Ad/Marketing Communication
This legal article/report forms part of my ongoing legal commentary on the use of artificial intelligence within the justice system. It supports my work in teaching, lecturing, and writing about AI and the law and is published to promote my practice. Not legal advice. Not Direct/Public Access. All instructions via clerks at Doughty Street Chambers. This legal article concerns AI hallucinations UK court cases. Subscribe here.
Introduction
I sometimes wonder whether readers are growing weary of discussions about AI hallucinations in legal work. As of today, we have close to 900 reported cases worldwide and today I will be discussing three more. In the UK alone there are now 37 recorded incidents (with some more to write up shortly). I remain concerned about the direction of travel and about how I can continue to keep pace while balancing a busy practice with the many complex and genuinely fascinating AI issues we explore on this blog.
That said, my recent post on AI hallucinations, which can be read here, may have attracted more readers and contributors than any other piece I have published. I therefore want to offer a sincere thank you, first to everyone who took the time to read and share that article and secondly to those of you who continue to send me decisions to read, consider and track. Your engagement and generosity of thought make this work both possible and worthwhile.
I would also like to extend my thanks to the thoughtful speakers and attendees at the Public Law Project event, “Morning update: AI and automated decision making, litigation and law reform”. It was a real pleasure to meet so many of you in person and to hear about how AI is being used across different areas of practice, sometimes with clear benefits and sometimes raising more difficult questions. The discussions at that event have given me a great deal to reflect on and there will certainly be more to write about in due course.
Professional Standards Authority for Health and Social Care v Nursing and Midwifery Council
The appeal focused on whether a nine-month suspension was wrong in principle and whether a strike-off should have been imposed instead. Alongside that central issue, the judgment records a notable moment about the written submissions filed for the second respondent by an unregulated representative (acting free of charge). The judge recorded that the submissions:
“…were partly generated by Artificial Intelligence (AI) via Microsoft’s Co-Pilot software. In particular, he relied on three authorities, two of which did not contain the words he attributed to them and none of which supported the propositions which he said they supported…” [the Judge set out the hallucinations]
The judge recorded that the representative used AI-generated citations:
“…This was pointed out to him at the hearing. He immediately admitted what he had done and that the references were phantoms created by AI. He promised not to use AI to generate submissions in future and to check his references personally…”
There was a warning about possible contempt but no further steps were taken in that case.
Folarin v The Immigration Services Commissioner
This was an appeal pursuant to section 87(2) of the Immigration and Asylum Act 1999 which concerns a decision made by the Immigration Services Commissioner. The court noted:
28. In presenting his case, and in particular in the section of his witness statement dealing with the legal framework, [the Appellant] referred to a number of decided cases in support of his points. The Tribunal asked questions in the hearing about these cases, because no or inaccurate citations and no copies of these cases were provided and the panel had struggled to locate some of these cases when preparing for the hearing. When asked directly if he had used artificial intelligence (AI) in compiling his legal submissions, [the Appellant] said that he had. He said he had used Westlaw and a paid subscription to a legal portal on ChatGPT in doing so. Upon further questioning, he explained that when making a point in his submissions he had asked ChatGPT to provide him with the top 10 relevant cases supporting the points he wished to make and asked it to produce an extract from the case law which he then had edited and abbreviated in his submissions. Accordingly, the cases on which he had relied were ones which AI had located for him. He accepted that he had not read the judgments in the underlying cases and had used ChatGPT to provide him with a summary so he could understand what they say. He also used AI for additional research and to polish the wording. He said that he had not intended to mislead the Tribunal but that he did not have the means to employ a lawyer and wanted to present his case to the best possible standard.
The Tribunal returned to the AI use later in the judgment:
“49.The Tribunal also had significant concerns about [the Appellant’s] integrity as a result of the way in which he presented his appeal and, in particular, in relation to his use of AI. He presented a number of cases to the Tribunal as settled law when these had been located for him by AI. In preparing for the hearing, the panel was unable to locate several of these cases in any database and could not therefore be satisfied that they actually exist. This would have had the effect of misleading the Tribunal had the Tribunal not challenged the point. We note that section 83(5)(a) states that the Commissioner must exercise his functions to so as to secure, so far as reasonably practicable, that those who provide immigration advice or immigration services do not knowingly mislead any court, tribunal or adjudicator in the UK, so potential misleading of a tribunal is a serious matter. We note [the Appellant] evidence that he did not intend to mislead the Tribunal, but also that he was aware that he had not verified the case law on which he relied.”
The Tribunal then considered the principles in Ayinde v LBH and the distinction between negligence and the knowledge required for contempt:
“50. The use of AI for legal research in the context of litigation has been recently considered by the High Court in the case of R on the Application of Frederick Ayinde v London Borough of Haringey [2025] EWHC 1383 (Admin). The decision of Dame Victoria Sharp P indicated at paragraph 26 that “Placing false material before the court with the intention that the court treats it as genuine may, depending on the person’s state of knowledge, amount to a contempt. That is because it deliberately interferes with the administration of justice.” Whilst mere negligence would not be sufficient to establish contempt, and knowledge that the information is false or a lack of honest belief that it was true would be required, she stated at paragraph 29 that “Where a lawyer places false citations before the court (whether because of the use of artificial intelligence without proper checks being made, or otherwise) that is likely to involve a breach of one or more of the regulatory requirements that we have set out above, and it is likely to be appropriate for the court to make a reference to the regulator.”
The Tribunal accepted that the appellant was not a qualified lawyer, but noted he had completed an LLB and the Legal Practice Course and was seeking registration as an immigration adviser, which would involve giving advice on aspects of immigration law. The Tribunal was concerned that, if the appellant was willing to present his own case relying on non-existent authorities and without properly checking what AI had produced, there was a real risk he might take the same approach when advising clients, including people who may be inexperienced in immigration matters or otherwise vulnerable. The Tribunal added that, if he had already been registered and had conducted himself in that way, it would in any event have had to consider reporting him to the regulator. On that basis, it concluded that the appellant was not currently fit to provide immigration advice and that the refusal of his application was correct.
Huish v The Commissioners for HMRC
This was an appeal against discovery assessments made under section 29 Taxes Management Act 1970 (“TMA”). The Appellant represented himself and the court noted at paragraph 5:
“.For one of his arguments, the appellant relied upon four cases which he said had been decided in 2021 and supported his position namely, [cases named]. We granted an adjournment so that we, and HMRC, could check those cases since the names were not familiar. On reconvening we asked the appellant if he had used AI to assist in preparation for the hearing and he confirmed that he had done so. He was horrified when we pointed out, and Ms Aziz confirmed, that those alleged cases did not exist. We attach no blame to him, since he is a litigant in person but we have recorded the names so that others do not fall into the same trap.”
Comment
There are some important takeaways from each of the decisions discussed above and I want to reflect on them carefully, while remaining mindful of the competing considerations that judges and practitioners are rightly balancing in this developing area.
Readers of my legal writing will know that for some time I have been concerned about judgments that reproduce fabricated citations in full. I have had thoughtful and constructive conversations with colleagues about whether those concerns are well founded. Some take the view that the public interest in setting out the full picture outweighs the risk I describe. I understand that position and indeed my own work depends on judges being willing to make careful and transparent observations of this kind. Others accept that there is a theoretical risk, but consider it far less serious in practice than those who share my concerns suggest.
These are all fair and well made points. Despite the strength of those arguments, I remain concerned that this is a genuine issue which the profession would benefit from addressing sooner rather than later, ideally through some form of shared and consistent approach. I have been thinking about how that might be achieved in a way that preserves openness while reducing the risk of unintended consequences. One possibility, which seems to have been adopted in Huish above, is simply to set out truncated names rather than full fabricated citations.
However, it seems to me that a more structured solution may be to maintain a public database. This could be called the “False Citation Authorities Register”, or something similar. The idea would be a simple one. Where a judgment identifies that a party has relied on hallucinated authorities, the judge could record that fact by stating, for example, that an AI hallucination was identified in submissions from a particular party and that it has been logged as a numbered entry on the False Citation Authorities Register, rather than setting out the fabricated material in full within the judgment itself.
In this way, judges would be able to be open about what occurred and why it mattered, while avoiding the risk of giving fabricated material a permanence or visibility that might lead to it being reused, misunderstood, or treated as legitimate in the future. The court could also remind parties, before documents are submitted, to check the False Citation Authorities Register. This could serve a dual purpose. It would allow parties to see whether a false citation has already appeared through AI use and it would also focus minds on the fact that the court expects cited authorities to be accurate and capable of verification.
My hope is that an approach of this kind could encourage shared learning across the profession, while keeping the integrity of the legal record firmly intact. I would be very interested to hear your thoughts on the idea and in particular what you see as its potential benefits and its potential pitfalls.
