SRA Warning Notice on the Misuse of AI: What Solicitors Need to Know

On 17 August 2026 the Solicitors Regulation Authority (SRA) published a warning notice titled "Misuse of AI". It applies to all firms and individuals the SRA regulates and is likely to be relevant to other authorised persons working in SRA-regulated firms. The SRA says it will have regard to the notice when exercising its regulatory functions. This guide summarises what the notice says, based on the SRA’s own page. It is general information and not legal advice. For practical drafting and review habits, see our guide to AI-assisted drafting and document review for solicitors.

What the notice is for

The notice helps firms understand obligations that already exist. It does not present itself as creating new rules, and the SRA’s page says that if you fail to have proper regard to the warning notice you are at risk of disciplinary action. The SRA also says it can and will act where it finds evidence of rule breaches, and it points readers to its Enforcement Strategy.

The risks the SRA highlights

The notice focuses on four areas. First, hallucinated citations: AI can generate fictitious cases or references, and court documents containing false or incorrect information are a primary concern. Second, confidentiality: entering client information into public AI tools may breach confidentiality, data protection requirements and legal professional privilege, and both free and paid tools may lack adequate safeguards. Third, supervision: supervisors remain accountable for work done through junior or non-authorised staff, and firms need effective governance, systems and controls. Fourth, client care: inaccurate AI-generated information can harm clients and lead to a poor outcome in their matter.

The cases the notice refers to

The notice cites several decisions. In R (Ayinde) v Haringey LBC [2025] EWHC 1383 (Admin), fabricated citations were submitted, and the notice says the solicitor and barrister faced wasted costs and a regulator referral and that reliance on AI would not be a suitable defence. It also refers to a 2026 Upper Tribunal decision that raised concerns about public AI tools and client confidentiality, a 2026 Family Court case involving an unregistered barrister who misled the court with AI hallucinations (the court found no deliberate intent to mislead), and Cork and another v Smith [2026] EWHC 1199 (Ch) on the responsibilities of authorised persons when unchecked AI errors occur. The notice does not describe outcomes of any SRA investigations or sanctions.

What the SRA expects

• Verify that all named case law is genuine, relevant and has a verifiable citation.
• Apply human oversight and professional judgement, using a proportionate, risk-based approach.
• Maintain governance, systems and controls that cover AI risks.
• Before entering client information into an AI system, ensure contractual, technical and organisational safeguards exist, so that data stays in a secure environment, is not used to train models without authorisation and is not kept longer than necessary.
• Remember that using AI does not reduce or transfer professional responsibility.

For in-house solicitors, the notice says to be cautious about tools not designed for legal work and to consider conflicts between their organisation’s AI interests and their professional duties.

Rules the notice points to

The notice refers to several paragraphs of the SRA Code of Conduct, including paragraphs 2.1, 3.2, 3.5, 3.6, 4.2, 4.3, 4.4 and 9.1, which covers the duties of the Compliance Officer for Legal Practice. It also mentions Authorisation of Firms Rule 9.4, a three-year practising requirement for at least one supervisor of regulated work, which the notice says does not apply to all supervisors. Our article on effective supervision in legal practice looks at the wider supervision expectations.

Our suggestions for firms

These are our own practical suggestions, not statements from the SRA. Keep a record of which AI tools are approved and for what tasks. Make citation checking against an authoritative source a named step before anything goes to court. Make sure trainees and paralegals know the rules on client data and AI tools. Review the contract terms of any tool before client information goes into it.

What the notice does not say

The SRA’s page gives no count of reports and no time period. It does not specify approved AI tools or verification methods, and it does not say how self-reporting will be treated. Its client care section is brief and does not set out specific steps.

Frequently asked questions

Does the notice create new obligations?

The SRA frames it as helping firms understand existing obligations rather than creating new ones.

Is relying on AI a defence if a citation is wrong?

The notice says that, in the Ayinde case, reliance on AI would not be a suitable defence.

Can client information go into a public AI tool?

The notice warns that this may breach confidentiality, data protection requirements and privilege, and says safeguards should be in place first.

Who does the notice apply to?

All firms and individuals the SRA regulates.

Next steps

To keep your team up to date on professional regulation, browse our CPD resources, and read the full notice on the SRA website before updating your policies.