SRA warning notice · 17 August 2026
The SRA's warning notice doesn't ban AI. It expects oversight and controls. Graft reads the systems your firm already runs on, shows which AI tools have access to your data and where AI would pay off, and gives your COLP a clear picture to work from.
The SRA received 42 reports of potential AI misuse between July 2025 and July 2026, then published a warning notice on the misuse of AI. It names two areas of concern, and it is clear about who carries the responsibility.
AI tools can produce cases and citations that look convincing and have “no basis in fact”. The notice cites a judgment saying a referral to the regulator is “likely to be appropriate” when false citations reach a court. Supervisors stay accountable for work that goes out under them.
Putting client information into a public AI tool “will likely breach client confidentiality”, and privilege “may be permanently waived”. The SRA expects “appropriate contractual, technical and organisational safeguards” first.
“The use of AI does not diminish or transfer your professional responsibilities.” Firms need “effective governance structures, systems and controls” to manage the risk.
Sources: SRA, Misuse of AI: warning notice (17 August 2026); SRA news release (17 August 2026).
General information, not legal advice. Graft is not a law firm and is not affiliated with or endorsed by the SRA.
You can’t govern AI use you can’t see. Graft starts with a real picture of which AI tools have access to firm data and how widely, then shows where AI would pay its way.
Graft lists the AI tools that have been given access to your Google Workspace, and how many people granted each one. You start from the facts, not a guess.
Your COLP sees which AI tools can reach firm data, how widely, and where AI would pay off. Oversight and controls are much easier to set when you can see what you are overseeing.
Your COLP, COFA and solicitors keep their duties under the SRA Standards and Regulations. Graft shows them which AI tools can reach firm data, so they can act on it.
Plenty of vendors will tell you where AI could help. We build it, run it, and put a number on what it returned. That number is how you defend it to a client, a regulator or the partners.
We connect to the systems your firm already runs on and read how work really moves, instead of asking fee-earners to fill in a survey.
You get a ranked shortlist of where AI pays off in your firm, and what each opportunity is worth.
We build the agents to capture it. Fixed scope, no open-ended retainer.
They work against real matters, with a solicitor in the loop and an audit trail behind every step.
We measure what they save in pounds, every month. If it isn't working, you'll know first.
Every firm is different, so the assessment finds the real opportunities in yours. These are the areas that come up most often.
Triaging large document sets and flagging what a person needs to read first.
Routine correspondence and standard clauses, drafted for a fee-earner to check and finish.
Structuring new enquiries and surfacing the checks a matter needs before it opens.
Turning the day's work into accurate, well-written time narratives.
Finding the right precedent or past advice across the firm's own knowledge.
Assembling first-pass bundles and case chronologies from the matter file.
The trap firms fall into, and the response that lowers risk without banning AI.
Read itHow to find the AI use already happening off your network, before it finds you.
Read itWhat happens to client data in a public model, the privilege risk, and the tier distinction that matters.
Read itWhen AI invents case law, what the courts have done about it, and the one check that prevents it.
Read it