As to the Folarin judgment, I found it particularly valuable for the insight it offers into how litigants in person may use generative AI and how easily errors can creep in when that happens. The appellant described a very specific workflow: using Westlaw alongside a paid “legal portal” within ChatGPT, asking ChatGPT for a “top 10” list of cases, then asking it to generate extracts, and finally editing and abbreviating those extracts into submissions. He had not read the underlying judgments, instead relying on ChatGPT’s summaries to understand what the cases were said to decide. However, “…he did not have the means to employ a lawyer and wanted to present his case to the best possible standard.”
Crucially, the judgment treated this as more than a drafting misstep. It became a relevant consideration of a live issue in the case itself closely linked to the statutory aim of preventing advisers from misleading courts and tribunals and it fed directly into the conclusions. It will be interesting to see in future judgments how far an individual’s use of AI directly relates to substantive issues in the case.
I remain grateful to everyone who continues to read, engage and think through these issues with me. The tone of the judicial responses in these cases shows real care and restraint and it is precisely that care which makes these questions worth exploring openly and calmly as we work out what responsible practice should look like in this new landscape.
I have included below a section from my global hallucination tracker (AI and non‑AI hallucinated citations) that focuses solely on the UK position. I would be very grateful if readers who are aware of further cases not yet listed could share them with me, so that I can keep the tracker as accurate and up to date as possible. There is more analysis to follow shortly. If you find these articles helpful, please consider subscribing below.
Final Words from Chat GPT 5.2 Pro and Gemini 3 Pro
Chat GPT:
“AI risk is not limited to hallucinating whole “cases” — it can fabricate quotations and invert what real authorities stand for. If you file AI-assisted submissions, the court may publicly label your conduct as “Misuse of Artificial Intelligence,” and that label will sit in the judgment forever. You can be warned about contempt even when the court cannot refer you to a regulator.”
Gemini:
“The judicial patience for ‘AI innocence’ is evaporating. The warning is clear: if you allow an algorithm to read and understand the law for you, you risk being deemed unfit to advocate for others.“
Tracker
The full international AI and non‑AI fabricated/false citations tracker can be here.
The UK only one is here.
The section below provides only a summary of the UK cases involving AI and non AI fabricated or false citations. You can click on each case name to read my detailed discussion of it on this blog and from there you can use the hyperlinks to obtain the full judgment. The table will update automatically as the main tracker is expanded, even if the surrounding blog text remains unchanged over time.
| Number | Date | Case Name | Outcome | Judicial Quotes | Source |
|---|---|---|---|---|---|
| 87 | 01/10/2026 | FJ v SOSHD PA/95637/2025 | No AI related outcome | "31. In his skeleton argument the Appellant expressly relies on the Respondent’s policy documents. There are also two apparent references to Upper Tribunal decisions in the skeleton but these would appear to be AI hallucinations [full case names redacted]. I therefore focus on the Respondent’s current policy documents as cited in the review as being the most current and relevant sources of country information." | |
| 86 | 22/09/2026 | R v Willesden County Court [2026] EWHC 2484 (Admin) | suspected AI not held against C. No seperate AI sanction was imposed | "I note in passing that the claimant's grounds cited four cases. One is a well-known Supreme Court decision under the Equality Act 2010. I could not find the other three cases when I searched the citation references provided via the usual online resources and I wonder whether they are the product of AI hallucinations. However, as I have not had the opportunity to explore this with the claimant, I will not hold it against him." | |
| 85 | 21/09/2026 | NK (Namibia) v SOSHD | Costs tbc and self referal to SRA | "As we will explain, this appeal has an unhappy history, thanks to the use by the appellant's solicitors of AI to generate the grounds of appeal and skeleton argument. Two individuals have reported themselves to the Solicitors Regulation Authority and so we will say nothing about the "professional conduct issues involved. However, this decision illustrates the considerable procedural difficulties which can arise when inappropriate use is made of AI. | |
| 84 | 04/09/2026 | HT v SOSHD UI-2026-000991 | No AI related outcome | "...I assume the reliance on it here for a completely different principle is the result of an AI hallucination. The principles set out in Karanakaran will be discussed below, but they do not include any such simple and prescriptive rule." | |
| 83 | 25/08/2026 | Solicitors Regulation Authority Ltd v Abhishek Kumar (12884-2026) | Struck off the Register of Foreign Lawyers; no order as to costs. The Tribunal would have imposed the same sanction for either allegation alone. | "Mistakes happen. Not every mistake will be a matter serious enough to engage the Regulator or this Tribunal. In his response to the Rule 14 Statement, the Respondent had categorised his wrongdoing as “technological negligence”. His insistence, in his Rebuttal Statement of 28 July and in oral submissions, that he ‘mistakenly believed’ that AI was a “lawful drafting aid” reflected a persistent lack of insight into his misconduct. He blamed having been out of legal practice and being, “unaware of recent regulatory warnings or procedural directives regarding AI research tools”. The Tribunal found that the Respondent had embarked on the use of AI without giving any proper thought to the consequences of so doing. He had failed to take proper steps to ensure his work output was accurate or to satisfy himself that the electronic tools he was seeking to use were reliable. Having done so, the Respondent’s culpability was very high because it was repeated after errors had been pointed out and explained. Rather than take active steps to rectify his mistakes, the Respondent repeated them. The Tribunal gave very substantial weight to the repeated nature of the Respondent’s misconduct." | Read decision |
| 82 | 24/07/2026 | Lodhia v Twelve Trees (Bromley-by-Bow) Management Company Ltd and others [2026] EWHC 1889 (KB) | Indemnity costs, a three-year general civil restraint order and referral to the Attorney General’s Office formed part of the wider response to his conduct. | "...did not explain the 22 April documents which he sent to rradar, I asked him about them at the hearing. He said that all of the false evidence and documents were created by AI and were AI hallucinations. He provided a more detailed explanation in his 27th witness statement which was essentially to the same effect i.e. that he had mistakenly sent the wrong documents, AI having generated the false ones without his knowledge and without him realising what had happened. His 27th witness statement includes the detailed prompts which he says he put into the AI tool and exhibits from his output folder, but his explanation requires the court to accept that, having been told on 20 April that his oral criminal law assessment was on 12 May, on the following day he genuinely believed that it was on 6 May. It also requires acceptance that the AI tool generated documents which were false in other respects without any instruction or request from him to do so, that he then mistakenly sent off the wrong versions twice and that he did so without reading them and realising that they were false at any point before Steyn J's Order..." | Read decision |
| 81 | 23/07/2026 | iHeat.me Ltd v HMRC [2026] UKFTT 1089 (TC) | No AI related outcome | "...It has been accepted that this document was prepared using an AI tool and included fictitious references which were not checked. Some of these errors have been corrected, HMRC continue to question some other references. I make no finding in respect of asserted errors..." | |
| 80 | 14/07/2026 | Unnamed Moroccan asylum appeal (Guardian report, 28 July 2026) | The report describes a procedural irregularity and an ongoing appeal. No final disposal is established from the available report. | TBC | Read report |
| 79 | 08/07/2026 | Tobosaru v Court of Law Craiolva, Romania; Tofan v Secuava Local Court, Romania [2026] EWHC 1720 (Admin) | Apology accepted and incident recorded publicly. Errors were identified before the appeal hearing and had no effect on the argument or judgment. | 105. As far as the appeal was concerned the errors came to light prior to the hearing and therefore had no impact on argument or the court’s judgment. I have accepted the apology given on behalf of the Crown Prosecution Service and the assurance that there was no attempt to mislead. I have noted the steps taken to ensure that there is no repetition but considered that it was necessary to set out what had taken place in this judgment given the serious consequences that an error of this nature might have had in other circumstances. It would be naive to assume that there will not be an increasing use of artificial intelligence in legal work in future; indeed, that may be both necessary and beneficial. The episode highlights the risks of its use without appropriate oversight particularly for legal research.“ | Read decision |
| 78 | 02/07/2026 | YF (Feiner) v Secretary of State for Work and Pensions [2026] UKUT 255 (AAC) | Warning to verify authorities. The appeal was dismissed on its substantive grounds; no separate AI-related sanction is recorded. | "In this appeal Mr Feiner’s representative made reference to a number of cases by way of authority and an extract from Hansard which are either irrelevant to this appeal or, as far as I can tell, do not exist. Ms Gratrex pointed this out in her submission but the representative has not responded. I do not know how the representative found these references but it may well be that he used AI to assist in preparing the written submissions. If so, this should serve as a lesson to him and to other litigants in person and lay representatives as to the dangers of using AI. Any references suggested by AI should always be checked so as to be satisfied that they exist and that they support the proposition contended for. There are ample on-line resources for doing so." | Read decision |
