Most of the worry about AI in a law firm runs one way, towards the risks of using it rather than the negligence of not using it. Partners ask what happens if a tool invents a case, leaks a client's file or produces advice that turns out to be wrong. A legal statement published on 7 July 2026 by the UK Jurisdiction Taskforce asks the question from the other side as well. If a competent solicitor in your position would have used an AI tool on a matter, and you did not, that failure can itself amount to a breach of your duty of care.
What the taskforce published
The UK Jurisdiction Taskforce sits within LawtechUK and is chaired by Sir Geoffrey Vos, the Master of the Rolls. Its Legal Statement on Liability for AI Harms followed a public consultation that ran from 14 January to 13 February 2026, with a drafting committee chaired by Matthew Lavy KC. The statement works across negligence, vicarious liability, professional liability, product liability and the law of false statements, and it arrives at a settled conclusion. English private law, as it stands, can resolve most disputes about harm caused by AI, and no AI-specific liability regime is needed to do it. Sir Geoffrey described the exercise as providing "much needed legal certainty and legal predictability" in an area where few cases have yet reached the courts. Two points of detail carry weight for smaller firms. A business cannot escape responsibility on the ground that a chatbot produced the false information, and the developer of a general-purpose model will not usually answer for downstream uses it had no way of foreseeing. Responsibility tends to settle on the person who chose to deploy the tool.
The duty runs in both directions
For professionals the statement is blunt. Your obligation to exercise reasonable skill and care applies to your use of AI exactly as it applies to every other part of your work. A solicitor can be negligent for using AI badly. Selecting an unsuitable model, skipping any real diligence on the tool, failing to explain to a client how AI was used on their matter, or sending out work without checking the output for error and bias would each qualify. The taskforce then says the same duty can bite the other way, and that a professional can be liable for failing to use AI where a competent member of the profession would have done so. Neither proposition is new law. What has changed is that a body of this standing has written the second one down, in terms a claimant's solicitor will read.
How the standard gets set
The test is the ordinary one. Whether a failure to use AI breaches the duty falls to be judged against what a reasonable professional of comparable seniority and specialism would have done in the same position. That standard does not stand still. It moves as tools become unremarkable within a field, which is why the statement bears on your firm's next few years rather than on this afternoon. A solicitor who reviews eight boxes of disclosure by hand today is unlikely to be criticised for it. The same choice in three years, once clients and opponents assume the first pass was done with software, invites a harder question about cost and delay. The duty to advise comes into view alongside it. Where a tool would materially reduce what a piece of work costs the client, saying nothing about it is a decision your firm should be able to defend.
The other risk has not gone away
None of this softens the duty to check, and the week supplied its own illustration. On 10 July, Legal Futures reported that the Crown Prosecution Service had acknowledged putting cases that do not exist before the High Court in an extradition appeal, the fabricated authorities having come from AI. The service told the court that the failure lay in not verifying the authorities relied on in formal submissions. The two stories belong together. A firm that stays away from AI carries one kind of exposure, and a firm that trusts it without verification carries another, sharper and closer to hand. The taskforce has now put both on the same map.
The practical answer is a written record of where you stand. Take the categories of work your firm does most often. For each one, note whether a tool exists that a competent practitioner would now reach for, whether you have tested it, and why you have adopted it or turned it down. That note costs an afternoon of somebody's time. It answers a negligence claim from either direction, because it shows the choice was made rather than avoided. Set it beside the checking discipline you already apply to a trainee's first draft, and the two risks the taskforce identified stop pulling against one another.
The SRA's Risk Outlook on AI in the legal market weighs the same balance of risk in both directions.
If you have neither adopted these tools nor recorded why not, the note that protects you takes a day to build with us: start with a conversation.