The Immigration and Refugee Board of Canada has issued a practice notice on the use of artificial intelligence in its proceedings. It takes effect September 7, 2026, and binds every party before all four divisions, including those without a lawyer.

On July 10, 2026, the Chairperson of the Immigration and Refugee Board of Canada signed a practice notice on the use of artificial intelligence in IRB proceedings. It comes into force on September 7, 2026.

The notice applies to all four divisions of the Board and to everyone who communicates with it. It reaches parties, their counsel, and self-represented parties alike. For a forum that decides refugee claims, refugee and immigration appeals, detention reviews, and admissibility hearings, that reach is broad by design.

With this notice, the IRB joins a fast-growing group of Canadian courts and tribunals that have published formal direction on AI. We track those policies across 249 courts and tribunals in our Can I Use AI in Court? tracker.

What the notice requires

The practice notice sets out a short list of obligations. The central ones:

• Declare AI use. A party who uses AI to generate or materially change content in a document must include a set declaration confirming that the content, and the authenticity of any cited case law, has been reviewed and verified by a named person.
• Verify against reliable sources. Parties must confirm legal principles and citations against reliable sources such as CanLII, official government websites, and established legal publishers. The notice is explicit that AI-generated summaries found within those sources do not count.
• Do not use AI for a person’s own evidence. Basis of claim narratives, affidavits, and witness statements must rest on the person’s own knowledge and experience. Using AI to draft, rewrite, or alter that evidence, including photographs, videos, and other corroborative material, is prohibited.
• Disclose AI in translation. Any use of AI for transcription or translation must always be disclosed, and translated documents still require a declaration from a human translator.
• Consider privacy, and do not pad. Parties are told to weigh the confidentiality risks of AI tools and not to fill submissions with AI-generated material unrelated to the facts of the case.

The required declaration reads:

Artificial intelligence (AI) was used to create or materially change text in this document. All content generated by AI, and the authenticity of any cited case law or other legal authorities, has been reviewed and verified by [name].

The consequences for non-compliance are set out as well. The Board may decline to accept or rely on a document, draw a negative inference about the credibility of the evidence, report counsel to their professional regulator, or restrict a representative from appearing before it.