The risk of losing privilege by using AI to help with legal disputes

As AI tools become more embedded in day-to-day business operations, many organisations are starting to use them to help with early dispute analysis, drafting and strategy. While these tools can be efficient and insightful, their use carries a significant risk: the accidental loss of privilege. For businesses facing contentious issues, understanding how AI interacts with privilege is essential.

What is privilege

Privilege is a legal protection that keeps certain communications confidential. When it applies, those communications do not have to be shared with other parties, including regulators and the courts.

For most businesses, the two most relevant types of privilege are:

• Legal advice privilege, which protects confidential written or oral communications between a lawyer and their client, where the document was prepared for the purpose of giving or receiving legal advice.
• Litigation privilege, which protects confidential written or oral communications between a lawyer and client, or between one of them and a third party. This is provided the communications are created for the dominant purpose of litigation that is in contemplation or has already commenced.

For privilege to apply, it is important that the communication remains confidential and fits within these boundaries. Once confidentiality is broken, intentionally or accidentally, privilege can be lost, often permanently.

What are the risks of using AI

AI tools can be useful for summarising documents, spotting patterns, and drafting internal notes. However, in the context of legal disputes, they raise several risks that can unintentionally undermine privilege, particularly when draft legal documents and legal advice is uploaded into these systems.

One of the most obvious risks to privilege occurs when someone uploads privileged legal advice, including draft submissions or privileged communications, into an AI tool, particularly those that are free or publicly available, such as ChatGPT.  This is often likely to be inconsistent with maintaining confidentiality, waiving privilege. Privilege can be lost if protected information is voluntarily shared with a third party who is not part of the confidential relationship. Where an AI service provider’s terms of use expressly permit them to access, store or disclose user data, it becomes increasingly difficult to maintain that the information has been kept confidential and to argue that privilege remains intact.

Another important consideration is that legal advice (or draft documents) generated by AI tools cannot attract legal advice privilege. As the advice is generated without lawyer involvement, it will clearly fall outside the lawyer-client protected relationship. Even if a lawyer is later instructed, it is arguable that privilege was waived when the content was first shared to the AI tool, or that the AI generated content should be treated as a pre-existing document, which are generally not protected by privilege.

Litigation privilege is broader, as it covers client communications with third parties. Therefore, it could be argued that the AI chatbot is a third party for these purposes, although this is not clear. In any event, whether these communications are privileged is likely to hinge on whether they are confidential.

It is also important to note that some AI providers store or learn from the user’s input. If that includes privileged material, it could be accessible in ways the business cannot control.

How can those risks be mitigated?

It is important that those using AI understand that this comes with a risk of waiving privilege. Those that wish to use AI to assist them should:

• Ensure that they are familiar with the terms and conditions of the AI tool or chatbot that they are using, particularly those relating to confidentiality and how the data will be used;
• Avoid using public AI tools for legal disputes or issues that may result in in litigation. In particular, care should be taken to avoid uploading sensitive material into public AI tools; and
• Educate teams and put policies in place on the use of AI. This could include setting out what counts as privileged, how easily it can be lost and when and how AI tools may or may not be used.

Summary

AI can be a valuable support tool but using it in the context of legal disputes creates real risks around confidentiality and privilege. Uploading privileged material into public AI platforms may amount to disclosure to a third party and is likely to fall outside the protected relationship. Organisations wishing to benefit from AI without undermining their legal position should proceed with caution, carefully reviewing AI platform terms and avoiding use of public tools for dispute related matters.

How we can help?

For further information and advice on resolving your disputes contact our dispute resolution team who will be able to provide assistance as necessary.