Are UK law firms using AI? Most lawyers now use some AI personally, but few small firms have built it into how the firm works. Vendor surveys from 2026 put AI adoption at around nine in ten legal professionals. The figures for AI embedded in a firm's strategy and operations are much lower: 13% of small law firms against 57% of large ones in LexisNexis's mid-2026 survey. That gap is where the regulatory risk now sits, because the courts and the SRA have made firm leaders responsible for how AI is used.
This guide is for managing partners, COOs and practice managers at firms of 10 to 200 people, and for chambers' practice directors. It covers where AI is used in UK legal practice now, a task-by-task map of what the SRA, the Bar Council, the BSB, the Law Society and the judiciary expect, what the courts have said about AI-invented citations, whether AI will replace lawyers, and a 90-day plan for adopting it safely.
Key Takeaway
UK regulators allow AI but have not lowered any standard for it. Every citation must be genuine and checked, client confidentiality applies to paid tools as well as free ones, and supervisors can be found in breach for a junior's unchecked AI work. In June 2025 the High Court said managing partners and heads of chambers must take practical steps. Whitehat's reading: for a small or mid-sized firm, the first AI decision is a supervision policy, not a tool.
How many UK law firms use AI?
There is no profession-wide figure from the SRA, the Law Society, the Bar Council or the Legal Services Board. What exists is a handful of surveys, most run by companies that sell legal AI, each measuring something slightly different. Read them as a range, not a single number.
94%
UK legal professionals who have adopted AI
LexisNexis vendor survey, mid-2026
13%
Small law firms with AI embedded in strategy and operations
Same survey (57% of large firms)
14%
AI adoption among sole practitioners and small firms
SRA research, June 2025
| Source | Who was asked | Result | Type |
|---|---|---|---|
| LexisNexis, The AI-dependent lawyer (mid-2026 survey) | 543 UK legal professionals, including in-house lawyers, the Bar, public sector and academics | 94% have adopted AI; 68% of small-firm lawyers use it weekly; AI embedded at 13% of small firms and 57% of large firms | Vendor (sells Lexis+ AI) |
| Clio, UK & Ireland Legal Insights Report (Apr 2026) | More than 500 legal professionals and 500 members of the public across the UK and Ireland | 89% use AI tools "in some capacity"; 81% of firms say they tell clients at least occasionally, but only 7% of clients recall being told | Vendor (UK and Ireland combined) |
| SRA, small-firm technology research by Thinks Insight & Strategy (Jun 2025) | Around 150 solicitors at sole practices and small firms | "Moderate adoption rates (14%)" for AI | Regulator-commissioned; small sample |
| Thomson Reuters, Future of Professionals (Sep 2025) | 172 UK legal professionals within a global survey of 2,275 | 87% expect AI to have a high or transformational impact on the legal profession within five years | Vendor; international study |
The spread comes from three things. The high figures count any personal use by a person; the low ones ask whether a firm has adopted AI. The vendor samples include in-house teams and large firms, which use AI more. And none of the vendor surveys publish their question wording or fieldwork dates. For a 30-person firm, the useful reading is the gap between the two: most of your people are probably using AI already, and the firm may not have decided how.
Where is AI used in UK legal practice now?
The uses fall into five groups. Two are established and court-recognised; three are newer and carry most of the risk.
- Document review and e-disclosure. Technology-assisted review predates generative AI. The judiciary's own guidance says "TAR is now part of the landscape of approaches to electronic disclosure", and the High Court noted in 2025 that AI is used to manage large disclosure exercises in the Business and Property Courts.
- Summarising and admin. Summarising documents and emails, transcribing meetings and drafting routine correspondence. The judicial guidance lists these administrative tasks as potentially useful.
- First drafts. Letters, contract clauses, attendance notes and skeleton outlines, always reviewed by the fee-earner responsible.
- Legal research. The highest-risk use. Public chatbots invent cases that look real, and the UK misuse cases reported so far centre on research and citations.
