On This Page You Will Find
- What Canada’s new IRB artificial intelligence rules prohibit
- Which immigration and refugee proceedings the rules cover
- When applicants can still use AI
- When AI use must be disclosed to the tribunal
- What can happen if applicants or representatives break the rules
- Why the new rules do not amount to a general IRCC ban on AI
Canada’s immigration tribunal has introduced strict new rules preventing refugee claimants, immigrants and their representatives from using artificial intelligence to create or materially rewrite personal evidence.
The Immigration and Refugee Board of Canada (IRB) says parties must not use AI to draft or substantially alter evidence that reflects a person’s own experiences or testimony.
That includes Basis of Claim narratives, affidavits, witness statements and supporting evidence such as photographs, videos and screenshots.
The rule took effect on September 7, 2026, and applies across all four divisions of the IRB. It covers applicants and appellants, their lawyers and other counsel, as well as people representing themselves.
However, the change does not amount to a general ban on the use of AI tools in Canada’s immigration system.
The notice applies specifically to proceedings before the IRB. It does not create a new AI prohibition for routine applications submitted to Immigration, Refugees and Citizenship Canada (IRCC), including Express Entry applications, study permits, work permits, visitor visas, citizenship applications or most permanent residence applications.
AI Cannot Be Used To Write A Claimant’s Story
The strongest part of the new policy deals with personal evidence.
The IRB says parties must not use artificial intelligence to generate or materially alter material that reports or reflects someone’s evidence.
Personal narratives and testimony must instead remain based on the individual’s own knowledge and experience.
The IRB specifically prohibits using AI to:
- Draft a Basis of Claim narrative
- Draft an affidavit
- Draft a witness statement
- Rewrite a person’s story
- Create or alter photographs
- Create or alter videos
- Create or alter screenshots
- Create or change other corroborating evidence
The reference to rewriting a person’s story is particularly important.
It means the restriction goes beyond fabricated evidence. A claimant cannot simply provide the facts to an AI tool and ask it to produce a polished or substantially rewritten version of their personal account for submission as evidence.
The IRB says Basis of Claim narratives, affidavits and witness statements must reflect the person’s own evidence rather than content generated by an AI system.
Spellcheck And Grammar Tools Are Still Allowed
The policy does not prohibit every use of technology when preparing personal evidence.
Applicants and representatives can continue using basic assistive functions such as spellchecking, grammar correction and formatting assistance.
However, those tools must not generate the evidence or materially change its meaning or wording.
This creates an important distinction.
Using a tool to fix spelling errors in a refugee claimant’s written account may be acceptable. Asking an AI chatbot to rewrite that account, improve the narrative or generate passages based on the claimant’s experiences may breach the rules.
Other Uses Of AI Must Be Declared
The IRB does allow AI to be used for some other material submitted during immigration and refugee proceedings.
But parties must disclose the use of AI when it generates or materially changes content in a document.
The required declaration must state that AI was used and confirm that a named person reviewed and verified the generated material, including any cited case law or other legal authorities.
Parties can also provide further information explaining which AI software they used, why they used it and whether it affected the whole document or only certain sections.
The IRB places responsibility on the party submitting the material to ensure that AI-generated information is accurate, verifiable and trustworthy.
That requirement is particularly important for legal arguments and case citations because generative AI systems can produce inaccurate or nonexistent legal authorities.
The IRB says parties should systematically verify legal principles and case citations against reliable sources such as official government websites, legal databases and trusted legal publishers.
AI Translation And Transcription Must Be Disclosed
The rules also address translation and transcription.
Anyone using AI to help transcribe or translate material must disclose that use.
IRB rules already require translated documents to include a declaration from a human translator. Under the new AI policy, that translator must also state when artificial intelligence assisted with the translation.
Applicants should also be aware that requirements can be stricter for particular types of evidence.
For example, the Refugee Protection Division’s current instructions say documents that are not in English or French must be translated by a human and that AI tools or online translation services cannot be used to produce the required translation.
