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 Law.
The Complete Legal Timeline, How Pervasive is the Problem in Ireland and Recent Workplace Relations Commission Guidance
Introduction
This legal article is jointly written by Tomás Keys and me. Tomás is a Barrister practising from The Law Library in, Ireland. He recently authored an article for The Bar Review in October 2025 titled “Artificial Intelligence: What Happens When AI Goes Rogue” where he examined the impact of the increased use of generative AI within the legal profession and the repercussions. He also addressed three of the cases below.
The use and misuse of generative AI and the consequent false citations/AI hallucinations which are produced continues apace in many common law jurisdictions.
While many of the cases highlighted on the Natural and Artificial Intelligence in Law blog relate primarily to the principal author’s jurisdiction of practice, namely England and Wales, the blog also examines international alleged incidents of false citations or AI hallucinations highlighting instances of such occurrences in litigation across other jurisdictions.
This week’s legal article examines cases in which alleged false citations, whether suspected AI assistance or otherwise, have been presented before the courts and tribunals of Ireland, exploring how and why these incidents occurred and what role, if any, generative AI played. Two earlier cases are also noted for context, as they appear to mark the first mention of artificial intelligence in separate Irish tribunals, though they did not involve any alleged false citations or fabrications. As always, this is not legal advice, and readers are encouraged to consult the full judgments for context, as the authors of this article can only comment on what is contained in the written decisions.
Before addressing the Irish cases involving false citations, two cases deserve mention as they appear to mark the first recorded reference to artificial intelligence in quais judicial tribunals namely the Valuation Tribunal and Workplace Relations Commission respectively
Cullinan Funeral Directors v Commissioner of Valuation (no alleged false citations/AI hallucinations)
[Appeal No: VA23/5/0250] (Valuation Tribunal John Stewart, 24 Jun 2024)
It is mentioned here solely because the applicant had used ChatGPT to query whether a funeral parlour or a funeral home is defined as a shop for valuation purposes. For that reason, it is not included in the count of hallucination or fabrication cases.
Varghese v Viatel Technology Ltd (no alleged false citations/AI hallucinations)
[ADJ-00050448] (Workplace Relations Commission, Orla Jones, 9 Aug 2024)
This case does not involve any false citations or fabrications. It is mentioned here solely because it includes an early reference to the use of AI, where it was merely suspected that AI had assisted in preparing a document. For that reason, it is not included in the count of hallucination or fabrication cases.
Coulston & Ors v Elliott & Anor (First Suspected Incident)
[2024] IEHC 697 (HC, Nolan J, 10 Dec 2024).
The main claim was a possession claim by a receiver appointed over the defendants’ shop and dwelling. However, from paragraph 85 the court was concerned by a new argument in written submissions. A litigant in person informed the court that he was unable to explain one aspect of his submissions as “a friend” had helped him prepare them. Nolan J. formed the view that it was highly likely that either the litigant in person went to a person who purportedly held themselves out as a lawyer and was not, or that he used an AI generative tool:
“When I asked the First Named Defendant where this argument came from, he confirmed that he had asked a friend to prepare the submissions. This has caused the court significant disruption, since I made it clear that no new issue should be raised in his submissions. When I asked him to explain the submissions, he said that he could not. It seems to me therefore that it is highly likely that one of two things happened, either the Defendants went to somebody who purported to be a lawyer or they used a generative AI program to generate the submissions.” (Paragraph 86)
The court noted that if the Defendant went to somebody purporting to be a solicitor or barrister, and they were not, this would amount to a criminal offence. On the other hand:
“… if they used a generative AI program, they have been fooled. Such programs often sound persuasive but can be fatally flawed. Sadly, in this case the argument was indeed fatally flawed. The general public should be warned against the use of generative AI devices and programs in matters of law…” (Paragraph 87)
Reddan v An Bord Pleanála (Approved) (Second Suspected Incident )
[2025] IEHC 172 (HC, Nolan J, 13 Mar 2025)
This was an application to bring judicial review proceedings against the decision to grant planning permission relating to a golf club. When considering ground 6, Nolan J. was highly critical of an unfounded allegation made by a litigant in person that an architect had committed perjury and noted the phrase “subordination to perjury” was unfamiliar in Irish courts. Nolan J observed:
“…When challenged, he says it was something he discovered during his online research. This sounds like something that derived from an artificial intelligence source. It has all the hallmarks of ChatGPT, or some similar AI tool…” Paragraph 78
Nolan J was also unimpressed by the seventh ground:
“This ground looks as if it is a cut and paste from some form of pseudo legal online text. The fact that it contains his name suggests to me that it is being generated by a third party or, as I indicated above, by some form of artificial intelligence tool”
Malone & McEvoy v Laois County Council et all (Third Suspected Incident)
[2025] IEHC 345 (HC, Holland J, 23 Jun 2025)
In this case, litigants in person sought an injunction to restrain alleged unauthorised development and require remediation at a quarry and concrete products facility. They also advanced EU‑law declaratory claims. An issue arose where the litigant in person provided written legal submissions on the day of the trial, which included quotation marks and italics to convey that the text was from a case of the Court of Justice of the European Union (CJEU). Given that the submissions had only been provided on the date of trial, nobody had an opportunity to check whether the quotation was accurate. The judge was unable to find the quotation in the reference case or any case. Upon further enquiries, the litigant in person confirmed by email that it was “not a direct quotation from a single source but a well-established concept reflected in multiple decisions of the…[CJEU]”. When the judge sought further clarification on who or what had produced the text placed in quotation marks, the party:
“..did not reply. I refrain from inferring how it was generated.