| 77 | 01/07/2026 | Mostafa v Julier and Slater [2026] EAT 122 | Renewed permission application refused. The judge queried a nonexistent authority without criticism of the appellant; no separate AI-related sanction is recorded. | "...Ground (4) refers to a case which I am unable to find, and Judge Glyn was unable to find. I asked the appellant whether he was able to locate it or whether there was a possibility of it having been generated by AI, ( without criticism of him). I am satisfied that such a case does not exist, but in any event that does not assist in circumstances of the changed statutory regime, Practice Direction and Joint Presidential Guidance..." | Read decision |
| 76 | 12/06/2026 | Aisha Habib and Emadullah Habib v Secretary of State for the Home Department (UI-2026-000882; UI-2026-000883) | Show-cause directions to solicitors and counsel. Wasted costs and a regulatory referral were under consideration, not yet ordered in this decision. | "51. I have not set out the names of the cases which appear to me to be likely to be hallucinations because this decision will be published online on the Tribunals Decisions website. The inclusion of cases which are, or may be, fake may perpetuate the loop of such cases being cited by AI if this decision is scraped by a Large Language Model." | Read decision |
| 75 | 22/05/2026 | Cork and another v Smith [2026] EWHC 1199 (Ch) | Public criticism; firm’s self-referral to the SRA recorded; judgment to be supplied to the SRA. The court declined to pursue contempt further. | 95. I do not intend to set out the words of the President in Ayinde again but it will be clear from what she said that legal professionals bear ultimate responsibility for their work and cannot outsource the process of legal research or of legal reasoning to an AI. It is a tool to be used with caution. AI has the potential to be wholly unreliable. AI may of course provide a jumping off point for research and legal reasoning but it does not, at least at present, do away with the need for proper research and thought on the part of a legal professional, even a very junior legal professional | Read decision |
| 74 | 19/05/2026 | Brookes v Executive Telecommunications Case No. 6022493/2024 | No AI related outcome | 16.Mr Grundy also highlighted to me that I should exercise caution when considering the authorities referred to by the Claimant. AI appeared to have been used in some form. Authorities did not necessarily support the proposition advanced, and he was not able to locate two of the authorities and therefore doubted their authenticity. I took note of this. The Claimant did not dispute this. I don’t repeat the names of such authorities here as current ‘thinking’ is that referencing such authorities in a judgment creates a risk of validating them in a public document and thus replicating their AI use. | |
| 73 | 13/05/2026 | World Kuk Sool Association Inc v Paul (BL O/0411/26) | Untraceable reference withdrawn and disregarded; warning that fabrication can attract sanctions. | “With regard to the case referred to in the final paragraph of this quotation, I was unable to locate any such judgment with that citation. As a result, after the hearing, I wrote to [redacted name] to ask for a copy of that judgment. In response, he asked that I disregard the reference. It is important for parties to be aware that they have a duty not to mislead the tribunal and that the fabrication of authorities (whether intentionally or because references generated through Artificial Intelligence have not been verified) can lead to sanction. I have, as indicated, disregarded the reference in question.” | Read decision |
| 72 | 06/05/2026 | Rafique v HMRC [2026] UKFTT 673 (TC) | Reinstatement and challenge to strike-out refused. Warning about fabricated authorities; no contempt referral made. | 100. While litigants in person in the FTT are (entirely appropriately) given a greater degree of latitude in their conduct of the proceedings than represented parties, no person has immunity from sanction when committing a serious breach of court or tribunal rules. That is well demonstrated in Olsen v Finansiel Stabilitet A/S [2025] EWHC 42 (KB) where the High Court only “narrowly and somewhat reluctantly” decided against causing a summons for contempt of court to be issued to two elderly litigants in person who had provided a fabricated citation to the court. | Read decision |
| 71 | 05/05/2026 | Tuyunuklu v Akmal [2026] UKUT 174 (LC) | Incorrect legal proposition rejected; appeal dismissed for the reasons given in the judgment. No separate AI-related sanction recorded. | 16. In considering the parties’ arguments I start from two propositions of law. The first is that Rakusen v Jepsen is not authority for the idea that [redacted principle]. The Supreme Court did not say that. From references to AI in the transcript I wonder if the appellant was misled by an AI hallucination. Whatever the reason, his point about Rakusen v Jepsen is without foundation | Read decision |
| 70 | 23/04/2026 | Rodney v Gee’z Micro Bar (M00DD215) | Solicitors referred to the SRA; public judgment. The judge declined to initiate contempt proceedings. | "It is important that there is a public record of this incident should anything remotely similar occur again involving [redacted firm] or these solicitors. I therefore direct that a transcript of this judgment be prepared at public expense and be published on the Judiciary website. A copy shall be sent to the Solicitors Regulation Authority alongside the witness statements filed by both solicitors and my order of 5 March 2026." | Read decision |
| 69 | 13/04/2026 | Brook House (Henley) RTM Company Ltd v Henley Brook Ltd (CAM/38UD/LRM/2025/0008) | No unreasonable-conduct costs order made at this stage; any such application required a separate procedure. | 31...it appeared the Respondent had relied upon a large language model to prepare both the counter-notice and the statement of case. He noted that the Respondent had cited legislative provisions inaccurately, including [redacted]... Mr Castle submitted that these errors were indicative not of a litigant in person misunderstanding the law, but of material generated by artificial intelligence, confidently stating entirely untrue facts which could not reasonably be taken from the wording of the authorities named. | Read decision |
| 68 | 24/03/2026 | Nandra and McLeod v Dabbs and Ahmed (LON/00AW/LSC/2025/0813) | Benchmarking evidence was given no weight. Any costs applications were to be determined separately; no AI-specific sanction was imposed in this decision. | The Applicants also state that the Respondents’ comments consist of, in the main, a list of purported “Tribunal Benchmarks” for costs. These were provided without any source or reference, and the Applicants have been unable to verify their origin. The Respondents have also supplied a long list of names of potential alternative service providers with associated cost numbers for the various service charge items. These costs, the Applicants say, are also unevidenced and the Applicants suggest that they are likely to have been fabricated by AI tools. The Respondents also offer approximately 20 legal citations, the vast majority of which cannot be identified by the Applicants from standard sources or in their submission are not relevant to the issue of payability or reasonableness of the service charges for the Property. The Applicants also note that the directions warn explicitly against the citation of irrelevant legal authorities. They add that, with one exception, the Respondents have provided no verifiable alternative quotation or estimate for any service charge item, nor have they provided any evidence of value to assess the reasonableness of any service charge item. The exception is that they have obtained alternative estimates for management fees, but in the Applicants’ view these quotes support the reasonableness of the charges levied by the existing managing agent. | Read decision |
| 67 | 16/03/2026 | Jeffs v London Borough of Lewisham (LON/00AZ/LSC/2025/1032) | Costs protection and reimbursement of application fees refused. The tribunal included apparent AI-related errors among its reasons. | “32…(b) The Applicant has misunderstood the law and relied on inaccurate statements which the Tribunal understands to have originated from AI hallucinations which included false quotes of legislation and cases that were not relevant or did not exist.” | Read decision |
| 66 | 11/03/2026 | Kamal v Tax Policy Associates Ltd and another [2026] EWHC 551 (KB) | Public judicial criticism of hallucinated material in the context of wider findings about the conduct of the litigation. | 189. At a more granular level, the Defendants object to the variety of different addresses, in different jurisdictions, [claimant] has used on his claim form, particulars of claim, application notices and witness statement. They object to his imposition of arbitrary and oppressive deadlines for responses after long periods of delay. They object to a range of features of his correspondence including misapplying the ‘without prejudice’ designation to letters containing no settlement proposals, and the deployment of AI-generated ‘hallucinatory’ references to non-existent cases causing unnecessary effort in cross-checking (there being already a growing jurisprudence of court deprecation of this phenomenon, including in exercise of the Hamid jurisdiction – see R (Ayinde) v Haringey LBC [2025] EWHC 1383 (Admin)). | Read decision |
| 65 | 05/03/2026 | Re A, B, C, D (Extension of assessment; Use of AI: hallucinations) [2026] EWFC 71 (B) | Public criticism and naming of the unregistered barrister, who represented herself; her self-report to the BSB was recorded. No regulatory outcome is established by this judgment. | It is the duty to the court owed by [Representative} as a litigant in person to ensure that the cases cited in legal argument are genuine and provide authority for the proposition advance with which I am concerned. She accepts that she did not do so, and apologises. She says that the mistakes were unintentional, and seeks to justify herself by saying the main authority upon which she relied was sent to me (as it was). Having heard from her, and read her submissions, I absolve her of any intention to mislead the court, but remain concerned that [Representative] minimises the seriousness of misleading the court and goes so far as to assert that criticising use of AI risks setting a harmful precedent for disabled litigants in person and will discourage access to justice. | Read decision |
