AI can write a polished paragraph in seconds.
But if that paragraph is supposed to tell your personal immigration or refugee story, Canada’s Immigration and Refugee Board (IRB) is drawing a very clear line: AI cannot create or materially change your evidence.
The IRB’s new Practice Notice on the Use of Artificial Intelligence in IRB Proceedings came into effect on September 7, 2026 and applies across all four IRB divisions. The notice is designed to address the risks of AI-generated or AI-altered material, particularly where inaccurate, fabricated, or misleading information could enter an immigration or refugee proceeding.
That means applicants, lawyers, consultants, and self-represented individuals need to be much more careful about where AI fits into an immigration application, refugee claim, affidavit, witness statement, or supporting evidence.
And no, this does not mean the IRB has banned every use of AI.
The important question is:
What exactly is AI being used to do?
What the New IRB AI Rules Actually Say
The biggest change is straightforward.
The IRB says parties must not use AI to generate or materially change content that reports or reflects a person’s evidence.
That includes documents such as:
- Basis of Claim narratives
- Affidavits
- Witness statements
- Personal stories or accounts
- Photographs
- Videos
- Screenshots
- Other corroborative evidence
The reason matters.
A refugee or immigration proceeding can depend heavily on whether a person’s evidence is authentic, consistent, and based on their own knowledge and experience. If an AI tool rewrites that story, adds information, changes details, or creates supporting evidence, the record may no longer accurately reflect what the person actually experienced.
So using AI to make your story “sound better” can create a much bigger problem than a few grammar mistakes ever would.
Can You Use AI to Write a Refugee Claim?
This is where applicants need to be especially careful.
According to the IRB’s practice notice, AI must not be used to draft a Basis of Claim narrative, affidavit, or witness statement.
It also specifically prohibits using AI to rewrite a person’s story.
For example, imagine someone writes:
“I left my country because I was threatened by people connected to my political activities.”
Using AI to completely rewrite that personal account into a detailed, emotionally polished narrative could potentially introduce wording, facts, timing, motivations, or details that did not actually come from the claimant.
That is precisely the type of risk the new rule is addressing.
Your refugee claim narrative should reflect your own knowledge and experience.
AI should not become the author of your personal evidence.
What About AI-Generated Photos, Videos or Screenshots?
This is another major red flag.
The IRB expressly says parties cannot use AI to create or alter photographs, videos, screenshots, or other corroborative evidence.
That matters because supporting evidence can sometimes be used to establish important facts in an immigration or refugee proceeding.
Changing an image with AI might seem harmless if the goal is simply to make it clearer. But if the alteration materially changes what the evidence shows, it can raise serious authenticity concerns.
The safer approach is simple:
Do not manufacture or materially alter evidence to make your case appear stronger.
Authenticity matters more than presentation.
So, Is All AI Use Banned by the IRB?
No.
This is one of the most important distinctions in the new IRB AI policy.
The IRB allows certain minor assistive functions, including:
- Spellcheck
- Grammar correction
- Formatting assistance
But there is an important condition: these tools cannot generate the evidence or materially change its wording.
Think of it this way:
Grammar correction: potentially permitted.
AI rewriting your personal story: prohibited.
That difference is critical.
If you wrote your own statement and a basic tool catches a spelling mistake, that is very different from asking an AI system to rewrite the entire statement more persuasively or emotionally.
AI Translation and Transcription Have Their Own Rules
There is another area applicants should not overlook: AI translation and transcription.
The IRB says the use of AI for transcription or translation must always be disclosed.
The notice also reminds parties that translated documents must include a declaration from a human translator under the applicable IRB rules. If AI helped with the translation, the human translator must declare that assistance.
So if you’re preparing documents for an immigration hearing, refugee claim, or IRB proceeding, don’t assume that AI translation is invisible or doesn’t need to be mentioned.
Disclosure matters.
When Do You Have to Declare AI Use?
The rules go beyond personal evidence.
