Sign in. It’s quick, free and it’s up to you.

An account is an optional way to support the work we do. Find out more.

THE HIGH COURT has introduced new rules governing the use of artificial intelligence in legal proceedings, warning that lawyers who submit inaccurate AI-generated material could face sanctions, personal costs orders or even referral to the national legal regulator.

The new practice direction, signed by High Court President David Barniville, comes into force on 1 September and applies to all civil proceedings before the High Court.

While acknowledging that generative AI can reduce costs and improve access to justice, the court warns that the technology carries significant risks, including the creation of fabricated legal authorities, inaccurate summaries of genuine cases and so-called “hallucinations”.

“GenAI Tools may also generate incorrect legal propositions or inaccurately summarise genuine sources,” a document on the High Court directions said.

Under the new rules, anyone responsible for court documents must independently verify any material generated with AI before it is used in proceedings.

The practice direction makes clear that checking information with another AI tool is not sufficient, and that lawyers remain personally responsible for the accuracy of everything filed with the court.

It also introduces new requirements for affidavits, witness statements and expert reports.

Unless otherwise directed by the court, anyone swearing an affidavit or making a witness statement must declare that AI has not been used to generate or rewrite the substance of their evidence.

Expert witnesses must similarly disclose any material use of AI in preparing their reports.

The court warns that parties who fail to comply with the practice direction could face adverse cost orders, have court documents rejected, or in some cases have their claim or defence struck out.

Lawyers who submit non-existent legal authorities or inaccurately cite genuine ones may also be referred to the Legal Services Regulatory Authority or their professional body.

The move follows several recent cases in which Irish judges expressed concern about the use of artificial intelligence in court proceedings.

In March, the Court of Appeal criticised former journalist Gemma O’Doherty’s use of AI after written submissions filed in a defamation appeal cited legal authorities that “simply did not exist”.

Judge Caroline Costello described the false citations as “hallucinations”, noting they had forced the opposing side to spend time and money attempting to locate cases that had never existed.

The judge also issued guidance that AI could be used to assist with legal research, but only where users independently verified the material before relying on it in court.

Separately, in October last year, High Court judge Mary Rose Gearty raised concerns after a litigant admitted using an AI tool to prepare court documents in proceedings involving an anonymity order.

The judge questioned whether feeding details of the case into an online AI platform risked identifying the alleged victim or breaching the court’s anonymity order.

The new practice direction states that information subject to court orders restricting its disclosure should not be entered into AI systems where doing so could risk confidential information being shared or disseminated.

To embed this post, copy the code below on your site