42.It is necessary to be clear: it is not acceptable to depict text in written submissions as a verbatim quotation from an authority where it is not such. A similar action by a professional lawyer would be misconduct –see, for example, the recent and somewhat analogous case of Ayinde. The principle is essentially the same -though I hasten to say that I would not push the analogy too far as to a factual comparison of the present case with that case and the error in the present case is not of the order of the misconduct that case. However, appreciable judicial time was wasted on the issue –not least trying to find the source of the quotation. And it does illustrate:
- The vital importance of precision and accuracy in written submissions. That duty lies on lay litigants as much as on lawyers.
- That text in submissions formatted so as to convey that it is a direct and verbatim quotation from an identified source must be exactly that. Of course, it is permissible to edit the text (for example to exclude irrelevant content or by underlining for emphasis) but, if so, that it has been done must be apparent on the face of the document.
- That opposing parties are entitled to written submissions in good time to check them.”
Oliveira v Ryanair DAC (Fourth Suspected Incident “may have used AI”)
[ADJ-00055225] (Workplace Relations Commission, Patricia Owens, 1 Oct 2025)
A recent decision of the Workplace Relations Commission (the body which hears employment disputes at first instance), called Oliveira v Ryanair DAC [ADJ-00055225], appears to be the first written decision in Ireland where a party has expressly admitted they may have used AI.
In the case, the Complainant who was self-represented submitted a complaint under Section 77 of the Employment Equality Acts (1998-2015) that he was subject to discrimination.
He alleged that during the course of his employment with the Respondent as a cabin crew member that he was subject to discrimination on race and family grounds as well as alleging harassment and sexual harassment in the workplace.
The matter was heard over the course of two days in the workplace relations commission. The complaint was ultimately refused on the basis that the Complainant had failed to establish a prima facie case of discrimination and where he failed to produce cogent evidence to support claims of victimisation, harassment and sexual harassment.
Allegation of use of Artificial Intelligence
The respondent raised concerns about the complainants submissions. In particular the Respondent suggested:
That it appeared that the Complainant’s submissions may have been generated with the assistance of Artificial Intelligence. This is particularly so where the case citations relied upon do not appear to give the outcome which the Complainant relied upon.
That the first case the Complainant relied upon is DEC-E2016-105 and he asserted that it related to racial harassment. It was a claim for discriminatory dismissal and made no reference to harassment save by reference to another authority relied upon in the decision. The Complainant asserted that €15,000 was awarded, however, the case was dismissed for want of jurisdiction.
That the second case ADJ-00028765, was a constructive dismissal case, which was again dismissed.
That ADJ-00021453 related to an industrial relations claim related to wages and was dismissed as the Complainant did not appear.
That ADJ-00039821 and ADJ-00040112 appeared to be phantom determinations.
That ADJ-00030217 was a claim for penalisation and did not appear to contain the word “warn” or “warning” whatsoever. Notably there was no finding of penalisation.
That ADJ-00035678 was once again an Industrial Relations complaint and at it’s height related to a claim that sick leave was due to issues in the Complainant’s employment. Even if it did apply the Adjudicator noted “This is a unique gesture particular to these unique set of circumstances and without precedent.”
That ADJ-00029543 was a case related to alleged harassment on the basis of religion. Harassment was not established and no award was made.