| 64 | 24/02/2026 | Chevalier-Hernandez v Poplar Dock Management Company Ltd (LON/00BG/LSC/2025/0917) | Untraceable authority withdrawn; warning to check sources. No separate AI-related sanction recorded. | 62. The tribunal is concerned about the proliferation of AI generated “fake” legal authorities and reminds all parties, including litigants in person, of their duty to check that any legal authorities upon which they rely are genuine. Litigants in person can check on: https://onlinelibrary.london.ac.uk/resources/databases/bailii | Read decision |
| 63 | 17/02/2026 | Brightwaters Energy Ltd v Eroton Exploration and Production Company Ltd | Warning about relying on AI without verification. Prompt correction by counsel was expressly acknowledged. | 48. The reference to the 2014 case was the result of AI hallucination. The case does not exist, and the text which it was supposed to contain was invented. That [D's representative] referred to a non-existent authority was the result of the short time he had to prepare, and I acknowledge that he drew the attention of the court to the error promptly after identifying it. The incident does, however, demonstrate vividly the dangers of relying on the product of AI without verification. | Read decision |
| 62 | 15/02/2026 | A v British Transport Police Authority (2602720/2023) | Questioned authorities removed; claimant apologised and stopped using that adviser. The tribunal considered no further action necessary. | "77. The Tribunal have not set out the citations in this judgment because doing so may itself generate further ‘hallucinations." | Read decision |
| 61 | 11/02/2026 | Nwabueze v Simons Rodkin Solicitors LLP and others (3307551/2023; 3301414/2024) | Judicial observation about inaccurate references in an email. The claims were dismissed for the reasons set out in the substantive judgment. | "... We observe, however, that the email appeared to have been drafted with the “assistance” of AI, or Artificial Intelligence, since [redacted case] did not go to the Court of Appeal, (2) there does not appear to be a case with the neutral citation of [redacted neutral citation], and (3) [redacted] was not about an anonymity order or the effect of a settlement agreement on an application for an anonymity order. In any event, the email was as follows.” | Read decision |
| 60 | 30/01/2026 | Professional Standards Authority for Health and Social Care v Nursing and Midwifery Council and another [2026] EWHC 141 (Admin) | Contempt warning; no further step taken against the representative. The nurse’s separate removal from the register concerned the substantive disciplinary appeal, not the AI incident. | “…This was pointed out to him at the hearing. He immediately admitted what he had done and that the references were phantoms created by AI. He promised not to use AI to generate submissions in future and to check his references personally…” | Read decision |
| 59 | 30/01/2026 | Chandler v Chief Constable of Kent Police (2306189/2023) | Reliance on internet AI advice did not justify extending time. The claim was struck out as out of time, not as a penalty for using AI. | "The Claimant was a police officer and therefore should have appreciated the dangers of relying on AI and independently researching the legal requirements. The advice he sets out in his submissions includes a section “Legal Advice: Consider seeking advice from a solicitor or other legal professional if you’re unsure about your rights or the process”. It appears that the Claimant did not seek advice although he references limited advice from the Police Federation in his submissions so presumably received some advice. This is not a legitimate reason for me to extend time for presentation of the claim." | Read decision |
| 58 | 29/01/2026 | Folarin v Immigration Services Commissioner [2026] UKFTT 135 (GRC) | Appeal against refusal of registration dismissed. Unverified AI authorities contributed to concerns about integrity and fitness to advise. | The decision of Dame Victoria Sharp P indicated at paragraph 26 that “Placing false material before the court with the intention that the court treats it as genuine may, depending on the person’s state of knowledge, amount to a contempt. That is because it deliberately interferes with the administration of justice.” Whilst mere negligence would not be sufficient to establish contempt, and knowledge that the information is false or a lack of honest belief that it was true would be required, | Read decision |
| 57 | 27/01/2026 | Aly v Esplen and others [2026] UKFTT 167 (PC) | Warning to obtain, read and explain authorities. The land-registration application failed on its merits; costs applications were left for a later stage. | The reference to case law came about in the following way. Ms Matthews produced a skeleton argument ahead of the hearing, which at first blush presents a reasonable understanding of the law of adverse possession. However, on closer inspection, it does betray some errors. Counsel for the First Respondent suggested it had been generated using Artificial Intelligence (AI) and took objection to this. For her part, Ms Matthews explained that she had no legal background and readily accepted that she had used AI in carrying out her research. | Read decision |
| 56 | 26/01/2026 | Sunder v Royal Mail Ltd and others (1401270/2024; 2226866/2024) | Incorrect passages and references rejected. Reconsideration refused because the applicable threshold was not met. | "The Claimant was unable to explain this ground to me with any clarity. I suspect this was generated by AI. In his submission 28. Who or whatever produced the grounds, it is misconceived. At the May hearing I was not determining or adjudicating any of these claims, I was determining two discreet matters referred to me by Employment Judge Self and case managing the matter to Final Hearing...." | Read decision |
| 55 | 19/01/2026 | Your Home Partners v Kellichan and Hood [2026] SC KDY 34 | False references withdrawn; no contempt proceedings initiated. The claim was dismissed for lack of jurisdiction and competence. | [20] There is no reason why that duty should not also apply to lay persons, as otherwise the management of cases in courts and therefore the course of justice could be significantly obstructed. However, having heard from the claimant, I am satisfied that the claimant has not knowingly attempted to interfere with the administration of justice in this manner, by wasting court time and misleading the court. However, there was a degree of recklessness in the claimant delaying verifying that the references produced by Artificial Intelligence were genuine, until only after the relative submissions were lodged. | Read decision |
| 54 | 16/01/2026 | Hesabi v Gastony International Ltd (LON/00AU/HMG/2025/0636) | Incorrect proposition abandoned when the statute was considered. Application dismissed on its merits; the respondent confirmed it would not seek costs. | "14. The Applicant is of the view that the Property should not have been let until it had a licence, and she believes that she has read something to this effect. She is not legally qualified, and she confirmed that she carried out a general internet for search as part of her preparation for the final hearing. The Tribunal notes that the AI summary to such an internet search can, in the Tribunal’s experience, bring up inaccurate information. When directed to the express wording of the statute, the Applicant did not seek to argue that the defence provided for at section 95(3) of the 2016 Act did not apply on the facts of this case." | Read decision |
| 53 | 16/01/2026 | Huish v HMRC [2026] UKFTT 129 (TC) | The tribunal identified nonexistent cases and expressly attached no blame to the unrepresented appellant. | "...We attach no blame to him, since he is a litigant in person but we have recorded the names so that others do not fall into the same trap.” | Read decision |
| 52 | 12/01/2026 | Green v Imprint Creative Print Solutions Ltd (1809293/2024) | Nonexistent authorities disregarded. No separate AI-related sanction recorded. | "The claimant responded to the respondent’s email of 2 December 2025 raising the time limit issue, by email. The case law she referred to therein could not be identified by the Tribunal or counsel for the respondent. The claimant readily accepted that she had used AI to generate the submissions. I am satisfied that the authorities referred to do not exist and should be disregarded by this Tribunal." | Read decision |
| 51 | 08/01/2026 | Elden v HMRC [2026] UKFTT 41 (TC) | Strike-out application refused; detailed directions required copies of authorities and statements verifying future submissions. A potential regulatory referral was mentioned, not made in this decision. | 93. In further submissions, the Representative said 'The suggestion that citing a published authority amounts to providing false material is misconceived. A court decision is a matter of public record. Whether a case applies is a matter of legal argument and opinion, not misrepresentation. It is entirely proper for parties to put forward different interpretations for the Tribunal to consider. To characterise this as "false material" is both unfounded and inappropriate.' It is not clear who the representative is quoting as saying false material was used. The wording used by HMRC was 'inaccurate use of AI/inaccurate authorities'. | Read decision |
| 50 | 08/01/2026 | AN Concrete Ltd v Wasserman Boxing Ltd (BL O/0010/26) | The hearing officer expressly declined an AI-related sanction. Ordinary scale costs of £2,400 followed the result of the oppositions. | “70 Whilst there have been recent cases in which the courts and the Appointed Person have warned parties of the risks of using artificial intelligence for legal research, drafting skeleton arguments or written submissions, and I sense that [Redacted] might have used some sort of artificial intelligence tool to draft its submissions in lieu, I accept that it might have misunderstood the references, the lack of experience being a mitigation. Hence, I do not think it would be appropriate to apply any sanction in this case” | Read decision |
