Irish WRC Issues Strict Rules on AI Use in Legal Submissions

Irish Tribunal Issues AI Warning After ‘Hallucinated’ Cases Cited

Ireland’s employment tribunal, the Workplace Relations Commission (WRC), has issued a stern and timely warning on the use of Artificial Intelligence in preparing legal submissions, following a high-profile case where AI-generated documents cited entirely fabricated legal precedents. The formal guidance was prompted by the recent case of Fernando Oliveira v Ryanair DAC, which brought the perils of unchecked reliance on AI into sharp focus and has sent a clear message to litigants: you are wholly responsible for the content you submit, regardless of its origin.

The incident, which gained significant media attention, involved a former cabin crew member whose submissions against the airline were drafted using an AI tool. The Adjudication Officer presiding over the case criticised the submission for containing “citations that were not relevant, mis-quoted and in many instances, non-existent.” This led to what the officer described as a considerable waste of time for both the tribunal and the opposing party, as they were forced to verify legal citations that were, in effect, digital ghosts created by the AI model.

The Perils of Digital Delusion

At the heart of the WRC’s guidance is a warning against the phenomenon known as ‘AI hallucination’. This occurs when a large language model, in its attempt to provide a coherent and confident-sounding answer, invents facts, figures, or, in this instance, legal case law. The guidance explicitly states that AI tools are not trained specifically on the nuances of Irish employment and equality law or the unique procedures of the WRC. Consequently, they can produce arguments that, while appearing plausible, do not reflect the legal reality in Ireland.

In the landmark Ryanair case, at least two of the case citations relied upon by the complainant, ADJ-00039821 and ADJ-00040112, were found to be complete fabrications with no corresponding WRC decision. The guidance note cautions that such inaccuracies can severely undermine a party’s arguments and credibility. It stresses that AI-generated text should not be mistaken for legal advice, as the technology lacks the critical reasoning and jurisdictional awareness of a qualified legal professional.

Confidentiality and Ultimate Responsibility

Beyond the risk of factual inaccuracy, the WRC highlights a critical concern regarding data privacy and confidentiality. The guidance warns that inputting sensitive personal data or commercially sensitive information into public AI tools is fraught with risk, as these platforms may store or use the provided information for their own purposes, such as model training. This could lead to inadvertent and serious data breaches.

Crucially, the guidance document makes it unequivocally clear that accountability rests solely with the party submitting the documents. The WRC will treat all submissions as the party’s own work, and any errors, misleading statements, or fabricated information will be attributed directly to them. The defence that “the AI did it” will hold no water. This principle of ultimate responsibility means that litigants must meticulously review, verify, and be prepared to stand over every single claim and citation in their submissions, whether drafted by themselves, a legal representative, or an algorithm.

A Call for Transparency

In a move towards promoting greater transparency and assisting Adjudication Officers, the WRC has proposed an optional disclosure statement for parties who utilise AI in their preparations. While not mandatory, the commission recommends including the following declaration: “Parts of this submission were drafted using an AI writing tool. I have reviewed and confirmed the accuracy of all content.”

The inclusion of this statement serves a dual purpose. Firstly, it signals to the Adjudication Officer how the document was prepared, providing important context. Secondly, and more importantly, it acts as a confirmation that the submitting party has performed their due diligence, actively verifying the AI-generated content rather than blindly trusting it. This small measure encourages a more conscious and responsible approach to integrating these powerful new tools into the legal process.

The High Cost of Inaccuracy

The ramifications of submitting flawed, AI-generated material are severe and multi-faceted. As underscored by the Adjudication Officer in the Ryanair case, it can cause significant and costly delays. Furthermore, referencing another case, Erdogan v Workplace Relations Commission, the guidance reminds parties that an Adjudication Officer has the authority to ensure a hearing “progresses expeditiously” and can refuse to admit evidence that is incorrect or irrelevant.

In serious instances, the inclusion of misleading or fabricated information can do irreparable damage to a party’s credibility. In a legal forum where trust and accuracy are paramount, being seen as careless or, worse, deceptive, can fatally undermine an entire case. The WRC’s position is clear: the potential time-saving benefits of AI are quickly negated if the output is unreliable and leads to procedural chaos and a loss of trust from the tribunal.

A Tool, Not a Barrister: A Guide for Litigants

The WRC’s message is not an outright ban on AI, but rather a robust framework for its responsible use. The guidance effectively provides a checklist of best practices. Litigants are urged not to include any material they do not fully understand and cannot defend if questioned. AI can be a helpful assistant for organising thoughts or structuring a first draft, but it is not a substitute for bespoke legal strategy, which requires an assessment of a case’s specific strengths and weaknesses.

The core takeaway is that while AI can be a powerful tool in the preparatory stages of litigation, it cannot replace the foundational elements of legal practice: deep subject-matter expertise, rigorous verification, and professional accountability. The guidance concludes by strongly recommending that individuals seek specific, expert legal advice to navigate the complexities of their cases and avoid the significant risks associated with the misuse of artificial intelligence. The commission’s intervention serves as a vital reminder that in the pursuit of justice, human oversight remains indispensable.