That ADJ-00026789 was a case related to a multitude of claims in a similar vein to the Complainant’s case. Despite the various limbs to the claim no breach of the Employment Equality Act was determined to have occurred and no order for redress was made. “Once again there was no reference to searches whatsoever…”
Counsel for the Respondent also
“…raised concern that much of the submission provided appeared to be AI generated and was used by the Complainant to mislead the hearing and to create a myriad of complaints.”
Initially, the Complainant responded to the suggestion that the AI may have been used to generate written submissions by suggesting that the respondent’s insinuation that the complainant’s submissions “may have been generated with the assistance of artificial intelligence” was a baseless, unprofessional, and ad hominem attack designed to distract from the merits of the case. The substance of the legal arguments and the veracity of the evidence are paramount, not the method of [drafting].
However, on the second day of the hearing the adjudication officer noted that the Complainant
“…acknowledged that he may have used AI and became defensive about his use…”
The adjudication officer was less than impressed with the actions of the self represented litigant and while not criticising the use of AI per se, did criticise the product of said use where she stated:
“While I’m not particularly concerned about whether the Complainant used AI or not I am clear that parties making submissions to the WRC have an obligation to ensure that their submissions are relevant and accurate and do not set out to mislead either the other party or the Adjudication Officer. These submissions were rife with citations that were not relevant, mis-quoted and in many instances, non-existent. The Complainant wasted a considerable amount of time of the Respondent and the Adjudication Officer in seeking to establish the veracity or otherwise of legal citations.
The Complainants attempts to bring in new allegations and claims late in the day and to seek to rely on phantom citations to support his claims can only be described as egregious and an abuse of process.”
However, because the issue was not vital to the case nor was the underlying proposition controversial, the adjudicator was inclined to accept that there was no attempt or intention or mislead and accepted the apology. It did not affect the outcome.
Conclusion
Overall, it appears that there have been four reported incidents of false citations where it was suspected generative AI was involved in the court/tribunals in Ireland. While no practice directions have yet been published for the courts in Ireland, with regard to the use of AI in litigation, the issues as identified in Ayinde are well known and indeed were expressly cited in Malone & McEvoy v Laois County Council above.
On 30th October 2025 the Workplace Relations Commission published guidance for litigants on the use of AI tools to prepare materials for submissions to the Workplace Relations Commission.
Both the Law Society of Ireland[1] and the Professional Practices Committee of the Bar of Ireland have published guidance for its members on the use and misuse of generative AI.
The guidance states:
These tools may help you draft text or explain concepts, but they should not be relied upon as legal advice. You should always check that you are relying on the most up-to-date law. (emphasis in original text)
The guidance does not prohibit the use of AI in preparation for a hearing but instead states:
If you use an AI tool in preparation for a WRC hearing, you should:
- Double-check all legal content: Make sure case names, legislation, and principles are real, relevant, and from Irish law.
- Understand what you are submitting: Do not include material you do not fully understand or cannot explain if questioned or cross examined in the hearing.
- Avoid including sensitive personal data: particularly in public AI tools (e.g. names, PPS numbers, health details).
- Do not rely on AI for legal strategy or outcomes: AI cannot assess the strengths or weaknesses of your specific case.
While the Workplace Relations Commission does not require a disclosure statement that AI has been used in the preparation of submissions, the guidance suggests that parties may wish to include an optional statement in the submission such as:
“Parts of this submission were drafted using an AI writing tool. I have reviewed and confirmed the accuracy of all content.”
The guidance also suggests that if parties are citing legal authorities that are not included on the Workplace Relations Commissions’ own core list of authorities that the party provides copies of the decisions referred to or hyperlinks to the decisions.
Interestingly, the guidance from the Workplace Relations Commission discloses that AI was used in part in the preparation of the guidance:
“The WRC is committed to identifying ethical, responsible and meaningful use of AI capabilities. We will be transparent as we explore, adopt and integrate AI technology to benefit staff and external stakeholders. To demonstrate this commitment and how AI can be used usefully with transparency and responsibility, parts of this guidance were drafted using AI to assist with layout and formatting.”
Even without formal practice directions from the Irish courts, the obligations on lawyers is very clear. As stated recently by the Court of Appeal of Louisiana (Fifth Circuit) in the case of In Re Sanctions Order of Kenney (no. 25-C-389) delivered on 23 October 2025
“…read the law you cite. Read the Code. Read the statutes. Read the cases. Citation to fake or fabricated cases in pleadings is prima facie evidence that the attorney who signs the pleadings has failed her duty…”
If you find these legal articles interesting, please consider subscribing below:
[1] https://lsi.ent.sirsidynix.net.uk/client/search/asset/1065685
