FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Aipedia: Barristers
Barristers should not put client material into free public AI tools, because the BSB says such tools are "unlikely" to meet the confidentiality conditions. Self-employed barristers are typically their own data controllers, so the choice of tool and its terms is personal.
Part of: AI in Barristers
In short
Free AI tools often claim rights over inputs and prompts. The BSB's May 2026 guidance therefore limits them to the lowest-risk tasks, such as general learning or training with no case details. In Munir the Upper Tribunal said uploading client letters or Home Office decisions to a public AI tool puts them in the public domain, breaching confidentiality and waiving privilege. CD6 and rC15 protect confidentiality, rC86 treats AI as outsourcing, and data protection duties apply. BMIF minimum terms cover third-party loss but not first-party cyber costs, so a leak can fall between policies.
The BSB's AI guidance points out that the terms of free public tools often let providers claim rights to inputs and prompts, and may allow them to be used for training. On that basis the BSB says it is "unlikely" that free, publicly available AI systems meet the confidentiality conditions, and limits them to the lowest-risk tasks: general learning and training exercises with no case details. The Bar Council's updated considerations list confidential data being used in training among the main risks.
In Munir [2026] UKUT 81 (IAC) the Upper Tribunal said at ¶21 that uploading client letters or Home Office decisions to a public AI chatbot puts them in the public domain, breaching confidentiality and waiving privilege, and that firms should consider reporting to the regulator and the ICO. In a speech on 22 April 2026, the Chancellor of the High Court said confidentiality is a prerequisite for privilege, that public tools appear not to be confidential, and that with secure tools it is "hard to see" any impact on privilege; the speech framed questions rather than deciding them. See AI in immigration appeals for the tribunal context.
The BSB's governance points give a checklist: who owns inputs and outputs, whether data is used for training, whether data leaves the UK, deletion and prompt-retention policies, audit logs, privacy settings and access restrictions on shared chambers systems, whether the tool meets ICO standards, and what happens to data if the vendor fails. Check the instructing solicitor's position too; see AI bundle review and AI notetakers.
Removing names before pasting material into a tool reduces risk but rarely removes it. Facts, dates, places and the shape of a dispute can identify a client, and documents such as decisions or expert reports are often identifiable from their content alone. Anonymised material may still be confidential and privileged. Anonymising is a sensible extra precaution with an approved tool, not a way to make a free public tool safe for case work.
Illustrative: counsel pastes a confidential expert report into a free chatbot to get a plain-English summary for a lay client. The tool's terms allow inputs to be used for training. The instructing solicitor learns of it and the opposing party later asks whether privilege in related material has been waived. The client asks who will pay for the satellite dispute.
A client's claim for loss caused by a confidentiality breach is usually a third-party civil liability claim within BMIF cover under the BSB minimum terms, which contain no AI exclusion. The BSB's PI page says first-party cyber cover is not required, so costs of investigating, notifying and responding to the leak may be uninsured without a cyber policy. Fines are excluded. On notifying insurers without worsening a privilege problem, see preserving privilege when notifying insurers and does BMIF cover AI mistakes.
Apex arranges PI for barristers and can explain how BMIF and a cyber policy would share a confidentiality incident. More on the AI at the Bar hub.
Yes, for the lowest-risk tasks the BSB describes, such as general learning or training exercises that contain no case details. Do not put client names, documents or instructions into a free public tool.
The Upper Tribunal in Munir said uploading client letters or Home Office decisions to a public AI chatbot waives privilege. The Chancellor of the High Court said secure tools are unlikely to affect privilege, but the wider question has not been decided.
It is not a reliable fix. Facts and documents can identify a client even without names, and anonymised material can still be confidential and privileged. The BSB limits free public tools to general learning and training with no case details.
A client's claim for resulting loss is usually within BMIF cover, subject to the terms. First-party costs such as investigation and notification are not required by the minimum terms and may need a cyber policy. Regulatory fines are excluded.
Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.
Start a PI quote Request a callbackApex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.
