High Court Practice Direction HC 142: Generative AI in Irish Court Documents
What Irish High Court Practice Direction HC 142 requires on generative AI in court documents: verification, declarations, disclosure and consequences.
The Irish courts have issued practice directions on the responsible use of generative AI in court documents: HC 142 for the High Court and CA 16 for the Court of Appeal. HC 142 is dated 29 July 2026 and came into operation on 1 September 2026. This guide summarises the High Court direction, based on its published text on courts.ie. It is general information and not legal advice. For wider good practice on AI-assisted legal work, see our guide to AI-assisted drafting and document review.
Scope
HC 142 applies to court documents used in civil proceedings, including pleadings, submissions, affidavits, witness statements and expert reports. It covers parties, legal practitioners, witnesses, experts and third parties required to produce documents, and it also applies to litigants in person. It covers proceedings begun before or after commencement, and court documents prepared after commencement. It defines generative AI as technology capable of generating new content such as text, images, audio, video, data or computer code, usually in response to user prompts. The direction refers to civil proceedings only; it does not say whether it applies to criminal proceedings.
The risks the direction names
The direction lists inaccuracy and hallucinations, including fabricated cases, legislation and references, bias and incompleteness from training data, factual errors including incorrect confirmations of accuracy, confidentiality and privacy risks because information entered may be retained, processed or disclosed, and legal privilege risks, because privileged information put into non-private tools may lose its status and may be disclosable.
Verification duties
Material generated wholly or partly by generative AI must be independently verified before use, and any limits on verification must be disclosed to the court. Using the same or another AI tool to check accuracy is not sufficient. References to case law, legislation, textbooks and articles must be confirmed to exist and to support the stated proposition, and quotations must be accurate and correctly attributed. Exhibits, images and recordings must be authentic, and fabricated or manipulated material must never be presented. Practitioners remain bound by professional duties to act competently and not to mislead the court. Litigants in person are personally responsible for the accuracy of their documents and are told to verify output using authoritative sources.
Declarations and disclosure
- Affidavits and witness statements: unless the court directs otherwise, a declaration near the start states that generative AI was not used to generate the substantive content, including by altering, embellishing or rephrasing evidence, and that the document reflects the deponent’s own knowledge and recollection. Administrative use such as formatting or spell-checking does not prevent the declaration.
- Expert reports: a declaration at the start states that generative AI was not used to generate the substantive content, which reflects the expert’s own opinions, save as otherwise expressly stated. Where AI was materially used in drafting, analysing or interpreting, the nature and extent of that use must be disclosed.
- Other court documents: use of generative AI does not of itself require disclosure. Disclosure is required where the court directs it, and everyone must be able to explain their use if asked.
The direction also says generative AI should not be used on documents subject to a court order or an express or implied undertaking where that would risk disclosure contrary to the order or undertaking.
Errors and consequences
If a party spots a problem in another party’s document, it should raise it with that party and, if there is no appropriate response within a reasonable period, with the court. If a party finds an accuracy problem in its own documents, it must notify the other parties and the court without delay and provide clean and marked-up versions, and it bears the costs unless the court orders otherwise. The direction says non-compliance can lead to adverse costs orders, rejection of documents in whole or in part, or striking out of a claim or defence where appropriate. For practitioners, it mentions possible referral to the Legal Services Regulatory Authority or their professional body, and possible personal costs orders. The court may also take misuse into account in case management and costs decisions.
Our suggestions
These are our own practical suggestions, not part of the direction. Build the affidavit and expert declarations into your templates. Keep a short internal note of how any AI tool was used on each document. Check every authority against an official source, not another AI tool. Agree with clients and witnesses, in writing, that they must not use AI to reword their own evidence.
What the sources do not say
The direction does not define "non-private" tools, set a fixed time for responding to a problem in another party’s documents, or say when the court will direct disclosure of AI use. We have not read the Courts Service guidance for non-lawyers, and we have not summarised CA 16 separately. A courts.ie news notice dated 31 August 2026 announced the directions, while the High Court direction itself is dated 29 July 2026.
Frequently asked questions
When did HC 142 take effect?
It came into operation on 1 September 2026.
Does it apply to litigants in person?
Yes, according to the direction.
Is it enough to check AI output with another AI tool?
No. The direction says that is not sufficient.
Do all documents need an AI declaration?
No. The declarations apply to affidavits, witness statements and expert reports; other documents need disclosure only where the court directs it.
Next steps
Browse our CPD resources to keep up with developments, and read the full direction on courts.ie before changing your court-document processes.
This page was last updated:
Learnsignal Healthcare Education Team
The Learnsignal Healthcare Education Team creates CPD and compliance training content for nurses, allied health professionals, and care providers, drawing on current regulatory guidance from bodies including NMBI and equivalent professional regulators.
View all posts by Learnsignal Healthcare Education Team