- Client-facing tools. Intake, triage and guided processes. In May 2025 the SRA authorised Garfield.Law, the first purely AI-based firm in England and Wales, to help clients recover unpaid debts through the small claims process.
For billing and time recording, AI features increasingly sit inside practice management software, but we found no independent UK data on how widely they are used.
What UK regulators expect: a task-by-task map
No UK body has written new rules for AI. Each has explained how existing duties apply. Read together, they give a clear picture of what a firm can hand to AI and what it must keep. The map below is Whitehat's synthesis of the published guidance; the verdict column is our reading, not the regulators'.
| Task | What the guidance says | Duty that applies | Our verdict |
|---|---|---|---|
| Legal research and citations | The High Court: free generative AI tools "are not capable of conducting reliable legal research" (Ayinde, Jun 2025). The SRA: you should be assured all cited authorities "are genuine, relevant, have a verifiable citation" (Aug 2026). The judiciary: AI is "a poor way of conducting research to find new information you cannot verify independently" (Oct 2025). | Check every authority against an authoritative source before it is used | Assist only, with recorded verification |
| Drafting letters, contracts and submissions | The SRA: solicitors using AI "remain accountable for their work and outputs" (Aug 2026). The Bar Council: "The ultimate responsibility for all legal work remains with the barrister" (Nov 2025). The Law Society: review outputs "for accuracy and factual correctness". | Personal responsibility for the finished work | Assist: first drafts, reviewed by the responsible fee-earner |
| Document review and e-disclosure | The judiciary: "TAR is now part of the landscape of approaches to electronic disclosure" (Oct 2025). | Disclosure duties and supervision of the process | Automate, with a documented, supervised process |
| Anything involving client data | The SRA: client information only where "appropriate contractual, technical and organisational safeguards are in place", and "Both free to use and paid for AI systems may pose risks" (Aug 2026). The judiciary: anything put into a public chatbot "should be seen as being published to all the world" (Oct 2025). | Confidentiality, privilege and data protection | Approved tools only, with contracts checked |
| Client-facing tools and intake | The SRA: "It should always be made clear to clients where they are interfacing with AI" (compliance tips, Feb 2026). When it authorised Garfield.Law, the SRA noted the system cannot propose case law and only takes a step "where the client has approved it" (May 2025). | Transparency; named solicitors stay accountable | Assist, with clear disclosure |
| Supervising others' AI use | The SRA: supervisors of junior or non-authorised colleagues "may also be found to have breached regulatory requirements" (Aug 2026). The High Court: leaders such as managing partners and heads of chambers must take practical measures (Jun 2025). | Supervision and firm governance | Keep with a named partner |
| Telling clients and the court | The SRA's 2023 Risk Outlook: "Tell clients when you will be using AI with their case". The BSB: be transparent about AI use, "especially if asked by your client, the court, or the BSB" (May 2026). The judiciary: no need to mention responsible use to the court, "but this is dependent upon context" (Oct 2025). | Honesty and client information | Say so in your client care letter |
Two sources matter most for a firm's policy. The SRA's warning notice on the misuse of AI (17 August 2026) states the duties above and warns that failing to have proper regard to it puts solicitors "at risk of disciplinary action". For chambers, the BSB's guidance on AI and other technologies (valid from 18 May 2026) is the regulator's text; the Bar Council's note is professional guidance rather than BSB guidance.
What have UK courts said about AI hallucinations?
The pattern across the reported cases is the same: AI-generated research went into a court document and nobody checked it. The consequences have grown from embarrassment to regulatory referral.
- Harber v HMRC, First-tier Tribunal (December 2023). A litigant in person relied on nine tribunal decisions that "had instead been generated by artificial intelligence". It was an early warning, not a lawyer's error.
- Ayinde v Haringey and Al-Haroun v Qatar National Bank, High Court (6 June 2025). In Al-Haroun, 18 of 45 citations put before the court did not exist. The Divisional Court referred lawyers to their regulators, said referral is likely to be appropriate when false citations reach a court, and said guidance on its own "is insufficient to address the misuse of artificial intelligence".
