Since 2 August 2026, Article 50 of the EU AI Act has imposed targeted transparency obligations concerning certain uses of AI.
Now, the High Court’s new Practice Direction HC 142 adds specific requirements for the use of generative AI in civil proceedings from 1 September 2026.
The two regimes address different questions, and neither requires lawyers to disclose every use of AI.
An Irish solicitor or law firm using an AI system professionally will ordinarily be a “deployer” under the AI Act. Article 50 does not, however, require every AI-assisted letter, opinion or submission to be labelled.
The obligation depends on what the system is doing. Where an AI system interacts directly with a client through a legal chatbot, for example, Article 50 requires the provider to ensure that the client is told that they are interacting with AI, unless this is obvious. Although that duty rests on the provider, the deploying firm should still ensure that clients are not misled into believing they are communicating with a solicitor.
A lawyer acting as deployer has more direct duties when using emotion-recognition or biometric-categorisation systems, or AI-generated or manipulated material constituting a deepfake. Article 50 may also require disclosure where AI-generated or manipulated text is published to inform the public about a matter of public interest. That obligation does not apply where the text has undergone human review or editorial control, and a person or organisation accepts editorial responsibility for it.
What does HC 142 add?
HC 142 applies to documents prepared or generated on or after 1 September 2026 for use in High Court civil proceedings, including proceedings already in existence.
It does not generally require disclosure simply because generative AI has been used to assist in preparing a court document. However, anyone responsible for the document must be able to explain the nature and extent of any such use if directed to do so by the Court.
There are additional requirements where AI is used in relation to evidence:
- Affidavits and witness statements must ordinarily include a declaration that generative AI was not used to generate or alter their substantive content.
- Expert reports must contain a similar declaration, with any material use of generative AI expressly disclosed.
Any AI-generated content must be independently verified under appropriate human control and supervision. Asking the same, or another, AI system to confirm the accuracy of the content is expressly insufficient.
Authorities, quotations, facts and exhibits must therefore be checked against the original and authoritative sources.
Failure to comply with HC 142 may have serious consequences, including adverse costs orders, rejection of court documents, striking out, personal costs orders, or referral to the Legal Services Regulatory Authority or the practitioner’s professional body.
The Practical Point
For ordinary internal drafting, research or document review, Article 50 creates no general disclosure duty. HC 142 similarly avoids a blanket disclosure rule for court documents, but it introduces specific declarations, potential court-ordered disclosure and rigorous verification requirements.
The use of AI is therefore not automatically something to announce. It is always something to supervise.
PS: This article has been humanly reviewed and its sources independently checked.
