over information inputted therein, before uploading confidential material. 

While there hasn’t been an authoritative ruling by Australian courts, recent judicial consideration supports the position that uploading documents into AI could constitute a waiver of legal professional privilege. 

In Helmold v Mariya (No 2),[1] the court observed that the ‘input of documents arising out of the proceedings into a generative AI program which stores, collates and replicates data may waive privilege or fall foul of the requirements that certain matters be treated as commercial in confidence’.[2]  The court emphasised that these risks warrant ‘extreme caution’[3] by practitioners.

Similarly, in Mertz v Mertz (No 3),[4] the court warned that ‘there is a risk that entering draft documents into an AI program will…give rise of a waiver of legal professional privilege’.[5]

Regulators have reinforced this position. The joint statement issued by the Victorian Legal Services Board and Commissioner, the Law Society of NSW and the Legal Practice Board of WA make it clear that ‘lawyers cannot safely enter confidential, sensitive or privileged client information into public AI chatbots/co-pilots (like ChatGPT) or any other public tools’.[6] It warns that if practitioners elect to use generative AI tools, they must carefully review the contractual terms of the platform to ensure that the information will be kept secure.

In line with our ongoing commitment to Smarter Law, Hall & Wilcox is committed to harnessing the benefits of specialised, in house and legally focused AI tools to provide seamless and efficient outcomes for clients. 

We would be pleased to discuss safeguarding legal professional privilege through the use of AI, as well as the responsible use of AI at Hall & Wilcox.  Please reach out to Catie Moore, Partner or Lauren Separovich, Special Counsel should you wish to discuss. 

This article was prepared with the assistance of Shivaani Sivaraj, Seasonal Clerk.