Aipedia: data

AI misuse cases and sanctions in UK professions

Aipedia data, by the Apex broking team · Last updated 2026-10-30

UK courts, tribunals and regulators have sanctioned or warned about AI misuse in professional work since 2023, mostly over fake citations in legal and tax proceedings. This list gives each matter with its date, profession, outcome and source.

Part of: Aipedia

In short

Since Harber v HMRC in December 2023, UK courts and tribunals have repeatedly dealt with fake AI-generated citations, including Ayinde (June 2025), Munir (November 2025) and Elden (January 2026). Outcomes so far are wasted costs, regulator referrals, a finding of professional incompetence and one reported strike-off. Regulators have added warnings: the SRA logged 42 reports of potential AI misuse in a year. The ICO has acted on AI deployment without proper risk assessment. Most professions still have no verified AI case.

Scope of this list

This page lists every UK court or tribunal decision, regulator warning and enforcement step involving AI in professional work published as at 30 October 2026. It draws on the facts sheets behind the 57 profession hubs in Aipedia. Trade press reports are labelled. International examples are kept separate. Each row links to the primary source where one exists.

UK cases, tribunal decisions and regulator action

The table below runs in date order. Profession names link to the relevant Aipedia hub.

Date	Case or action	Profession	What happened	Outcome	Citation
4 Dec 2023	Harber v HMRC [2023] UKFTT 1007 (TC)	Tax advisers	An unrepresented taxpayer cited nine First-tier Tribunal decisions. None were genuine; the Tribunal found they were AI-generated.	Appeal dismissed. Fake cases "promote cynicism" and waste resources (para 24).	National Archives
6 Oct 2023 to 21 May 2024	ICO investigation: Snap's generative AI chatbot	IT consultants and developers, Data protection consultants	Preliminary enforcement notice for not adequately assessing the risks of a generative AI chatbot before launch.	Closed after Snap completed a compliant risk assessment; the ICO warned it may use its full powers.	ICO
26 Mar 2024	Manjang v Uber Eats (EHRC-funded)	HR consultants, Recruitment agencies	Automated facial recognition checks repeatedly failed to verify a Black courier, and an automated process removed him from the platform.	The tribunal refused to strike out the claim; it then settled for an undisclosed sum.	EHRC
23 Jul 2024	ICO reprimand: Chelmer Valley High School	Schools, academies and governors	Facial recognition was used for canteen payments without a DPIA, relying on opt-out slips rather than consent.	Reprimand. "A DPIA is required by law".	ICO
6 Nov 2024	ICO audit of AI recruitment tools	Recruitment agencies	Some tools filtered candidates by protected characteristic, inferred gender or ethnicity from names, or kept data indefinitely.	Almost 300 recommendations to providers, all accepted or partly accepted.	ICO
3 Apr 2025	Zzaman v HMRC [2025] UKFTT 539 (TC)	Tax advisers	A statement of case prepared with AI contained citations that were inaccurate and did not support the propositions.	The Tribunal highlighted "the dangers of reliance on AI tools without human checks" (para 29).	National Archives
6 Jun 2025	R (Ayinde) v Haringey LBC; Al-Haroun v Qatar National Bank [2025] EWHC 1383 (Admin)	Barristers, Solicitors and law firms	Ayinde: five fictitious cases in judicial review grounds settled by a pupil barrister. Al-Haroun: 45 citations, 18 non-existent, generated by the client with public AI and relied on unchecked by the solicitor.	Wasted costs of £2,000 each against counsel and the law centre; contempt threshold met but not pursued; referrals to the BSB and SRA. Leaders must take "practical and effective measures" (para 9).	National Archives
12 Aug 2025	MS (Professional conduct; AI generated documents) Bangladesh [2025] UKUT 305 (IAC)	Barristers	A barrister used a generative AI chatbot and cited a non-existent case, "Y (China)".	Referred to the BSB.	National Archives
Reported 22 Sep 2025	HMRC v Gunnarsson [2025] UKUT 247 (TCC)	Tax advisers	A litigant in person's skeleton cited three tribunal decisions that appear not to exist.	The Upper Tribunal applied Ayinde to tribunals: unrepresented parties "remain responsible for the accuracy"; costs, wasted costs and strike-out are available.	ICAS summary
