One wrong click: how public AI tools can collapse privilege

The rapid rise of generative AI has fundamentally reshaped how legal work is performed, offering unprecedented speed and efficiency. But alongside that opportunity comes a growing body of authority warning of the risks - none more significant than the loss of confidentiality and legal professional privilege.

The recent decision of the Upper Tribunal (Immigration and Asylum Chamber) in Munir -v- Secretary of State for the Home Department [2026] UKUT 81 (IAC) is the first case in England and Wales to address directly whether using public AI platforms can jeopardise legal privilege, and the Tribunal’s conclusion leaves little room for doubt.

Although the case arose from concerns about practitioners submitting AI generated material containing hallucinated case citations, the Tribunal used the opportunity to provide important guidance on confidentiality in the digital age.

The Tribunal held that:

“Uploading confidential documents into an open source AI tool, such as ChatGPT, is to place this information on the internet in the public domain, and thus to breach client confidentiality and waive legal privilege.”

This is the first time an English court has explicitly linked the use of public AI platforms with the loss of privilege. The implications are profound.

Why this case matters

Public AI means public domain

The core issue is confidentiality. Privilege only applies where communications are kept confidential. Once confidentiality is compromised (whether deliberately or inadvertently) privilege falls away. The Tribunal treated the act of uploading privileged or confidential documents into a public AI tool as functionally equivalent to making them available to the public. It also drew a clear distinction between public/open AI tools (e.g., ChatGPT) and closed enterprise environments (e.g., Microsoft Copilot), observing that closed systems which do not place information in the public domain can be used without the same confidentiality risk.

Privilege once lost cannot be retrieved

The Tribunal’s comments reflect long established English common law: waiver of privilege is permanent. Once material enters the public domain, there is no mechanism to 're privilege' it:

'Once legal privilege is waived, it is lost. It cannot be recovered. There is no mechanism by which a party can retrospectively assert privilege over material that has entered the public domain.'

For organisations that rely heavily on the protection of privileged communications, including regulated entities, in house teams, and professional advisers, the message is clear: a single misstep involving AI can have irreversible consequences.

AI misuse and professional accountability

The judgment also forms part of a wider pattern of judicial scrutiny of AI use by legal professionals. The Tribunal criticised practitioners who had submitted hallucinated citations and stressed that:

• AI cannot replace legal analysis;

• lawyers remain responsible for supervising work, including AI assisted work; and

• failing to supervise properly may lead to regulatory referral.

This underscores that the convenience of AI does not dilute professional duties, and misuse may invite both forensic and regulatory consequences.

What this means for legal teams and organisations

The Munir decision is likely to shape the profession’s approach to AI. For law firms, in house legal teams and corporates, the key implications are:

• Do not input confidential or privileged material into public AI platforms.

• Where AI is necessary, use enterprise grade tools with contractual and technical safeguards preventing disclosure or training use.

• Train all staff (legal and non legal) on privilege, confidentiality, and approved AI workflows.

• Update governance frameworks and policies to align with the Tribunal’s findings and your risk appetite.

At Hill Dickinson we work alongside you with clarity, warmth and commitment, to deliver the right outcomes whenever you need us. Feel free to get in touch.