Breaking The Rules Could Affect Credibility
The IRB has given the new rules significant enforcement powers.
If someone fails to comply, the tribunal can refuse to accept a document or decide not to rely on it.
More seriously for applicants, the IRB can draw negative conclusions about the credibility of evidence submitted in the case.
For lawyers and other counsel, the consequences can go further.
The IRB says it can disclose information about a representative’s conduct to their professional regulatory body. It can also restrict or prohibit counsel from appearing before the tribunal.
The Board also reserves the right to take other action it considers necessary.
Applicants Remain Responsible For AI-Generated Material
The new practice notice makes clear that using AI does not transfer responsibility for errors to the technology.
Applicants, appellants and their representatives remain responsible for everything they submit to the tribunal.
They should therefore be prepared to answer questions about how they used AI and, if requested, establish the authenticity and accuracy of their documents.
The IRB also warns against using artificial intelligence simply to make legal submissions longer.
It says parties should not fill submissions with AI-generated material that has little connection to the facts of the individual case.
The Board also advises applicants and representatives to consider privacy risks before entering sensitive information into an AI service because some tools may not provide sufficient protection for confidential information.
New AI Rules Apply To The IRB – Not All IRCC Applications
The scope of the change is important.
The IRB is an independent administrative tribunal that decides refugee and immigration matters. The new practice notice applies to proceedings before that tribunal rather than to Canada’s immigration system as a whole.
It therefore affects matters such as refugee protection claims, refugee appeals, immigration appeals, admissibility proceedings and detention-related cases before the Board.
It does not establish a general prohibition against using AI while preparing an ordinary application submitted to IRCC.
For example, the notice does not itself prohibit someone from using an AI tool when working on an Express Entry profile, study permit application, work permit application, visitor visa application, citizenship application or other routine immigration application.
Applicants still need to ensure that anything they submit to IRCC is truthful and accurate, but the September 7 practice notice does not create a blanket AI ban for those applications.
Why Canada Introduced The New AI Restrictions
The IRB says artificial intelligence could help parties present cases but also carries risks because AI systems can produce false, fabricated or inaccurate information.
The Board introduced the rules to protect the integrity of its proceedings while still allowing limited uses of AI where applicants properly disclose and verify the material.
The practice notice was signed by IRB Chairperson Manon Brassard on July 10, 2026, before taking effect on September 7. The IRB webpage was modified on September 8.
The result is not a complete prohibition on artificial intelligence in immigration and refugee proceedings.
Instead, Canada has drawn a clear line around personal evidence: AI can help with minor technical corrections, but it cannot generate, reconstruct or materially rewrite someone’s personal story.
Frequently Asked Questions
Can I use ChatGPT to write my Basis of Claim form?
No. The IRB says parties must not use AI to draft or materially change a Basis of Claim narrative or other evidence reflecting someone’s personal experiences. Minor spellchecking, grammar correction and formatting are allowed, provided the tool does not generate evidence or materially change the person’s account.
Can I use AI to improve the wording of my refugee claim?
Only for minor corrections. The IRB specifically prohibits using AI to rewrite a person’s story. A tool may help with spelling, grammar or formatting, but applicants should not ask an AI system to substantially rewrite, expand or generate their personal account for submission as evidence.
Do I have to tell the IRB if I used artificial intelligence?
In many cases, yes. If AI generates or materially changes content in a document submitted to the IRB, the document must include a declaration explaining that AI was used and naming the person who reviewed and verified the material. Minor spellchecking and grammar assistance do not require disclosure.
Does Canada ban AI in Express Entry and other IRCC applications?
No. The practice notice applies to proceedings before the Immigration and Refugee Board. It does not establish a new general AI ban for Express Entry, visitor visas, work permits, study permits, citizenship applications or most other applications submitted directly to IRCC.
What happens if someone submits prohibited AI-generated evidence?
The IRB can refuse the document, decline to rely on it or draw negative conclusions about the credibility of the evidence. Representatives can also face referral to their professional regulator or restrictions on appearing before the Board. The IRB can take other action it considers necessary.