Last updated: 9 November 2026. This guide is based on the Solicitors Regulation Authority’s warning notice “Misuse of AI” and news release of 17 August 2026, its effective supervision guidance as updated on 12 June 2026, its May 2025 authorisation of Garfield.Law, the SRA Code of Conduct, the Divisional Court’s judgment in Ayinde and Al-Haroun [2025] EWHC 1383 (Admin), guidance from the Law Society and the Bar Council, the Civil Justice Council’s October 2026 final report on AI and court documents, and ABA Formal Opinion 512.
Short answer: The SRA has not written new AI rules. Its position is that existing duties apply whatever tool a solicitor uses: competence, supervision, confidentiality and never misleading the court. In August 2026 it turned that into a formal warning notice on fake citations and client data, backed by the threat of disciplinary action, after receiving 42 reports of possible AI misuse in a year.
Key takeaways
- On 17 August 2026 the SRA published a warning notice, “Misuse of AI”, aimed at two risks: fictitious cases or citations reaching the court, and client information being entered into AI tools without safeguards.
- The notice says “AI has no separate legal personality”, so the solicitor stays accountable for every output. Ignoring the notice puts solicitors “at risk of disciplinary action”.
- In June 2026 the SRA added an AI section to its supervision guidance: AI-assisted work needs “appropriate human review, scrutiny and professional judgement”.
- The SRA still supports AI use. It authorised Garfield.Law, the first AI-driven law firm, in May 2025 and joined the government’s Advisory AI Growth Lab in 2026.
Does the SRA have specific rules on AI for solicitors?
No. There is no AI chapter in the SRA Standards and Regulations, and the SRA has said it takes an outcomes-focused approach rather than prescribing particular tools or methods. Instead it applies the existing SRA Principles and the two Codes of Conduct, one for individual solicitors and one for firms, to AI in the same way it applies them to any other way of working.
What has changed is how specific the SRA has become since its November 2023 Risk Outlook report on AI. It has:
- authorised the first law firm built around an AI system (May 2025);
- updated its effective supervision guidance on 12 June 2026 with a section headed “Supervision and Artificial Intelligence (AI)”;
- joined the government-funded Advisory AI Growth Lab, launched for legal services on 5 August 2026, alongside the Legal Services Board, the Council for Licensed Conveyancers and the Information Commissioner’s Office;
- issued a formal warning notice on the misuse of AI on 17 August 2026.
The supervision guidance makes two points that matter for every firm. Firms “should consider what effective supervision of AI assisted or AI generated work looks like in practice”, and “an authorised individual retains ultimate responsibility for any legal services delivered with AI assistance.”
What does the SRA’s 2026 warning notice on AI say?
The warning notice applies to all firms and individuals the SRA regulates. Warning notices are not new rules, but they show how the SRA reads its existing rules.
The notice deals with two problems.
- Inaccurate court documents and citations. AI tools can produce “fictitious cases, references or seemingly factual assertions.” Before any named case reaches a court, the solicitor must make sure it is genuine, relevant and has a verifiable citation. The notice quotes the Ayinde judgment: reliance on an AI output “would not be a suitable defence.”
- Client confidentiality. Using public AI tools with client information “will likely breach client confidentiality”, and legal professional privilege “may be permanently waived and unable to be recovered.” Client data should stay in a secure environment, should not be used to train models without authority, and should not be kept longer than necessary.
The notice also cites 2026 decisions, including UK v Secretary of State for the Home Department [2026] UKUT 81 (IAC), where the Upper Tribunal said that putting client letters into an open AI tool “is to place this information on the internet in the public domain”.
In its news release the SRA said it received 42 reports of potential AI misuse between July 2025 and July 2026, and that open investigations involve inaccurate citations, supervision and confidentiality. Aileen Armstrong, its Executive Director of Strategy and Policy, said “the use of AI does not change the professional standards expected of solicitors and law firms.”
Which Code of Conduct duties apply when solicitors use AI?