If AI is used to generate or materially change content in a document submitted to the IRB, the document must include an AI-use declaration.
The IRB provides specific wording stating that AI was used and that the AI-generated content and the authenticity of cited case law or other legal authorities have been reviewed and verified by a named person. Parties can also explain which software they used, why they used it, and whether AI was used throughout the document or only in certain sections.
This creates an important practical rule:
Using AI doesn’t transfer responsibility to AI.
The person submitting the document remains responsible for its accuracy.
AI Doesn’t Get to Be Your Legal Researcher
Here’s another trap.
AI can produce an impressive-looking legal answer that sounds completely convincing—and still contain a nonexistent case, incorrect legal principle, outdated information, or a citation that doesn’t support the statement.
The IRB specifically requires parties using AI to verify legal principles and case citations against reliable sources. It identifies sources such as legal databases, official government websites, trusted public services, and CanLII as examples of reliable sources. AI-generated summaries are not considered reliable sources.
So if an AI tool gives you a case citation, don’t simply copy and paste it into an immigration submission.
Check it.
Find the actual decision.
Read the relevant section.
Confirm that the case actually says what your submission claims it says.
Don’t Use AI to Make Your Submission Longer
More pages do not automatically mean a stronger immigration case.
The IRB’s notice specifically warns parties against filling submissions with AI-generated information that isn’t relevant to the facts of their particular case.
The notice states that longer submissions are not better submissions and cites Messa c. Canada (Citoyenneté et Immigration), 2025 CF 1557, at paragraph 11.
This is particularly relevant now that generative AI makes it incredibly easy to produce 20 pages when five focused pages might communicate the issue more clearly.
For an IRB hearing or immigration proceeding, relevance matters.
Don’t bury the important facts under AI-generated filler.
What Happens If You Ignore the AI Rules?
This is where the new practice notice becomes more than just guidance.
If a party doesn’t follow the notice, the IRB says it may:
- Refuse to accept a document for filing or decline to rely on it
- Draw negative inferences about the credibility of evidence
- Disclose information to a counsel’s professional regulatory authority
- Restrict or prohibit counsel from appearing before the IRB
- Take other necessary action
That means an AI mistake isn’t necessarily just a technical formatting issue.
Depending on the circumstances, it can affect how evidence is treated and how credibility is viewed.
What Should Immigration Applicants Do Now?
The new rules make one thing clear: don’t let AI cross the line from assistance into authorship of your evidence.
If you’re preparing an immigration application, refugee claim, IRB submission, affidavit, or witness statement, use a careful process.
1. Tell your own story
Your personal evidence should come from your own knowledge and experience.
Don’t ask AI to invent details, expand events, create emotional language, or fill gaps in your memory.
2. Check every legal citation
If AI suggests a case or legal principle, verify it through an authoritative source before using it.
3. Keep supporting evidence authentic
Don’t use generative AI to create or materially alter photographs, videos, screenshots, or other corroborative evidence.
4. Disclose AI when required
If AI materially changes content in a document submitted to the IRB, follow the declaration requirements.
5. Be careful with translation
If AI assists with translation or transcription, make sure the required disclosure and human translator declaration are properly handled.
6. Don’t upload sensitive information carelessly
The IRB also warns parties to consider privacy and confidentiality risks when using AI tools because some systems may not have adequate security protections for sensitive information.
The Biggest Takeaway for Immigration and Refugee Applicants
The new IRB rule isn’t really about whether AI is “good” or “bad.”
It’s about accountability and authenticity.
AI can help with certain administrative or minor language tasks. But your personal evidence is different.
Your refugee claim, immigration evidence, affidavit, or witness statement needs to accurately represent the person’s own experience.
That’s why the line is so important:
AI can assist with some tasks—but it cannot become the source of your evidence.
And if you’re unsure whether a particular AI-assisted document crosses that line, don’t guess. Immigration and refugee proceedings can involve significant legal consequences, so the specific facts of the case and the applicable IRB rules matter.