- UK v SSHD, Upper Tribunal (November 2025). The tribunal said uploading confidential documents into an open-source AI tool such as ChatGPT places them in the public domain, and that supervisors are likely to be more culpable.
- Cork v Smith, High Court (May 2026). The court recorded that "a junior associate had used AI to research the point and the AI invented the quotation", and that neither the associate nor the supervising solicitors checked it.
The SRA's August 2026 warning notice says it acted after receiving reports of potential breaches "from senior members of the judiciary". For a firm, the practical lesson is that a supervision failure is now treated as seriously as the original error.
What we'd tell a managing partner: three questions, answered from the evidence
What are UK firms actually doing with AI? Using it personally far more than they have governed it. On LexisNexis's 2026 figures, 68% of small-firm lawyers use AI weekly, but only 13% of small firms have built it into strategy and operations. Clio's 2026 survey found 81% of firms say they tell clients about AI at least occasionally, while only 7% of clients recall being told. The work is happening; the policy, the client wording and the supervision often are not.
What is the most common mistake? Treating AI as an IT purchase rather than a supervision question. Each of the cases above turned on AI output nobody checked, and the courts and the SRA now look at who should have checked it. Paying for a tool does not solve this: the SRA says paid systems can also put confidentiality at risk.
What should a firm do first? Not buy a tool. In order:
- Name an owner. The SRA expects, as a minimum, the COLP to be responsible for regulatory compliance when new technology comes in.
- Write a one-page AI policy: permitted uses, prohibited data, and a rule that every authority is checked against a primary source, with the check recorded.
- Approve specific tools for specific tasks, after checking the contractual safeguards for client data.
- Train everyone before they get access, including trainees, paralegals and support staff, because supervisors carry the risk.
- Update your client care letter to say how you use AI, and ask your insurance broker in writing whether your planned use affects your cover.
- Then pilot one workflow, measure the time it saves, and widen it only once the checks work.
Whitehat's AI practice helps firms set up exactly this governance. See our AI consultancy and implementation service.
What are the risks of AI for a law firm?
- Invented authorities. The risk the courts have dealt with most. It ends in wasted costs, regulatory referral and reputational damage.
- Confidentiality and privilege. Client material entered into a tool without the right safeguards. The Bar Council tells barristers to "be extremely vigilant" about sharing privileged or confidential information.
- Unmanaged use. The SRA's 2023 Risk Outlook warned that "staff might still want to use ChatGPT or similar systems. You will need clear rules covering this."
- Supervision failures. A trainee's unchecked output becomes the supervisor's regulatory problem.
- Client expectations. Clients who were not told, or who expect AI to lower fees, may complain.
- Data protection. The ICO's guidance on AI and data protection is under review following the Data (Use and Access) Act, so check it before relying on it.
These risks are manageable. The SRA's Risk Outlook also made the opposite point: "The risk to firms might not come from adopting AI, but from failing to do so."
Will AI replace lawyers?
Not on the current evidence. It will change what lawyers and their support teams spend time on. The Department for Education's 2023 analysis ranked solicitors second among UK occupations for exposure to large language models. The report says it measures exposure, not whether a job "will be augmented (aided) or replaced (substituted) by AI", and solicitors are not on its list of high-automation occupations. IPPR's 2024 analysis of the whole economy found back-office, secretarial and entry-level roles most exposed, which matters for legal support staff more than for qualified lawyers.
Garfield.Law shows where the line currently sits. It is an AI-based firm, but the SRA authorised it for a narrow task, with no case-law suggestions, client approval for every step and named solicitors accountable. For a typical firm, the likely change is in the mix of work: less first-draft research and document handling, more checking, judgment and client advice. That has a training consequence, because trainees learn on exactly the tasks AI now does first.
The same pattern appears in other professions. Our guide to AI in accounting covers how UK accountancy firms are handling it.