The warning notice lists SRA Principles 1, 2, 4, 5 and 7 (the rule of law, public trust, honesty, integrity and clients’ best interests) and a long list of Code paragraphs. These are the ones that do most of the work.
| Duty | Where it sits | What it means for AI use |
|---|---|---|
| Not misleading the court or others | Code for Solicitors 1.4 | Filing an invented case or a false quotation can mislead the court even if no one intended it |
| Only properly arguable submissions | Code for Solicitors 2.4 | Submissions built on unchecked AI research may not be properly arguable |
| Drawing the court’s attention to relevant law | Code for Solicitors 2.7 | You must know the real authorities, including those against your client |
| Competent service | Code for Solicitors 3.2; Code for Firms 4.2 | Understanding a tool’s limits, and checking its output, is part of competence |
| Supervision | Code for Solicitors 3.5; Code for Firms 4.4 | Supervisors “remain accountable for the work carried out through them”, including AI-assisted work by juniors and support staff |
| Confidentiality | Code for Solicitors 6.3; Code for Firms 6.3 | Do not put client data into tools whose terms allow storage, reuse or training without safeguards |
| Governance and controls | Code for Firms 2.1 | Firms need systems covering AI risk, staff training and approved tools |
None of these duties is new; as the notice puts it, “the use of AI does not diminish or transfer your professional responsibilities.” For a practical way to set this out internally, see our guide to writing a law firm AI policy.
What did the High Court say in Ayinde and Al-Haroun?
The judgment that most shaped the SRA’s stance came from the Divisional Court of the King’s Bench Division on 6 June 2025. Dame Victoria Sharp, President of the King’s Bench Division, and Mr Justice Johnson heard two referred cases together under the court’s Hamid jurisdiction, which it uses to deal with lawyers who may have breached their duties to the court.
- In Ayinde v Haringey, a housing judicial review, the grounds cited five cases that did not exist. The court found the contempt threshold met for the barrister who signed them but chose not to start contempt proceedings, and referred her to the Bar Standards Board. It referred the solicitor at Haringey Law Centre to the SRA. In separate wasted costs proceedings, Ritchie J ordered the barrister and the law centre to pay £2,000 each.
- In Al-Haroun v Qatar National Bank, a schedule listed 45 citations, 18 of which did not exist, and many real cases did not say what was claimed. The solicitor had relied on research by his client. The court said “a lawyer is not entitled to rely on their lay client for the accuracy of citations of authority” and referred him to the SRA.
Two passages are now quoted in almost every piece of UK guidance. The court said “freely available generative artificial intelligence tools… are not capable of conducting reliable legal research”, and that lawyers “have a professional duty therefore to check the accuracy of such research by reference to authoritative sources.” It also said that guidance on its own “is insufficient”, that heads of chambers and managing partners have leadership responsibilities, and that “the profession can expect the court to inquire whether those leadership responsibilities have been fulfilled.” The full judgment is on the National Archives.
Our table of AI hallucination sanctions in 2026 shows how courts elsewhere have punished fake citations.
How did the SRA approach Garfield.Law, the first AI-driven law firm?
On 6 May 2025 the SRA authorised Garfield.Law, an AI-powered litigation assistant that helps small and medium-sized businesses recover unpaid debts through the small claims process. Its then chief executive, Paul Philip, called it “a landmark moment for legal services in this country.”
The conditions show what the SRA looks for in any AI-heavy service:
- the system cannot propose case law, which the SRA called a high-risk area for large language models;
- it is not autonomous and takes each step only with the client’s approval;
- named regulated solicitors remain ultimately accountable for its output;
- it has quality, confidentiality and conflict controls and the minimum insurance any regulated firm needs.
The SRA said it expected this to be “the first of many AI-driven law firms” and would monitor it closely. In August 2026 its current chief executive, Sarah Rapson, said innovation “has an important role to play in improving legal services” when announcing the Advisory AI Growth Lab, which was taking applications until 27 September 2026.
What do the Law Society, the Bar Council and the courts add?
- Law Society. Its guide Generative AI: the essentials, updated in September 2025 to cover Ayinde and Al-Haroun, advises solicitors to check outputs against authoritative sources and, for free tools with no contract with the vendor, to “not put any confidential data into the tool.” It also suggests agreeing with clients whether AI will be used on their matters.
- Bar Council. On 26 November 2025 it updated its guidance on ChatGPT and other large language models, adding legal-specific tools, recent case law and cyber security risks. It says responsibility for legal work stays with the barrister.
- Civil Justice Council. Its final report on AI and court documents, published in late October 2026, recommends no new declaration for lawyers’ statements of case or skeleton arguments, because existing duties already make them responsible. It does recommend declarations from witnesses of fact and experts where AI has generated or influenced the substance of their evidence, and a Practice Direction on AI misuse. These are recommendations for the Civil Procedure Rule Committee and are not yet in force.