| 49 | 30/12/2025 | Peiu v Hywel Dda University Local Health Board (1602474/2024; 1604381/2024) | Untraceable and unexplained authorities were not relied upon by the tribunal. | "Within her written submissions, the Claimant had provided a number of cases but was unable to provide citations or references for such cases and had not provided copies. Despite some efforts by Respondent counsel and the Tribunal to find some of the cases, this proved unfruitful. The Claimant was unable to provide references and indicated that she had researched through an AI applications, such as ChatGPT and was unable to provide references or give an indication of how they were relevant. She was informed that unless she was able to do so, they would not be relied upon and were not." | Read decision |
| 48 | 12/12/2025 | De Carvalho Ferreira v Magic Life Ltd and others (3304596/2025) | Warning to verify authorities. Interim relief was dismissed for procedural reasons, not as an AI sanction. | “[Claimant’s Rep] produced during this hearing a skeleton argument in response to the skeleton argument produced by Miss Martin. Amongst other things, this contained the following: “The Respondents’ assertion that the ET1 was premature is a red herring. The principle in [redacted case] provides that a technically premature unfair dismissal claim is treated as presented on the Effective Date of Termination (EDT).” This reference to [redacted case] seemed to me to be potentially relevant to some of the matters I had to decide, but on seeking the case at the reference given (or indeed any other reference) I could not find it. I understand Miss Martin embarked on a similar search without success. I asked [Claimant’s Rep] if she could clarify this reference or provide a copy of the case she had in mind. She could not. I suggested to her that this was sometimes the kind of problem that arose where AI was used in the production of a document, and she accepted that in the limited time she had had she had used AI in the production of the document. In principle there is no objection to the use of AI by litigants, but there is a problem where such AI use produces material that may mislead the tribunal. I urge the claimant and [Claimant’s Rep] to check and verify any material that they use in this case that has been produced by AI, and in particular any references to case law or statute that is produced by AI. The claimant and/or his representative remain accountable for materials they submit to the court, whether prepared with the assistance of AI or not.” | Read decision |
| 47 | 10/12/2025 | Harrison v May trading as Leeds Gymnastics Academy (6000187/2024) | Preparation-time order of £2,178 for wider unreasonable conduct. Unchecked ChatGPT references reinforced the tribunal’s conclusions. | "Those written representations from [Respondent] contained numerous case law references. At least half of those references were non-existent and [Respondent] admitted that he had just used Chat GPT to produce his representations without checking any of the results. He said he was reasonably entitled to conclude that everything that Chat GPT said was reliable, and in fact, he said there was no reason to fact check the internet at all." | Read decision |
| 46 | 09/12/2025 | D (A Child) (Recusal) [2025] EWCA Civ 1570 | Warning about verification; the court expressly absolved the mother of any intention to mislead. | “Finally, I return to the issue raised by the father’s representatives about the mother’s erroneous citation of authority (see in particular paragraph 54 above). I absolve the mother of any intention to mislead the court. Litigants in person are in a difficult position putting forward legal arguments. It is entirely understandable that they should resort to artificial intelligence for help. Used properly and responsibly, artificial intelligence can be of assistance to litigants and lawyers when preparing cases. But it is not an authoritative or infallible body of legal knowledge. There are a growing number of reports of “hallucinations” infecting legal arguments through the citation of cases for propositions for which they are not authority and, in some instances, the citation of cases that do not exist at all. At worst, this may lead to the other parties and the court being misled. In any event, it means that extra time is taken and costs are incurred in cross-checking and correcting the errors. All parties – represented and unrepresented – owe a duty to the court to ensure that cases cited in legal argument are genuine and provide authority for the proposition advanced.” | Read decision |
| 45 | 08/12/2025 | Peggie v Fife Health Board and Upton (4104864/2024) | Certificates of correction amended the judgment. No finding of AI use is established by the judgment or correction certificates. | TBC | Read decision |
| 44 | 05/12/2025 | Udokporo v Enrich International Ltd (BL O/1141/25) | Appeal dismissed; public criticism of fabricated references. An application for full costs was invited, with £750 additional scale costs proposed if none was made. | Litigants-in-person who put their name to a document before the registrar or the Appointed Person must be able to provide all the material cited by them and that material must relate to what they are saying, and likewise any quotation they rely upon must be accurate (albeit I accept that innocent transcription or typographical errors are not representative of improper conduct). If a party cannot provide the authorities they rely upon, their conduct is unreasonable within the meaning of Tribunal Practice Notice 1/2023 and “off-scale” costs are usually appropriate: see Pro Health Solutions (O/559/25), [23]-[24]. | Read decision |
| 43 | 03/12/2025 | Taiwo v Homelets of Bath Ltd and others | Public criticism of false citations and warning that any lawyer responsible could face referral. The broader costs orders concerned the overall litigation. | “…This case does not exist (albeit the bogus reference can be ‘recreated’ through Google’s AI Overview function). There is a 2016 case in the Bolton County Court between the two named parties, but there was no appeal in 2018 to the Court of Appeal and [redacted] is a false reference | Read decision |
| 42 | 02/12/2025 | Harwood-Allen v London United Busways Ltd (2304840/2022) | The tribunal declined to determine whether AI caused the errors. The respondent’s reconsideration application was refused on its merits. | Read decision | |
| 41 | 01/12/2025 | Lockesfield Management Company Ltd v leasehold and freehold owners (LON/00BG/LVL/2024/0601) | Inapplicable authorities disregarded; the tribunal expressly made no criticism of the respondent. | 20. [redacted] confirmed that an internet search had been undertaken as part of her preparation for the hearing which could well have included in the use of generative artificial intelligence, and she accepted the Tribunal’s preliminary view that these legal authorities were not applicable. For these reasons, the legal authorities were disregarded. However, the Tribunal accepts that [redacted] was doing her best as a self-representing litigant who was unaware of the risks of “AI hallucinations” and makes no criticism of her." | Read decision |
| 40 | 24/11/2025 | Oxford Hotel Investments Ltd v Great Yarmouth Borough Council | False quotation rejected; the misleading citation also contributed to distrust of new material. Appeal dismissed on the grounds explained in the judgment. | “…purported to quote at a little length from [18] of the judgment to the effect that a microwave satisfied the statutory definition. The problem is that the real [18] of Barker v Shokar says no such thing. Nor does any other part of the judgment in that case. [Director for the Appellant] ended up accepting that this misleading use of authority was the product of AI. It is one which illustrates again, in courts and tribunals, the dangers of using AI for legal research without any checks.” | Read decision |
| 39 | 21/11/2025 | Jennings v Natwest Group plc [2025] SAC (Civ) 41 | Judicial criticism of confusing submissions and apparently nonexistent cases; the appeal failed on the substantive grounds analysed in the decision. | “[10] These require caution, the appellant having made submissions using ChatGPT, an artificial-intelligence database (see appellant’s supplementary submission). That may explain the generality of the submissions, which largely comprise free-form legal propositions with only limited link to the facts. It has served to complicate and obscure the true analysis of the issues. At least three of the cases cited appear to be non-existent.” | Read decision |
| 38 | 17/11/2025 | 133 Blackstock Road (Hackney) RTM Company Ltd v Assethold Ltd (LON/00AM/LSC/2025/0945) | Respondent debarred after failing to comply with directions to explain authorities. Separate show-cause procedure for unreasonable-conduct costs; no SRA referral ordered in this decision. | “19.The Tribunal is extremely concerned that the Respondent has put material before it that is erroneous. [redacted] has failed to give any explanation as to how this error arose. One explanation might be the use of an AI LLM in the production of the Respondent’s statement of case. | Read decision |
| 37 | 17/11/2025 | UK v Secretary of State for the Home Department [2026] UKUT 81 (IAC), first proceeding | Self-reports to the IAA and SRA recorded; the tribunal made no duplicate referral. | 38. In our judgement, a supervisor who fails to ensure that the work of a more junior fee-earner does not contain false cases or citations is likely to be more culpable than a lawyer who fails to ensure that his own work is free from such “hallucinations”. An individual in the latter camp fails the tribunal, the public and his lay client, whereas an individual in the former camp fails, in addition, to aid the development of more junior lawyers.” | Read decision |
| 36 | 17/11/2025 | R (Munir) v Secretary of State for the Home Department [2026] UKUT 81 (IAC), second proceeding | Referral to the SRA for false citations and failures of supervision; regulatory outcome not determined in this decision. | 38. In our judgement, a supervisor who fails to ensure that the work of a more junior fee-earner does not contain false cases or citations is likely to be more culpable than a lawyer who fails to ensure that his own work is free from such “hallucinations”. An individual in the latter camp fails the tribunal, the public and his lay client, whereas an individual in the former camp fails, in addition, to aid the development of more junior lawyers.” | Read decision |
