FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Aipedia: Barristers
AI can help counsel plan cross-examination and practise questioning, and the BSB treats training exercises without case details as low-risk. Using real case material, or anything that touches a witness's own evidence, is where the confidentiality and conduct risks begin.
Part of: AI in Barristers
In short
The BSB's AI guidance gives training pupils to prepare for cross-examination, without actual case details, as an example of the lowest-risk use, suitable even for free tools. Case-specific planning involves confidential material, so CD6, rC15 and the BSB's warning about free tools apply. The guidance does not address using AI with a witness's own evidence, which raises separate witness-handling questions. The Civil Justice Council has proposed a declaration that AI was not used to generate the content of trial witness statements. Errors in AI-assisted preparation usually reach PI only through their consequences.
Uses range from generic to case-specific. At the generic end, counsel and pupils use AI to role-play a hostile witness, rehearse question technique or generate practice scenarios. At the case-specific end, counsel feed in witness statements and exhibits to produce lines of questioning, spot inconsistencies between accounts or anticipate re-examination. The second kind is more useful and much more sensitive.
Counsel also use AI to search transcripts of earlier hearings, interviews or depositions for statements a witness may contradict. This is efficient but brittle: transcription errors, speaker misattribution and paraphrase can produce an apparent inconsistency that is not real. Before putting a prior statement to a witness, check the exact words in the official transcript or recording and the page reference. Putting a misquoted prior statement to a witness is unfair to the witness and can mislead the court, which engages CD1 and rC9.
In its discussion of CD6, the BSB's AI guidance gives "general training exercises (e.g. training pupils to prepare for cross-examination) that do not involve actual case details" as the kind of task where even free public tools may be acceptable. Once real names, statements or exhibits go in, the confidentiality analysis changes. The BSB says free, publicly available tools are "unlikely" to meet the confidentiality conditions, so case-specific work belongs in a tool whose terms, data location and deletion settings counsel has checked. See free AI tools and confidentiality.
The BSB guidance does not deal with using AI alongside a witness's own account. Putting a witness through AI-generated practice questions about their evidence, or using a tool to suggest how their account might be expressed, raises separate witness-handling questions that the guidance does not resolve, and counsel should treat it with real caution. The Civil Justice Council's consultation proposed that a legal representative should declare that AI was not used to generate the content of trial witness statements; the final report is expected later in 2026. That direction of travel suggests courts will scrutinise AI involvement in witness evidence closely.
AI-generated questioning can reflect bias in training data. A tool may suggest lines of attack based on a witness's accent, background or disability that would be inappropriate or discriminatory. The BSB maps this to CD8 and rC12, which prohibit unlawful discrimination, and to CD1. Counsel should review any suggested line for fairness as well as relevance.
Illustrative: counsel pastes two witness statements into a free AI chatbot to generate a cross-examination plan. The plan wrongly reports an inconsistency that exists only because the tool merged two dates. Counsel puts the point, the witness corrects it convincingly, and the judge comments on the unfair question. Separately, the statements are now held by the tool's provider on terms counsel never checked.
A poor line of questioning is rarely a claim in itself. Claims more commonly arise from consequences: a lost hearing, a complaint, or a confidentiality breach. A client's negligence claim usually falls within compulsory BMIF cover under the BSB minimum terms, which contain no AI exclusion. First-party costs of a data incident are not required by those terms. See does BMIF cover AI mistakes.
Apex arranges PI for barristers and can explain how your cover treats claims and data incidents linked to AI-assisted trial preparation. More on the AI at the Bar hub and in AI bundle review.
The BSB guidance gives training pupils for cross-examination, with no actual case details, as an example of the lowest-risk use. Keep scenarios invented and do not paste in real statements, names or exhibits.
Only into a tool whose confidentiality terms and data handling you have checked. The BSB says free public tools are unlikely to meet the confidentiality conditions, and statements are confidential client material under CD6 and rC15.
The Civil Justice Council proposed that a legal representative declare that AI was not used to generate the content of trial witness statements. The consultation closed on 14 April 2026 and the final report is expected later in 2026.
It can be. The BSB maps AI bias to CD8 and rC12, which prohibit unlawful discrimination. Review suggested questions for assumptions based on a witness's background or characteristics before using them.
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.
