The AI leadership challenge in law has become one of the biggest issues facing UK law firms. Most firms already use artificial intelligence in some form. The debate is no longer about adoption but accountability. When AI makes a mistake, the SRA’s position remains clear: the solicitor is responsible. That makes AI a leadership and governance challenge before it is a technology issue.
AI is now embedded in mainstream legal work, from document review and due diligence to research and drafting. There is no separate SRA rulebook for it, but the existing duties, competence, confidentiality, and accountability, apply in full. The SRA is explicit that a solicitor remains responsible for the outputs of AI they use and cannot hand that accountability to a vendor or an IT team. For firm leaders the task is to put governance around AI: approved tools, mandatory human review, client transparency, data controls, training, and records, so the gains arrive without the regulatory and negligence risk.
How UK law firms are using AI
Adoption is no longer an early-adopter story. In its November 2023 risk outlook on AI in the legal market, the SRA reported that three-quarters of the largest firms were already using AI by the end of 2022, and that more than 60% of large firms were exploring generative systems. A later LexisNexis survey put generative-AI use among lawyers at 61% by September 2025, up from 46% at the start of that year, with the share of firms reporting no plans to adopt falling to single figures.
The everyday uses are concrete. Document review and due diligence that once took days now run in a fraction of the time. Research tools summarise authorities and surface relevant clauses. Drafting assistants produce first cuts of letters, contracts and memos. None of this is the point of tension. The tension sits in what happens when one of those outputs is wrong and a client, or a court, relies on it.
The AI Leadership Challenge in Law: Why It Is a Leadership Problem, Not an IT One
The instinct in many firms is to treat AI as a procurement decision: choose a tool, roll it out, let the technology team own it. That misreads where the regulatory risk lands. The SRA’s position is that accountability travels with the solicitor, not the software.
That line, from the SRA’s risk outlook, is the one firm leaders need to internalise. The regulator adds that a firm cannot delegate accountability to an IT team or an external provider, and that AI should support rather than replace human judgment. In practice, that turns questions people think of as technical, which tool, which settings, who checks the output, into governance decisions that partners and compliance officers own. If a system is configured so that unreviewed AI output can reach a client, someone chose that configuration, and under the rules that choice is the firm’s.
The risks the regulators name
The SRA and the Law Society have both set out, in plain terms, what goes wrong. Four risks recur.
| Risk | What it looks like in practice | Named by |
|---|---|---|
| Hallucination | AI produces plausible but false output; drafted arguments citing cases that do not exist | SRA, Law Society |
| Confidentiality | Staff paste client data into public tools such as free chatbots, with no vendor relationship | SRA, Law Society |
| Bias | Models reflect or amplify biases in their training data, producing unfair outcomes | SRA, Law Society |
| Accountability | Responsibility for an AI-assisted error still sits with the solicitor and the firm | SRA |
The bias risk is not abstract. The SRA’s report cites lending algorithms that were markedly more likely to reject applicants from minority ethnic backgrounds because they had learned from historically skewed data. In legal work the same mechanism can shape outcomes quietly, inside a process that looks neutral.
The consequences of getting it wrong have hardened. The Law Society’s guidance notes that the courts now treat the misuse of AI as serious enough that admonishment alone will rarely be sufficient, with sanctions running up to strike-out of a case or a referral for contempt. A fabricated citation is no longer an embarrassment. It is a professional-conduct and litigation exposure.
What the rules actually require
There is no AI-specific section of the SRA Standards and Regulations, and firms sometimes read that absence as a grey area. It is not one.
“There are no AI rules yet” is a common and risky misreading. The SRA has not written a separate AI rulebook, but that does not mean AI use is unregulated.
Your existing obligations apply in full: to act with competence, to keep client information confidential, to uphold the proper administration of justice, and to remain accountable for your work. The Law Society is clear that a solicitor’s professional duties apply to work carried out with AI exactly as they apply to work carried out without it. The tool is new. The duties are not.
Supervision is where this now bites hardest. The SRA’s revised guidance on effective supervision, published in June 2026, states that where a firm uses AI to deliver legal services the outputs need appropriate human review and an authorised individual retains ultimate responsibility. File review, escalation and sign-off for AI-assisted work are no longer optional good practice; they are part of how a firm evidences supervision. We covered that shift in detail in our guide to what your practice management systems must prove in 2026.
An AI Governance Framework for the AI Leadership Challenge in Law
The firms handling this well are not the ones with the most tools. They are the ones that decided, in advance, how AI may and may not be used, and wrote it down.
Six things an AI governance framework should cover
None of this requires a technology background. It requires decisions, made by the people who carry the accountability.
- An approved-tools list, so fee earners know which systems are sanctioned and which are off-limits, especially free public tools.
- A hard rule against putting confidential or personal client data into any tool the firm has not vetted.
- Mandatory human review of AI output before it reaches a client or a court, with the reviewer named.
- A position on client transparency: when, and how, you tell clients that AI was used in their matter.
- Training that covers the risks, not just the features, so staff can recognise a hallucination or a confidentiality breach.
- Records of what was generated, who reviewed it and when, so the firm can evidence oversight if asked.
What good AI leadership looks like
Leadership on AI is not a stance for or against the technology. It is the willingness to own the trade-off. A firm that bans AI outright loses ground on cost and speed and often just drives usage underground, onto personal accounts the firm cannot see. A firm that adopts without governance takes on risk it has not priced. The job of leadership is to sit between those, deciding which work AI touches, under what controls, and who signs off.
That is a partner and compliance-officer responsibility, not a delegated one. It means being able to answer three questions at any point: which tools are in use, what they are used for, and how the firm knows the output was checked. A firm that cannot answer those is not leading its AI adoption. It is hoping.
Frequently asked questions
Does the SRA regulate the use of AI by solicitors?
Can solicitors use ChatGPT or other public AI tools?
Who is responsible if AI produces a mistake in legal work?
What are the main risks of AI for law firms?
How many UK law firms use AI?
Do you have to tell clients you used AI?
What to do next
If your firm has no written AI position, that is the first job, and it is a leadership job, not an IT one. Decide which tools are approved, draw the confidentiality line, and require a named human review before AI-assisted work leaves the building. Then check it against how your supervision is actually evidenced, because from 2026 the SRA expects that link to be demonstrable. The firms that treat AI as a governance question will get the speed without the exposure. The ones that treat it as a gadget will find out, at the worst possible moment, who was accountable all along.
This article describes the regulatory position on AI use by solicitors in England and Wales as at 16 July 2026. It is general information, not legal advice. Regulatory guidance on AI is developing; confirm the current SRA and Law Society position before setting firm policy.