| 35 | 04/11/2025 | Choksi v IPS Law LLP | Explanatory witness statements required and the incident recorded in a public addendum. The addendum does not record a regulatory referral or separate sanction. | "...contains references to a number of cases that have wrong citations, wrong names or which simply do not exist. A number of the cases cited are wholly irrelevant and do not support the proposition in support of which they are cited...” | Read decision |
| 34 | 30/10/2025 | OscarTech UK Ltd v Orthofix S.r.l. (BL O/1013/25) | Two references withdrawn; warning about AI verification. Ordinary scale costs of £1,250 followed the opposition result. | “106… explained in the hearing that, he had used an AI tool to help him prepare for the hearing. 107. On the matter of the use of AI in legal research, it is appropriate to note that Mr Phillip Johnson, sitting as the Appointed Person in his recent decision BL O/0559/25, underlined the risks associated with AI tools; in particular, the fictitious case references that can be generated; and incorrect authorities cited in support of legal arguments. Mr Johnson made it clear that even litigants-in-person have a duty not to mislead the court [or tribunal] and, in observing that duty, they are urged to be alert to the risks associated with the use of ‘ChatGPT’ and the like.” | Read decision |
| 33 | 23/10/2025 | Victoria Place Flats RTM Company Ltd and others v Assethold Ltd (HAV/00HP/LSC/2024/0523) | Concerns raised with the regional judge. Fee reimbursement and costs-protection orders also made; no regulator referral is recorded. | "85. I then typed the same wording into M365 Copilot on an Android device but adding a question mark at the end which gave a similar response, although the phrasing was markedly different, and it referred to the Upper Tribunal decision cited by [landlord’s managing agent] rather than the ‘hallucinated’ Court of Appeal citation. Repeating the same question sometime later would not re-produce reference to the Upper Tribunal decision, showing that AI adapts and an earlier answer may no longer be returned as the algorithm learns, demonstrating the care that needs to be taking in using AI. The idiom ‘shifting sands’ comes to mind.” | Read decision |
| 32 | 17/10/2025 | Lee and others v Blackpool B&B Ltd and Tang (MAN/00EJ/HMG/2024/0011) | £227 costs order for unreasonable reliance on AI. The tribunal accepted honest belief and permitted continued participation. | "...I can only conclude that the ‘decision’ submitted to the Tribunal is a fabrication – whether or not it is the product of the injudicious use of artificial intelligence tools is unclear.” | Read decision |
| 31 | 14/10/2025 | Ndaryiyumvire v Birmingham City University (L20ZA723) | Wasted-costs liability against the firm; amount left for submissions. Public judgment. No additional individual SRA referral, contempt proceedings or police referral ordered. | 48. I do have to take account of the fact that, as was said in Ayinde, the use of AI is a large and growing problem and the citing of fictitious or fake authorities is a serious threat to the integrity of the justice system which depends upon courts being able to rely on lawyers putting before the courts, whether orally or in documents, accurate material and accurate statements of the law supported by genuine cases. Lawyers who cite fictitious cases must face serious consequences and in the current environment where this is a significant and growing problem, the guidance in Ayinde indicates that judges should take a fairly tough line. | Read decision |
| 30 | 22/10/2025 | Sriram (Mukti Roy) v Brittain | The judge declined to investigate deliberate use of false authorities and made no criticism of the applicant on that point. AI involvement is not established in the judgment. | “…rightly in my view, and I make no criticism of her. For what it is worth, I suspect, that, in common with many unrepresented parties, [Claimant] has resorted to research using the internet and has come up with false leads. The late Muir Hunter was an eminent member of the insolvency bar and the author for many years of an insolvency commentary that still bears his name. It is easy to see how his name could have come up in the course of an internet search and end up wrongly linked to a real case name and reference. The abbreviation BPIR stands for the Bankruptcy and Personal Insolvency Reports. They are not readily available to members of the public. It would have been difficult for [Claimant] to check the citation…” | Read decision |
| 29 | 13/10/2025 | Hassan v ABC International Bank plc (6005897/2024) | Reconsideration refused; £5,881.80 costs excluding VAT ordered. The findings distinguish admitted AI research from suspicion about other narrative drafting. | “On the use of AI in general, I happily accept that the internet is a resource many of us tend to rely on as providing expertise and knowledge where we lack it. Indeed, the facility for using a search engine has even been relied on in the EAT a reason for not granting an extension of time. I accept that AI is now at the forefront of internet searches. It might also be said that more intelligent and proficient users of the internet, like the Claimant, are more apt to use it in the way that the Claimant has i.e. to help construct arguments. I should not, and do not, approach the Claimant’s use of AI as in any way inherently negative” | Read decision |
| 28 | 10/10/2025 | Peters v Driver and Vehicle Standards Agency (3306403/2024) | Inaccurate summaries could not be relied upon; verification took hearing time and contributed to reservation of judgment. | “9. I raise this because: 9.1 An appreciable amount of hearing time was taken up with trying to obtain copies of various reports in order that respondent’s Counsel (and I) could check the accuracy of the AI generated summaries. 9.2 There was a significant risk I could have been misled had this not been done. 9.3 Because of the demonstrated inaccuracies, I was unable to rely on the summaries. 9.4 The delay involved also caused or contributed to my Judgment being reserved.” “…He is genuinely seeking to assist a claimant who would otherwise be unrepresented. Nonetheless, it is important that some basic checks are done to ensure that the material put before the Tribunal is accurate in order to avoid the above. I refer to R (on the application of Ayinde) v London Borough of Haringey [2025] EWHC 1383 which clearly identifies the risk of not undertaking such checks and the importance of doing so…” | Read decision |
| 27 | 06/10/2025 | AK v Secretary of State for the Home Department (UI-2025-002918) | Show-cause direction concerning a possible SRA referral. A referral had not yet been made in the published directions. | "What concerns me in this case is not merely that there were false citations in the grounds of appeal considered by Judge Saffer; it is that those false citations were then removed from the grounds of appeal which were placed in the composite bundle. The former actions are unprofessional, the latter are potentially dishonest because it suggests that there was an attempt to conceal the false citations..." | Read decision |
| 26 | 03/10/2025 | Warwick Econometrics Ltd v University of Warwick (BL O/0938/25) | Inconsistent explanations and untraceable case recorded. Applications treated as abandoned for failure to file a defence in time; not an AI-specific penalty. | “I then asked [Representative] if artificial intelligence or any form of artificial intelligence, such as Chat GPT, was used to produce the skeleton arguments of both the 3 September 2025 and the corrected skeleton argument of 4 September 2025. [Representative] denied this by simply stating “no”.” [paragraph 30] | Read decision |
| 25 | 30/09/2025 | Mboungueng v Severn Trent Water Ltd (1306955/2023) | Warning that repetition could lead to strike-out; no immediate AI-related sanction. The set-aside/reconsideration application was refused on its own grounds. | 30. Either way, there was a risk that the Tribunal would be misled. Moreover, either way the Tribunal – I – have had to spend a great deal of time trying to find a nonexistent case in order to satisfy myself that it definitely does not exist and even more time reading (or re-reading) from beginning to end the other cases the Claimant has referred to in order to make sure that there was not something in them, tucked away somewhere, that I was not previously aware of and that supports the Claimant’s submissions about them. It is not usually necessary for the Tribunal to do this, because normally people refer to cases reasonably accurately. If they refer to one I am familiar with, I don’t need to look it up in order to know that the reference is accurate (for example, the Claimant’s references to [case names redacted] in his email of 18 August 2025). If they refer to one I am not familiar with, I can look it up and quickly see that it says what it is alleged to say. | Read decision |
| 24 | 29/09/2025 | ANPV and SAPV v Secretary of State for the Home Department (UI-2025-003373; UI-2025-003374) | Show-cause order concerning a possible BSB referral; the decision does not record a completed referral. | “…suggested that the inaccuracies in the grounds were as a result of his drafting style. He accepted that there might have been some “confusion and vagueness” on his part; that he might “need to construct sentences in a more liberal way”; and that his drafting should perhaps “be a little more generous” when it came to making specific allegations about judges overlooking or failing to follow binding authorities. … The problems which I have detailed above are not matters of drafting style. The authorities which were cited in the grounds either did not exist or did not support the grounds of which were advanced. Where the cases did exist, they were often wholly irrelevant to the proposition of law which was given in the grounds.” (paragraphs 63 and 64) | Read decision |
