The SRA warning notice on AI misuse arrived on 17 August 2026, and it is a different kind of document from the compliance tips your firm has been reading since February. Guidance explains what the rules mean. A warning notice sets out the conduct the regulator expects and states plainly that a firm failing to have proper regard to it risks disciplinary action. The SRA published it alongside the figure that explains the timing, being 42 reports of possible AI misuse received between July 2025 and July 2026, with a number of investigations now open.
What the notice puts in issue
None of it is new law, and that is the point of it. The notice names the paragraphs of the Codes your practice already answers to, being competence, effective systems for supervising client matters, governance and controls, and confidentiality, then ties each one to a failure the regulator keeps meeting. False material reaching a court comes first, and the requirement is flat, that every authority put before a judge is checked as genuine and accurately cited before filing.
Client information comes second. The notice expects contractual, technical and organisational safeguards before client material enters any tool, and it is blunt that paid products as much as free ones store what you type, keep it longer than you assume, or use it to train a model. Supervision comes third, and there the notice shuts a door some firms have been leaning on, because the supervisor of a junior who used the tool answers for the output as fully as the person who wrote the prompt.
Underneath all three sits the proposition that using AI transfers nothing. You answer for the work reaching your client whichever tool produced the draft, and Aileen Armstrong, the SRA's executive director for strategy and policy, framed the expectation as firms and solicitors having appropriate oversight and controls in place. That is a question about your arrangements, not about your software.
What to do about it this week
Start with the record. The notice does not ask you to stop using these tools, it asks you to show what sits behind their use, and a firm holding that in the owner's head has nothing to show. Write down which tools your people may use, what they may put into them and who checks the output before it leaves the building. Then read the retention and training terms of everything already in use, since that is where the confidentiality half of the notice bites hardest in a small practice.
Fix the citation step next, because this is where firms have been caught in public. Every case and every statutory reference in a document going to court gets checked against the report itself before filing, by a named person, and the check gets recorded. The notice rests on decided cases, among them Cork and another v Smith, where a firm misled the court through AI nobody had checked, and the courts have shown no appetite for treating that as an honest slip.
Then tell your supervisors that the duty has landed on them. A junior who produces a fluent draft from a tool has not done the work, and the person signing it off carries the file. Where your firm keeps an AI use policy, read it against the notice this week. Where it keeps none, the notice reads as the specification for writing one, and the 42 reports tell you how long the quiet period is likely to last.
The warning notice is published free to read on the regulator's website as Misuse of AI, and the SRA set out the reporting figures in a press release of the same day.
If your firm uses these tools daily but the rules for them live in conversation rather than on paper, this is the gap we close with firms before a complaint tests it: talk it through with us.