Law360 Canada (July 7, 2026, 1:26 PM EDT) --

Sergio R. Karas

Artificial intelligence is no longer an emerging issue in Canadian immigration law. It is now firmly embedded in both immigration administration and the practice of immigration litigation.

Immigration, Refugees and Citizenship Canada (IRCC) uses automated tools such as Chinook to triage and process high volumes of applications, while lawyers increasingly rely on generative AI for legal research and drafting. Recent Federal Court decisions suggest that Canadian courts are developing a coherent framework for addressing both phenomena: AI may assist decision-makers and counsel, but it cannot replace human judgment, accountability or professional responsibility.

The most recent cases reveal two parallel trends. First, courts continue to uphold IRCC’s use of technology-assisted processing where a human officer remains responsible for the final decision. Second, courts are taking an increasingly strict approach toward lawyers who use generative AI without verifying its output. Together, these developments signal the emergence of what may be described as an “algorithmic administrative law” grounded in traditional principles of procedural fairness, reasonableness and accountability.

The Chinook cases: Automation is permitted, but human decision-making remains essential

The leading authority remains Haghshenas v. Canada (Minister of Citizenship and Immigration), 2023 FC 464. In that case, the applicant challenged the use of IRCC’s Chinook system in the refusal of a work permit application. The Federal Court rejected the argument that the mere use of technology rendered the decision unfair or unreasonable. The court emphasized that judicial review focuses on the legality of the decision itself, not on whether software assisted in preparing it, and held that the critical question is whether a human officer ultimately exercised the decision-making authority.

That principle has been reaffirmed in subsequent cases. The Federal Court has consistently declined invitations to invalidate decisions solely because Chinook or similar tools were involved in the processing of applications. Applicants must demonstrate an actual impact on procedural fairness or the reasonableness of the decision rather than relying on generalized objections to automation.

However, judicial acceptance of automation is not unconditional. The court has made clear that the presence of a technological tool does not insulate immigration decisions from scrutiny.

Mehrara: The court demands meaningful engagement with the evidence

A more significant recent development emerged in Mehrara v. Canada (Minister of Citizenship and Immigration), 2024 FC 1554. There, the court examined a refusal processed with assistance from Chinook 3+, an enhanced version of IRCC’s processing platform.

Justice Michael Battista set aside the refusal because the decision-maker failed to engage with detailed evidence submitted by the applicants. The decision is important not because the Federal Court questioned the legality of Chinook itself, but because it reinforced that technological assistance cannot excuse inadequate reasoning. Even where automation streamlines file management and information review, the resulting decision must still demonstrate that the officer considered the evidence and applied independent judgment.

The decision illustrates the court’s emerging approach. AI-assisted systems may facilitate administrative efficiency, but they do not alter the substantive requirements of Canadian administrative law. Decisions must remain intelligible, justified and responsive to the evidentiary record. The use of automation, therefore, increases rather than diminishes the importance of demonstrating genuine human engagement with the application.

The transparency problem

The jurisprudence also exposes an unresolved challenge: transparency.

IRCC maintains that officers retain final responsibility for decisions and that automated systems function primarily as triage and processing tools. Yet applicants often have limited information regarding the precise role played by these systems in individual cases. As a result, courts have frequently noted the difficulty of proving that automation materially affected an outcome.

This creates what many commentators describe as a “black box” problem. Applicants bear the burden of showing how technology affected the fairness or reasonableness of a decision, but the information necessary to establish that connection is often unavailable. Consequently, broad attacks on automation have generally failed, while challenges based on specific deficiencies in the decision-maker’s reasoning have achieved greater success.

As IRCC continues expanding automated processing capabilities, this tension between administrative efficiency and transparency is likely to become the central issue in future litigation.

Hussein: AI comes before the court

The most consequential recent case involving AI in immigration law may not concern immigration decision-making at all.

In Hussein v. Canada (Minister of Immigration, Refugees and Citizenship), 2025 FC 1060, the Federal Court confronted the use of generative AI by counsel. The applicant’s lawyer relied on Visto.ai, a legal research platform designed specifically for immigration and refugee practitioners. Unfortunately, the tool generated non-existent authorities and incorrect legal propositions. Counsel submitted the material without independently verifying the citations and failed to disclose the use of AI despite repeated inquiries from the court.

Associate Judge Catharine Moore delivered what is likely to become the leading Canadian statement on AI-assisted advocacy. The court held that generative AI is a legitimate professional tool, but its use must be disclosed, and its output must be verified by a human. The court emphasized that judges cannot be expected to search for authorities that do not exist or assess legal propositions generated by hallucinated sources.

The court further concluded that counsel’s conduct warranted costs consequences because the reliance on AI remained undisclosed until after multiple court directions had been issued. The case stands as a stark warning that AI does not diminish lawyers’ professional obligations of competence, candour and diligence.

A two-pronged legal framework is emerging

Viewed together, the recent cases demonstrate that Canadian courts are developing a two-pronged approach to artificial intelligence.

When the government uses AI to assist immigration processing, courts focus on whether a human decision-maker remained responsible and whether the resulting decision satisfies ordinary standards of procedural fairness and reasonableness. Technology itself is not suspect; what matters is accountability.

When lawyers use AI in litigation, courts impose an equally rigorous standard. Generative AI may improve efficiency, but professional responsibility remains entirely human. Lawyers must verify authorities, ensure accuracy and disclose AI use where required by court practice directions. Failure to do so may result in sanctions, costs awards and potentially professional discipline.

Looking ahead

The Federal Court is not creating a separate body of “AI law” for immigration matters. Instead, it is applying established legal principles to new technologies. Reasonableness remains reasonableness. Procedural fairness remains procedural fairness. Professional competence remains professional competence.

What has changed is the environment in which those principles operate. Immigration decisions increasingly pass through automated systems before a human officer sees them, and legal submissions increasingly incorporate AI-generated content. The courts have responded with a simple but significant message: artificial intelligence may assist, accelerate and organize, but it does not decide. Nor does it relieve decision-makers or counsel of their legal obligations.

As IRCC expands its use of automated processing and lawyers continue to integrate generative AI into their practices, legal questions will become more frequent and more complex. For now, however, the emerging jurisprudence provides a consistent answer: efficiency is welcome, but accountability remains indispensable.

Sergio R. Karas, principal of Karas Immigration Law Professional Corporation, is a certified specialist in Canadian Citizenship and Immigration Law by the Law Society of Ontario, Division Chair of the ABA International Law Section, past chair of the Ontario Bar Association Citizenship and Immigration Section, past chair of the International Bar Association Immigration and Nationality Committee, and a fellow of the American Bar Foundation.

The opinions expressed are those of the author(s) and do not necessarily reflect the views of the author’s firm, its clients, LexisNexis Canada, Law360 Canada or any of its or their respective affiliates. This article is for general information purposes and is not intended to be and should not be taken as legal advice.

Interested in writing for us? To learn more about how you can add your voice to Law360 Canada, contact Analysis Editor Richard Skinulis at Richard.Skinulis@lexisnexis.ca or call 437-828-6772.