| 23 | 08/09/2025 | Sulić v Marques (BL O/0820/25) | Public criticism of confused grounds; appeal considered on its merits. No separate AI-specific sanction recorded. | “6. The Grounds of Appeal in this case are entirely unsatisfactory. They make statements which mix up legal and factual matters, rely on irrelevant things, and the sole reference to a statutory provision is to one which does not exist. It does not matter whether these were drafted by the Appellant without any assistance or some form of generative AI was used. It is not acceptable, even from a litigant in person, to provide grounds which makes the task of appellate review so difficult. Indeed, the difficulty was made more profound because neither side provided any written submissions and both declined a hearing. This means this appeal has to be determined based on the Appellant’s confused set of criticisms of the Hearing Officer’s decision. 7. Nevertheless, I will address the Appellant’s challenges to the decision as far as is possible from the Grounds of Appeal as drafted.” | Read decision |
| 22 | 15/08/2025 | Kuzniar v General Dental Council (6009997/2024) | Unreasonable conduct found, but the tribunal declined to award costs, accepting honesty and efforts to correct the error. | "44. The Claimant explained that the problems arose from her using AI to carry out research. She had previously used AI/ChatGP to carry out research without problems in her litigation against Roxdent Ltd and so she expected to be able to do so again successfully in the instant case. She did not know about the problems with the citations when she told the Respondent’s solicitors about them, and when she found out about them, she did her best within the short time available to mitigate or reduce the problem. She did not act in bad faith or with any intent to place false information before the Tribunal. I accept this explanation. 45. The Claimant conducted the claim unreasonably as described above by referring to the Respondent a large number of nonsensical and in many cases non-existent citations without taking any or sufficient care to check them first. By not doing so she passed the work of checking them to the Respondent to have to do at short notice. My discretion to award costs is engaged. 46. However, I decline to award costs because AI is a relatively new tool which the public is still getting used to, the Claimant acted honestly (and furthermore has presented her case honestly to me over the last two days), and she tried to her best to rectify the situation as soon as she became aware of her mistake." | Read decision |
| 21 | 12/08/2025 | Holloway v Beckles and Beckles (CHI/24UD/LBC/2024/0006) | The errors reduced the respondents’ credibility. The tribunal found the inclusion of fake cases was not intentional. | "That leaves the matter of the fake cases. The Tribunal finds that this does amount to unreasonable conduct within rule 13(1)(b). It has decided that the misconduct is serious, being conduct that undermines civil litigation in the Tribunal. Therefore, the Tribunal determines that it should "make a costs order. It considers that the costs order should be proportionate to the additional costs caused. It has decided that the appropriate quantum is half the costs of counsel’s fees in attending the hearing of 14 May 2025. These amount to £750 and must be paid to the applicant within 28 days." | Read decision |
| 20 | 30/07/2025 | The Father v The Mother and others [2025] EWHC 2135 (Fam) | £5,900.63 costs order for wider litigation conduct, including reliance on unchecked fake cases. The whole order must not be attributed solely to AI. | “(16) The F then made a further application on a C2 asking that HHJ Bailey recuse herself on the basis of being biased against him and her not understanding ASD and the impacts of his diagnosis. This came before the Judge on 10 June 2025. In his written application to the court the F referred to a number of previous authorities, in particular relating to ASD. HHJ Bailey realised that many of these cases were not genuine, and the submission appeared to have been generated by Artificial Intelligence (“AI”). In light of the level of recent concern about litigants and lawyers using AI and referring to cases which are not genuine (as reflected in the Divisional Court decision R (Ayinde) v London Borough of Haringey [2025] EWHC 1383), HHJ Bailey referred the case to me as the Family Presiding Judge for the Midlands.” “The F relied upon faked cases without apparently making any effort to check their veracity. It is in my view important to note that the F is someone who is well capable of checking references and ensuring documents are accurate if it is in his interests to do so.” | Read decision |
| 19 | 30/07/2025 | Ilunga v Portico Cooperate Reception Management Ltd (2305959/2023) | Legal propositions in written closing submissions disregarded; factual and evidential submissions retained. Potential costs were discussed, not ordered in this passage. | "The claimant decided that she would either prefer to withdraw her written closing submissions or not rely on any of the assertions as to the law in those submissions. I determined that the fairest course of action to both parties was for me to have regard to the claimant’s submissions, insofar as they advanced factual / evidential points, but that I would disregard any legal principles asserted. I also heard oral closing submissions from the claimant and Mr Green. I reserved my decision as we had reached the end of Day 4 by this time.” | Read decision |
| 18 | 25/07/2025 | Chanda v Royal Mail Group Ltd (3311062/2023) | False citations amounted to unreasonable conduct, but the respondent’s strike-out application was refused. | “14. Whether generated using AI or not, we find that false citations of authority amounts to unreasonable conduct of proceedings.” | Read decision |
| 17 | 24/07/2025 | HMRC v Gunnarsson [2025] UKUT 247 (TCC) | Erroneous passages withdrawn; warning. The tribunal did not consider the respondent highly culpable. | "113. In this case, HMRC was put to the trouble of having to investigate the existence of the purported decisions relied upon by the Respondent. Fortunately, they did so. Depending on the circumstances, there may be occasions when the opposing party or the tribunal are not able to discover the errors relied upon. There may be others where an adjournment is required to investigate or address the inaccurate information. 114. On these facts, we do not consider the Respondent to be highly culpable because he is not legally trained or qualified, not subject to the same duties as a regulated lawyer or other professional representative and may not have understood that the information and submissions presented were not simply unreliable but fictitious. He was under time pressure given his other competing responsibilities and doing his best as a lay litigant seeking to assist the UT by preparing written submissions." | Read decision |
| 16 | 14/07/2025 | PS v London Borough of Wandsworth [2025] UKUT 239 (AAC) | Unreliable references did not assist the legal argument. The anonymity order remained in force; no separate AI-related sanction recorded. | We did not discuss these references at the hearing as I considered it unnecessary to do so as in my judgment the appellants arguments could properly be advanced without those references, but the local authority did raise the non-existence of these cases in its late written submissions. It may be these legal references were the product of AI generation as it is well known that AI ‘hallucinates’ the names of legal cases and legislation.” | Read decision |
| 15 | 07/07/2025 | Various Leaseholders of Napier House v Assethold Ltd (CAM/00KA/LSC/2023/0063 and linked references) | Permission to appeal refused on the grounds analysed in the decision; unnecessary research burden noted. | “15. The Respondent included two cases within their grounds for appeal which have been cited as…[False Case names] Having performed a search on BAILLI, Westlaw and Find Case Law, it has not been possible to find …[case names omitted. It may be that this case is not authentic and AI may have been used to reference this case….” On another case, the court noted the decision concerned the circumstances in which a parole board should hold an oral hearing. “When reading the full judgment it is difficult to see why the tribunal has been referred to this case…..” | Read decision |
| 14 | 01/07/2025 | MS (Bangladesh) v Secretary of State for the Home Department (UI-2025-001831) | Written explanations directed after an inaccurate citation was queried; no regulatory sanction recorded in this decision. | 13. We sought clarification regarding this citation and reference and asked for the relevant paragraph of the judgment being relied on. [counsel] was not able to specify this. [counsel] submitted that he understood, having used ChatGBT, that the Court of Appeal in Y (China) [2010] EWCA Civ 116 was presided by Pill LJ, LJ Sullivan LJ and Sir Paul Kennedy. However, the citation [2010] EWCA Civ 116 did not point to the case of Y (China) but to R (on the application of YH) v SSHD. We raised concern about this and referred [counsel] to the recent decision of the President of King’s Bench Division in Ayinde [2025] EWHC 1383 (Admin) on the use of Artificial Intelligence and fictitious cases, and directed him to make separate representations in writing. 14. In his subsequent written representations, [counsel] clarified that Y(China) was a typological error and he sought to rely on R (on the application of YH) v SSHD [2010] EWCA Civ 116 where, when discussing the meaning of ‘anxious scrutiny’ in asylum claims…” | Read decision |
| 13 | 20/06/2025 | Pro Health Solutions Ltd v ProHealth Inc (BL O/0559/25) | No appeal costs order, in light of conduct on both sides. No IPReg referral. The earlier hearing-officer costs order remained. | "As identified in Ayinde (including in the Appendix setting out domestic and overseas examples of attempts to rely on fake citations), fabrication of citations can involve making up a case entirely, making up quotes and attributing them to a real case, and also making up a legal proposition and attributing it to a real case even though the case is not relevant to the legal proposition being made (for instance, it deals with a completely different issue or area of law). It is not, however, fabrication to make an honest mistake as to what a court held in a particular case or to be genuinely mistaken as to the effect of a court’s judgment. In any event, it does not matter whether fabrication was arrived at with or without the aid of generative artificial intelligence. I therefore need to consider what if any sanction is appropriate.” | Read decision |