17 Nov 2025	UK and R (Munir) v SSHD [2026] UKUT 81 (IAC)	Solicitors and law firms, Chartered legal executives	Fake authorities appeared in grounds drafted by a junior caseworker.	The signing COLP was referred to the SRA. A supervisor who misses a junior's false citations is likely more culpable. Uploading client documents to an open AI tool puts them "in the public domain".	National Archives
8 Jan 2026	Elden v HMRC [2026] UKFTT 41 (TC)	Accountants, Tax advisers	A skeleton argument prepared by ICAEW-regulated chartered accountants used AI; its case summaries misdescribed authorities.	Lack of care "amounts to professional incompetence" (para 139). Future skeletons need a statement of truth and contributors' qualifications and regulator. No referral made.	National Archives
31 Mar 2026	ICO "Recruitment rewired"	HR consultants, Recruitment agencies	Evidence from more than 30 employers suggested solely automated hiring decisions are likely widespread, with weaker safeguards than the law requires.	Regulator findings; the ICO will scrutinise major employers and recruitment platforms.	ICO
2 Apr 2026	Financial Ombudsman Service response to the Mills Review	Claims management companies, Financial advisers	Professional representatives filed AI-assisted submissions of about 200 pages with multiple inaccuracies; AI likely contributed to up to 35% of responses to initial assessments in a small sample.	Regulator evidence; the FOS asked the FCA for clarity on records and human escalation.	FOS
20 Apr 2026	GPhC position statement on AI	Pharmacists and pharmacies	Warning that using AI to create full revalidation submissions, or to falsify information, is inappropriate.	Such use may be investigated.	GPhC
16 Jul 2026	CROSS-UK Report 1554	Structural and civil engineers, Architects	An architect's calculation report for heavy rooftop plant on an existing flat-slab building was, in the reviewing engineer's view, largely AI-written. It was fundamentally wrong, with overloading risk and no competent structural engineer verification.	Safety report. The panel said responsibility must stay with qualified structural engineers and designers should disclose AI-assisted outputs.	CROSS-UK
17 Aug 2026	SRA "Misuse of AI" warning notice	Solicitors and law firms	The SRA received 42 reports of potential AI misuse between July 2025 and July 2026, covering citations, supervision and confidentiality.	Investigations ongoing; failing to consider the notice risks disciplinary action.	SRA
2 Sep 2026	ASA ruling: Glow Up LLC	Aesthetics clinics, Marketing and PR agencies	Ads for an AI appearance-scoring app used AI-generated before-and-after transformations not shown to be genuine or typical.	Complaint upheld: ads misleading and irresponsible.	ASA
Reported 4 Sep 2026	Solicitors Disciplinary Tribunal: registered foreign lawyer	Solicitors and law firms	AI-generated false quotations and citations appeared in his defence to the SRA.	Struck off the register; no costs order as he was impecunious. Trade press report; the SDT judgment was not reviewed.	Legal Futures
17 Sep 2026	Hancox v Sutherland & Ors [2026] EAT 139	HR consultants	A litigant in person filed an AI-generated "skeleton argument" of about 300 pages and 132,000 words without checking it.	Not relied on; the judge warned such documents are likely to be rejected in future.	EAT judgment
23 Sep 2026	Planning Inspectorate AI guidance update	Planning consultants	Updated rules on declaring AI use in casework evidence.	Improper AI use may be unreasonable behaviour leading to an award of costs.	GOV.UK

What the UK cases show

Four themes run through the UK cases.

Fake citations dominate. Most decisions concern AI-generated authorities that do not exist, from Harber in 2023 to Elden in 2026. Courts and tribunals have applied the same standard to lawyers, accountants and litigants in person.
Supervisors and leaders carry the risk. Ayinde puts duties on heads of chambers and managing partners, and Munir says a supervisor who misses a junior's fake citation is likely more culpable. The SRA notice says supervisors stay accountable.
Confidentiality is the second front. Munir treats uploading client documents to an open AI tool as placing them in the public domain. The ICO actions on Snap, Chelmer Valley and recruitment tools turn on risk assessment and DPIAs before deployment.
Sanctions so far are referrals, costs and warnings. Wasted costs orders, regulator referrals and one reported strike-off are the main outcomes. Some tribunals now ask for statements of truth, and the Planning Inspectorate requires AI declarations.

