Blog
Does our professional indemnity insurance cover work done with AI?
In most cases yes, because your policy responds to the legal work you deliver, not the tools you used. The risk is not a gap in cover but a claim you could have avoided, and a renewal form you answered carelessly.
Alesis · · 5 min read
In most cases your professional indemnity cover responds to the advice and the work, not to the software behind it. If a fee earner gets something wrong and a client suffers loss, the claim is about the negligent work; whether a research tool, a precedent bank or an AI assistant was involved along the way is a question about how the error happened, not whether it is covered. The real exposure sits elsewhere: in renewal declarations you answer without thinking, in supervision that has not kept pace, and in the kinds of mistakes that are easy to make and hard to spot.
Why cover normally follows the work, not the tool
Solicitors' professional indemnity insurance in England and Wales is written to minimum terms set by the Solicitors Regulation Authority, and the point of those terms is to protect clients. Cover is framed broadly around civil liability arising from private legal practice. It does not usually list approved tools, and no one has ever suggested a firm loses cover for using a word processor, a case management system or an online research service.
An AI assistant sits in the same category. The output is not advice until a qualified person adopts it. If a solicitor signs off a letter containing an error, that is a solicitor's error, whatever produced the first draft.
That said, policies vary above the minimum terms, and top-up layers vary more. If you want certainty, read your own wording and ask your broker in writing. Two things are worth checking specifically:
- whether any exclusion or condition touches on automated processing, technology services, or work performed outside the firm
- whether your cyber policy, if you hold one separately, overlaps or conflicts with the professional indemnity policy on data incidents
Answer the renewal proposal form honestly
The more likely problem is not exclusion but non-disclosure. Proposal forms increasingly ask about the technology a firm uses, the volume of work handled without direct supervision, and the firm's risk controls. Some ask about AI directly. Some ask questions that touch on it without using the word.
Answer carefully. A vague or over-confident answer given in a hurry can cause more trouble than the underlying practice ever would. Before renewal, get a straight internal account of:
- which tools are actually in use, including anything staff signed up for themselves
- what those tools are used for, and what they are not used for
- who reviews output before it leaves the firm, and how that review is recorded
- what the firm tells clients, and where that appears in engagement letters
If your answer is that use is informal and undocumented, fix that before you write it down.
The risks that are genuinely different
Three failure modes deserve attention, because they are easier to walk into than the traditional ones.
The first is confident invention. A tool that produces fluent text can produce a case, a section number or a figure that does not exist, and it will look exactly like the ones that do. A fee earner under time pressure may not check. The defence against this is not insurance, it is verification against the source, every time, with the source open in front of you.
The second is confidentiality. Putting client material into a service whose terms you have not read, or whose data handling you cannot describe, is a data protection problem before it is an insurance problem. The Information Commissioner's Office expects firms to know where personal data goes and on what basis.
The third is silent scope creep. A tool bought for one purpose spreads to others because it is convenient. Summarising correspondence becomes drafting advice. Nobody decided this; it just happened. Insurers ask about controls, and the honest answer needs to reflect the firm as it is.
Practical steps before your next renewal
You do not need a lengthy policy document. You need something a supervising partner can enforce and a broker can read.
- Write a short statement of what AI tools the firm permits, for what tasks, and what is prohibited.
- Require that anything cited is checked against the official source before it goes anywhere near a client or the court.
- Record who reviewed and approved work, in the same way you record any other file review.
- Keep a note of what each tool is and where the firm's data sits, so you can answer a proposal question or a client due diligence question without scrambling.
- Tell your broker what you are doing. A firm that can describe its controls calmly is in a better position than one that says nothing.
If a client asks whether their matter is insured, the answer is about the firm's cover and the firm's work. That does not change because a tool helped.
Where Alesis fits
Alesis prepares drafts for a qualified person to review and sign off; it does not file, serve or email anything, and it does not provide legal advice. Every point it makes names its source, official sources open in one click, and when it cannot find support for a point it says what is missing instead of guessing. The firm's information is held and processed in the UK, kept apart from every other firm, and never used to train anything for anyone else, which makes the confidentiality question easier to answer on a proposal form.