| 12 | 18/06/2025 | UB v Secretary of State for the Home Department (UI-2025-000834) | Explanation and remedial training accepted; no further action. The tribunal expressly said the drafting concerns played no part in the appeal’s merits. | “…recognised this seriousness of this issue and has taken commendable steps to ensure it will not be repeated including (i) meeting with the caseworker who drafted the Grounds; (ii) holding a partners’ meeting to discuss adopting an AI policy and assigning the task of finalising an AI policy to a colleague in consultation with an AI professional; (iii) conducting relevant in-house training and issuing interim AI Guidance and (iv) planning for comprehensive staff training by an AI professional….” | Read decision |
| 11 | 06/06/2025 | Al-Haroun v Qatar National Bank QPSC and another [2025] EWHC 1383 (Admin) and previous order | Solicitor referred to the SRA, in addition to self-referral. No contempt proceedings and no court referral of reviewing counsel to the BSB. | "In CL-2024-000435, it appears from the Order of Mrs Justice Dias that correspondence was sent to the court, and witness statements were filed, citing authorities that do not exist and claiming that other authorities contained passages that they do not contain" Rt Hon. Dame Victoria Sharp | Read decision and earlier judgment |
| 10 | 06/06/2025 | R (Ayinde) v London Borough of Haringey [2025] EWHC 1383 (Admin) and [2025] EWHC 1040 (Admin) (Richie J) | BSB and SRA referrals; earlier wasted-costs orders recorded. The court declined to initiate contempt proceedings. | “ It is such a professional shame. The submission was a good one. The medical evidence was strong. The ground was potentially good. Why put a fake case in?” “I should say it is the responsibility of the legal team, including the solicitors, to see that the statement of facts and grounds are correct.” “…I consider that it would have been negligent for this barrister, if she used AI and did not check it, to put that text into her pleading.” Mr Justice Ritchie | Read decision and earlier judgment |
| 9 | 13/05/2025 | Bandla v Solicitors Regulation Authority [2025] EWHC 1167 (Admin) | Grounds struck out as abuse of process; indemnity costs of £24,727.20 ordered. His original strike-off predated this appeal and was not caused by the citation incident. | “I asked the Appellant why, in the light of this citation of non-existent authorities, the Court should not of its own motion strike out the grounds of appeal in this case, as being an abuse of the process of the Court. His answer was as follows. He claimed that the substance of the points which were being put forward in the grounds of appeal were sound, even if the authority which was being cited for those points did not exist. He was saying, on that basis, that the citation of non-existent (fake) authorities would not be a sufficient basis to concern the Court, at least to the extent of taking that course. I was wholly unpersuaded by that answer. In my judgment, the Court needs to take decisive action to protect the integrity of its processes against any citation of fake authority. There have been multiple examples of fake authorities cited by the Appellant to the Court, in these proceedings. They are non-existent cases. Here, moreover, they have been put forward by someone who was previously a practising solicitor. The citations were included, and maintained, in formal documents before the Court. They were never withdrawn. They were never explained. That, notwithstanding that they were pointed out by the SRA, well ahead of this hearing. This, in my judgment, constitutes a set of circumstances in which I should exercise – and so I will exercise – the power of the Court to strike out the grounds of appeal in this case as an abuse of process.” | Read decision |
| 8 | 22/04/2025 | Goshen Multiservice Ltd v Accuro Environmental Ltd (2304373/2024) | Untraceable authority could not be applied. No separate AI-related sanction recorded. | "...I cannot find such a case, and I am left wondering whether this case is an invention by the claimant or perhaps an artificial intelligence platform. As I explained in the hearing, I cannot apply authority which I have not seen. " | Read decision |
| 7 | 10/04/2025 | Unnamed County Court proceeding before HHJ Andrew Holmes, reported in Ayinde | Judge wrote to head of chambers; no immediate BSB referral after assurances. The incident was later included within the Divisional Court’s referral concerns. | “That was a case before the County Court … That counsel drew attention to the fact that the application before the judge contained false material: specifically the grounds of appeal and the skeleton argument settled … contained references to a number of cases that do not exist….” | Read decision |
| 6 | 03/04/2025 | Zzaman v HMRC [2025] UKFTT 539 (TC) | Warning to verify AI output; appeal dismissed on its merits. The tribunal expressly distinguished the errors from invented cases. | 29. However, our conclusion was that Mr Zzaman's statement of case, written with the assistance of AI, did not provide grounds for allowing his appeal. Although some of the case citations in Mr Zzaman's statement were inaccurate, the use of AI did not appear to have led to the citing of fictitious cases (in contrast to what had happened in Felicity Harber v HMRC [2023] UKFTT 1007 (TC) ). But our conclusion was that the cases cited did not provide authority for the propositions that were advanced. This highlights the dangers of reliance on AI tools without human checks to confirm that assertions the tool is generating are accurate. Litigants using AI tools for legal research would be well advised to check carefully what it produces and any authorities that are referenced. These tools may not have access to the authorities required to produce an accurate answer, may not fully "understand" what is being asked or may miss relevant materials. When this happens, AI tools may produce an answer that seems plausible, but which is not accurate. These tools may create fake authorities (as seemed to be the case in Harber ) or use the names of cases to which it does have access but which are not relevant to the answer being sought (as was the case in this appeal). There is no reliable way to stop this, but the dangers can be reduced by the use of clear prompts, asking the tool to cite specific paragraphs of authorities (so that it is easy to check if the paragraphs support the argument advanced), checking to see the tool has access to live internet data, asking the tool not to provide an answer if it is not sure and asking the tool for information on the shortcomings of the case being advanced. Otherwise there is a significant danger that the use of an AI tool may lead to material being put before the court that serves no one well, since it raises the expectations of litigants and wastes the court's time and that of opposing parties. | Read decision |
| 5 | 16/01/2025 | Olsen v Finansiel Stabilitet A/S [2025] EWHC 42 (KB) | No contempt summons issued. The fabricated case could affect a later costs decision; no final AI-related costs order recorded here. | "I have narrowly and somewhat reluctantly come to the conclusion that I should not cause a summons for contempt of court to be issued to the appellants under CPR rule 81.6. I do not think it likely that a judge (whether myself or another judge) could be sure, to the criminal standard of proof, that the appellants knew the case summary was a fake. They may have known but they could not be compelled to answer questions about the identity of the person who supplied it." Mr Justice Kerr | Read decision |
| 4 | 06/12/2024 | Crypto Open Patent Alliance v Wright | False citations were not considered further because the relevant principles were clear. No hallucination-specific sanction recorded in this decision. | “…referred to a series of authorities in support of arguments that reasonable adjustments should be made to enable a vulnerable litigant or witness to participate fairly in court proceedings. As COPA pointed out by reference to a series of examples, most of the authorities he has cited do not contain the passages attributed to them (or anything like those passages), and indeed most have nothing to do with adjustments for vulnerable witnesses. COPA suggested that it seems likely that they are AI “hallucinations” by ChatGPT (i.e. made-up references) rather than deliberately misleading inventions by Dr Wright. However, since the principles are clear and not in doubt, as set out above, it is not necessary to engage with his false citations any further.” | Read decision |
| 3 | 15/06/2024 | Mallon v Steer Energy Solutions Ltd (1403759/2021) | Earlier costs decision maintained on reconsideration in the context of wider unreasonable litigation conduct. AI-generated correspondence added time and cost. | "49. Whilst the claimant may believe that such an approach helps him, as he merely types in a question and then forwards the responses, the fact is those submissions simply generate a lot of work and confusion, and are often highly unreliable, literally creating non-existent case law and precedent. " | Read decision |
| 2 | 04/12/2023 | Harber v HMRC [2023] UKFTT 1007 (TC) | No AI-related penalty. The appeal failed on its merits; the tribunal expressly said reliance on fake cases did not affect that result. | “But that does not mean that citing invented judgments is harmless. It causes the Tribunal and HMRC to waste time and public money, and this reduces the resources available to progress the cases of other court users who are waiting for their appeals to be determined. As Judge Kastel said, the practice also "promotes cynicism" about judicial precedents, and this is important, because the use of precedent is "a cornerstone of our legal system" and "an indispensable foundation upon which to decide what is the law and its application to individual cases" | Read decision |
| 1 | 29/05/2023 | Unnamed Manchester civil case (Law Society Gazette report, 29 May 2023) | The Gazette reports that the judge accepted inadvertence and did not penalise the litigant. No published judgment has been verified. | N/A | Read report |