For most professions no AI-specific case had been published, including surveyors, architects, doctors, dentists, nurses, vets and insolvency practitioners. That may change: R3 reports a rise in AI-generated complaints to insolvency regulators (R3 report, August 2026). The SRA notice also cites Cork v Smith [2026] EWHC 1199 (Ch) on authorised persons' responsibility for unchecked hallucinations; we have not reviewed that judgment directly.

Related UK judgments on AI law

Some UK judgments concern AI itself rather than misuse by a professional. They shape how professionals advise on AI, so they are listed separately below.

Date	Judgment	Relevant to	Holding in brief	Citation
20 Dec 2023	Thaler v Comptroller-General [2023] UKSC 49	Patent and trade mark attorneys	An AI system cannot be named as an inventor; an inventor must be a natural person.	National Archives
4 Nov 2025	Getty Images v Stability AI [2025] EWHC 2863 (Ch)	Marketing and PR agencies, Interior designers	Secondary copyright claims failed because the model was not an "infringing copy"; limited trade mark infringement found for watermarks.	National Archives
11 Feb 2026	Emotional Perception AI v Comptroller-General [2026] UKSC 3	Patent and trade mark attorneys	The Supreme Court departed from the Aerotel test; neural network claims involving hardware are not excluded "as such".	National Archives

International examples (not UK)

The examples below are from outside the UK. They are not UK law, but they show how AI errors and claims about AI are being treated elsewhere. News reports are labelled as such.

Date	Country	Matter	Relevant to	What happened	Source
18 Mar 2024	United States	SEC: Delphia (USA) Inc. and Global Predictions Inc.	Wealth managers and discretionary fund managers, Financial advisers	Settled charges for false and misleading statements about their use of AI ("AI washing"); civil penalties of $225,000 and $175,000.	SEC
Oct 2025	Australia	Deloitte report for a federal department (news reports)	Management consultants, Accountants	A report containing a fabricated quote attributed to a court judgment and references to non-existent papers was revised to disclose generative AI use; Deloitte agreed a partial refund. Not a court decision.	CFO Dive
Nov 2025	Canada	Deloitte health workforce plan, Newfoundland and Labrador (news report)	Management consultants	Citations in a 526-page plan appeared not to exist; Deloitte acknowledged the errors and stood by the plan.	Global News

What this means for PI

Several outcomes above could lead to a PI claim or a costs exposure. How a policy responds usually depends on its wording.

Wasted costs orders are compensatory costs orders, not fines. Whether PI responds depends on the policy terms.
Regulatory fines and penalties are commonly uninsurable, and for charity trustees the Charities Act 2011 s.189(2)(a) bars insuring them.
A client who loses a case, an appeal or a planning costs award because of unchecked AI output may bring a negligence claim, which PI is designed to answer, subject to the terms.
Statements of truth and AI declarations, now required in some tribunals and planning casework, create a written record that a claim will be judged against.

The solicitors and tax advisers hubs explain the controls that respond to these cases.

Sources

Frequently asked

What was the first UK case on fake AI citations?

Harber v HMRC [2023] UKFTT 1007 (TC), decided on 4 December 2023, is the earliest in our list. An unrepresented taxpayer cited nine tribunal decisions that the First-tier Tribunal found were not genuine and had been generated by AI. The appeal was dismissed.

Has any UK professional been struck off for AI misuse?

Trade press reported in September 2026 that the Solicitors Disciplinary Tribunal struck off a registered foreign lawyer after AI-generated false quotations and citations appeared in his defence to the SRA. We have not reviewed the judgment. No other strike-off for AI misuse was verified.

What did the Ayinde judgment decide?

The Divisional Court addressed fake citations in two cases on 6 June 2025. It made wasted costs orders, found the contempt threshold met but did not pursue it, and referred lawyers to the BSB and SRA. It said leaders and regulators must take "practical and effective measures".

Does the Elden case affect accountants?

Yes. In Elden v HMRC the tribunal said unchecked AI-drafted submissions by an ICAEW-regulated firm amounted to "professional incompetence". It ordered future skeletons to carry a statement of truth and contributors' qualifications and regulator, and said it may refer people to professional bodies.

Are AI cases outside law and tax relevant?

Yes. The ICO acted on AI deployment by a technology company, a school and recruitment tool providers, and the ASA upheld a complaint about AI-generated before-and-after images. CROSS-UK reported an AI-written structural calculation. These show the risk is not limited to court documents.

Check your PI covers how you use AI

Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.

Start a PI quote Request a callback

Apex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.