A 90-day AI plan for a UK law firm
This plan suits a firm of 10 to 200 people or a set of chambers. It puts governance first because that is what the regulators ask for, and because it makes every later step faster.
Days 1–30: find out what is already happening
Ask every team which AI tools they use, for what, and with which client data. Name the owner (usually the COLP). Write a one-page policy covering approved tools, prohibited data and the citation-checking rule. Read the SRA warning notice, or the BSB guidance for chambers, against it.
Days 31–60: set the controls and train
Check the contractual and data safeguards of the tools you plan to approve. Train everyone before giving access, with supervisors trained on what to check. Update your client care letter, and speak to your insurer. Choose one pilot workflow, such as summarising long documents or first-draft correspondence.
Days 61–90: pilot, measure and decide
Run the pilot with recorded checks. Sample outputs for accuracy, measure the time saved, and report to partners. Widen use only where the checks held up, and review the policy against any new guidance.
If AI frees time, the next question is how to fill it with the right work. Our guide to law firm marketing covers how UK firms win more of the clients they want.
Frequently asked questions about AI in law firms
How many UK law firms use AI?
There is no official profession-wide figure. Vendor surveys in 2026 put AI adoption at 89–94% of legal professionals, while AI built into firm strategy and operations was reported at 13% of small firms and 57% of large firms. The SRA's 2025 research found 14% adoption among sole practitioners and small firms.
Can solicitors use ChatGPT?
Yes. The SRA lets firms use AI provided they meet its standards. Solicitors remain accountable for the output, must check that every cited authority is genuine, and should only enter client information where proper safeguards are in place. The judiciary's guidance treats anything typed into a public chatbot as published to the world.
Do law firms have to tell clients they use AI?
The SRA's 2023 Risk Outlook advises telling clients when you will use AI with their case, and its compliance tips say clients should know when they are dealing with AI. The SRA's 2026 warning notice does not set an express disclosure duty. The BSB expects barristers to be transparent, especially when asked. Saying it in your client care letter is the simplest approach.
Is there an SRA rule on AI?
There is no new rule. Existing standards apply. The SRA published a warning notice on the misuse of AI on 17 August 2026, which explains how its standards apply and warns of disciplinary action for solicitors who ignore it.
Will AI replace lawyers?
The evidence points to changing tasks, not replacing lawyers. Solicitors are highly exposed to large language models, but exposure is not replacement. Research, drafting and document handling change first; judgment, advice and accountability stay with qualified people.
Want more of the right clients for your firm?
We help UK law firms get found for the work they want, in Google and in AI answers. See how our SEO for solicitors works, or tell us about your firm.
Sources: SRA warning notice, Misuse of AI (17 Aug 2026); SRA compliance tips (updated 9 Feb 2026); SRA Risk Outlook on AI (20 Nov 2023); SRA, Garfield.Law authorised (6 May 2025); SRA small-firms research (19 Jun 2025); Bar Council guidance (reviewed Nov 2025); BSB guidance (May 2026); Law Society, Generative AI: the essentials; Judicial AI guidance (31 Oct 2025); Ayinde and Al-Haroun [2025] EWHC 1383 (Admin); Harber v HMRC [2023] UKFTT 1007 (TC); UK v SSHD [2026] UKUT 81 (IAC); Cork v Smith [2026] EWHC 1199 (Ch); LexisNexis, The AI-dependent lawyer (2026); Clio UK & Ireland Legal Insights Report (7 Apr 2026); Thomson Reuters, Future of Professionals (4 Sep 2025); DfE, The impact of AI on UK jobs and training (28 Nov 2023); IPPR (27 Mar 2024); ICO guidance on AI and data protection (under review).
Clwyd Probert
Founder, Whitehat
Clwyd Probert is the founder of Whitehat, a London-based SEO, AI and inbound marketing agency and HubSpot Diamond Partner. He works with B2B and professional-services firms on AI adoption and growth.