News
New IRB Rules on the Use of Artificial Intelligence: What Refugee Claimants and Other Parties Need to Know
August 18, 2026
Artificial intelligence tools such as ChatGPT are increasingly being used to help people research, write, translate, and organize information. Beginning 7 September 2026, the Immigration and Refugee Board of Canada (IRB) will introduce new rules governing how AI may be used in proceedings before the Board.
The new Practice Notice applies to all four divisions of the IRB, including the Refugee Protection Division (RPD) and Refugee Appeal Division (RAD). It applies both to people represented by a lawyer or consultant and to people representing themselves.
AI Cannot Be Used to Write a Claimant’s Personal Story or Evidence
Perhaps the most important rule for refugee claimants is that AI must not be used to generate or materially change personal evidence.
This means that AI should not be used to:
The reason is important. A refugee claim is based on the claimant’s own experiences and knowledge. The evidence presented to the IRB must remain the claimant’s evidence, rather than a version of events generated or materially rewritten by an AI system.
This does not mean that every use of technology is prohibited. Ordinary tools such as spellcheck, grammar correction, and formatting assistance may still be used, as long as they do not generate the evidence or materially change what the person is saying.
AI May Be Used for Other Documents, but It Must Be Disclosed
The Practice Notice does not prohibit AI altogether.
AI may still be used to assist with other documents submitted to the IRB, including legal submissions. However, where AI has been used to create or materially change the content of a document, the document must contain a declaration explaining that AI was used and confirming that the AI-generated content and any legal authorities have been reviewed and verified.
This is particularly important because AI systems can sometimes produce information that sounds convincing but is inaccurate or even invent cases, quotations, facts, or legal principles.
Lawyers, representatives, and self-represented parties therefore remain fully responsible for everything they submit to the IRB. Legal cases and authorities generated through AI must be independently checked against reliable sources such as CanLII, official government websites, and recognized legal databases.
Special Rules Apply to Translation and Transcription
The IRB also requires disclosure whenever AI is used for translation or transcription.
Importantly, documents requiring translation under the IRB’s Rules must still be accompanied by a declaration from a human translator. If the translator used AI to assist with the translation, that use must also be disclosed.
Refugee claimants should therefore be cautious about independently translating documents using ChatGPT, Google Translate, or similar services and then submitting those translations as evidence.
Claimants Should Also Think About Privacy
The IRB specifically warns parties about the privacy risks associated with AI.
Refugee proceedings frequently involve highly sensitive information, including details about persecution, political activities, family members, sexual or gender-based violence, medical information, identity documents, and the location of people who may still be at risk.
Before entering this type of information into an AI service, claimants and their representatives should understand how that service handles, stores, and protects the information.
What Happens If the Rules Are Not Followed?
Failure to comply with the Practice Notice can have serious consequences.
Depending on the circumstances, the IRB may refuse to accept a document, decide not to rely on it, or draw a negative inference concerning the credibility of evidence. In cases involving counsel, the IRB may also report concerns to the lawyer’s or representative’s professional regulator or restrict counsel from appearing before the Board.
What Does This Mean for Refugee Claimants?
AI can be a useful tool, but it should not become the author of a refugee claimant’s story.
A claimant’s narrative, affidavit, and testimony must continue to reflect their own memories, experiences, and words. AI should never be used to invent details, fill gaps in memory, make a story sound more persuasive, or change evidence to address perceived weaknesses in a case.
Claimants who are represented should also tell their lawyer if they have used AI to prepare, rewrite, or translate any document connected to their case. This allows counsel to determine whether the document can properly be submitted and whether disclosure of the AI use is required.
The central message of the new Practice Notice is straightforward: AI may assist with certain aspects of an IRB case, but the people presenting the case remain responsible for its accuracy, authenticity, and reliability.
The IRB’s Practice Notice comes into force on 7 September 2026.
Read the IRB Practice Notice: Use of artificial intelligence in IRB proceedings
