Chancellor warns AI use could put legal professional privilege at risk

In a speech to the City of London Law Society this week (22nd April), he warned that while artificial intelligence has the potential to transform legal work, it does not dilute the long‑established principles that underpin privilege, nor justify a relaxation of professional standards.

The Chancellor made clear that AI does not change the legal test for privilege. Legal advice privilege and litigation privilege remain grounded in confidentiality and dominant legal purpose. However, he warned that AI tools introduce new and often poorly understood risks of waiver, particularly where information is shared with third‑party systems or processed outside secure legal environments.

A key concern, the Chancellor said, was the tendency for users to treat AI as a neutral “tool”, rather than as a third party. Where confidential or legally sensitive material is input into externally hosted or consumer‑grade AI systems, privilege may be lost if confidentiality cannot be guaranteed, regardless of the user’s intention.

This, he suggested, is not a theoretical risk. Developments in case law and regulatory guidance increasingly indicate that courts will examine how AI tools operate in practice, including their data retention policies and terms of service, when determining whether privilege has been waived.

The Chancellor cautioned that if AI tools are used without clear governance, there is a real danger that privileged material could be disclosed inadvertently, with potentially irreversible consequences.

Once privilege is lost, it cannot be selectively reclaimed, increasing the risk of disclosure in litigation, public inquiries or under freedom of information and data protection regimes.

The Chancellor rejected any suggestion that AI should be treated as a special category exempt from established doctrine. He stressed that the courts have historically been reluctant to extend privilege beyond the narrow lawyer–client relationship and that AI does not alter this position.

Communications generated by or shared with AI systems will therefore be assessed by reference to orthodox principles, including whether confidentiality has been maintained and whether the communication was genuinely created for the purpose of giving or obtaining legal advice.

In practice, this means that lawyers and public bodies cannot assume that AI‑assisted drafting or analysis will automatically attract privilege, particularly where the system is not operating under the direct control of the legal team.

The Chancellor emphasised that protecting privilege in the AI age is primarily an issue of governance rather than technology. Organisations must put in place clear policies governing when and how AI may be used, what types of information may be input, and which tools have been formally approved. He also underlined the importance of training both lawyers and non‑lawyer officers to understand the legal consequences of using AI systems inappropriately.

Derek Bedlow