How does the SRA’s position compare with ABA Formal Opinion 512?
The American Bar Association published Formal Opinion 512 on 29 July 2024. Like the SRA, it applies existing rules, and the conclusions are close, but the ABA goes further on client consent and fees.
| Issue | SRA (England and Wales) | ABA Formal Opinion 512 (US) |
|---|---|---|
| Legal basis | SRA Principles and Codes of Conduct, plus a 2026 warning notice | ABA Model Rules (advisory; each state adopts its own rules) |
| Checking output | Named cases must be genuine and verifiable before reaching the court | Competence and candour duties require review of output |
| Confidentiality | Public tools with client data “will likely breach” confidentiality; privilege may be lost | Informed client consent needed before inputting client information into self-learning tools; boilerplate consent is not enough |
| Supervision | Supervisors accountable for AI-assisted work of juniors and staff | Managerial and supervisory duties extend to AI use |
| Fees | Not addressed in the warning notice | Lawyers should not bill clients for time spent learning a tool |
| Enforcement | Disciplinary action; court referrals to the SRA | State bar discipline; court sanctions |
Our explainer on ABA Formal Opinion 512 covers the US opinion rule by rule, and our guide to the EU AI Act for law firms covers firms with clients or offices in the EU.
What should a law firm in England and Wales do now?
- Approve tools. Check vendors’ terms on data storage, retention and training, and restrict public chatbots for client work.
- Make citation checks mandatory. A named person checks every authority in a court document against an authoritative source before filing.
- Write AI into supervision. Supervisors should know when juniors and support staff have used AI and review that work.
- Train people. The warning notice expects firms to have staff competence and training covering AI risk.
- Talk to clients. Agree whether and how AI will be used on their matters, as the Law Society suggests.
- Keep records. A written policy, tools list and training records help show the firm met its governance duties.
If you are choosing tools, our directory of legal tech tools and our guide to AI for lawyers compare the main research, drafting and review products.
Frequently asked questions
Can solicitors in England and Wales use ChatGPT or other AI tools?
Yes. The SRA does not ban any tool. Solicitors must keep to their existing duties: check outputs, supervise AI-assisted work, protect client confidentiality and never mislead the court. The SRA warns that putting client information into public AI tools will likely breach confidentiality.
What is the SRA’s warning notice on the misuse of AI?
Published on 17 August 2026, it tells all SRA-regulated firms and individuals how the SRA expects existing rules to apply to AI. It focuses on fake citations in court documents and on client confidentiality, and warns that failing to have regard to it risks disciplinary action.
Can a solicitor blame AI for a fake case citation?
No. The SRA says “AI has no separate legal personality” and the solicitor remains accountable. Quoting the Ayinde judgment, its warning notice says reliance on an AI output “would not be a suitable defence” where false citations reach a court.
What happened in Ayinde v Haringey?
In June 2025 the Divisional Court found that grounds for judicial review cited five non-existent cases. It referred the barrister to the Bar Standards Board and the solicitor to the SRA, and warned that it would ask whether firm and chambers leaders had met their responsibilities in future cases.
Is Garfield.Law regulated by the SRA?
Yes. The SRA authorised Garfield.Law in May 2025 as the first AI-driven law firm. Its system helps businesses recover debts through small claims, cannot propose case law, acts only with client approval, and named regulated solicitors remain accountable for its work.
Do lawyers have to tell the court they used AI?
Not under current Civil Procedure Rules. The Civil Justice Council’s October 2026 final report recommends no new declaration for lawyers’ statements of case or skeleton arguments, but does recommend declarations from witnesses and experts. Those proposals are not yet in force.
This article is general information about the regulation of solicitors in England and Wales as of 9 November 2026. It is not legal or regulatory advice. Check the SRA’s current guidance and warning notices before relying on any point.
Where the legal industry reads first.
Enjoyed this article? Get the biggest legal industry updates, deals, appointments, insights and expert interviews in your inbox, free.
No spam. Unsubscribe anytime.From Legal Desire
Is your firm the one being cited, or the one being skipped?
We ran the test. On several everyday legal questions, software companies are answering and law firms are not. We help firms fix that, and we build the technology behind the practice.
Lawyers and firms: get featured on Legal Desire
Share your deals, lateral moves, firm news and insights with 1M+ monthly readers: founders, law firm partners, general counsel and practising lawyers across the US, UK, EU and GCC.
Get featuredSubmit a press release
