SRA Warnings on AI: A staunch reminder of the need for human input
The Solicitors Regulation Authority issued a formal Warning Notice on 17 August 2026 on the misuse of AI in legal practice, flagging two risks: hallucinated content and client confidentiality breaches.
The Solicitors Regulation Authority issued a formal Warning Notice on 17 August 2026 on the misuse of AI in legal practice, flagging two risks: hallucinated content and client confidentiality breaches. The SRA isn't banning AI use. It's requiring firms to verify AI output before it reaches a client or a court, and to vet AI vendor contracts properly. The Law Gazette reports the regulator is now looking into dozens of reports of AI misuse by firms and solicitors.
The warning follows a run of real cases. In Ayinde v London Borough of Haringey and the linked Al-Haroun v Qatar National Bank [2025] EWHC 1383, fabricated case citations reached court submissions. In Pro Health Solutions Ltd v ProHealth Inc (UKIPO, June 2025), ChatGPT produced fake quotes attributed to real legal authorities. An unrepresented litigant in HMRC v Gunnarsson filed three tribunal decisions that don't exist. A former solicitor in Bandla v SRA cited multiple fake authorities and had the appeal struck out.
The same pattern is showing up on the compliance side. Automated AML systems have frozen accounts over legitimate activity misread as suspicious, in one case twice, at two different banks. A chaplain working the 2024 Paris Olympics had his account closed mid-tournament with no explanation. French officials confirmed his case wasn't isolated. In several documented disputes, banks initially denied AI involvement in a closure, then reversed that position once shown their own compliance records.
The UK's biggest AML fine so far came from the opposite failure: a screening gap, not a screening block. The FCA fined Starling Bank £28.96m in 2024 after its automated sanctions screening system checked customers against only a fraction of the full sanctions list, a flaw that ran undetected from 2017 to January 2023. Over that same window the bank also opened more than 54,000 accounts for around 49,000 high-risk customers, in breach of an agreement restricting exactly that. The FCA called the controls "shockingly lax."
Direct Compliance. UK risk & compliance roles, applied for directly. directcompliancejobs.com
Sources:
SRA warns solicitors over AI hallucinations and client confidentiality — Family Law
SRA Warning Notice on "Misuse of AI" dated 17 August 2026 — 4 New Square Chambers
SRA looking into dozens of reports of AI misuse by law firms and solicitors — Law Gazette
12 False Citations/AI Hallucinations in UK Courts (Ayinde) — Natural and Artificial Law
FCA fines Starling Bank £29m for failings in their financial crime systems and controls — FCA