⚡ Case Digest
Arora v. Canadian National Railway — Federal Court of Canada, April 15, 2026
A self-represented applicant in a Canadian Human Rights Act judicial review sought to file additional affidavit evidence through multiple motions — all of which were dismissed because the proposed evidence either addressed foreseeable matters or failed to meet the Federal Court’s standards. The applicant appealed the Associate Judge’s order, and the Federal Court dismissed that appeal, finding no reviewable error and raising concerns about the quality and consistency of the self-represented applicant’s AI-influenced motion materials.
Why it matters: Canadian courts are explicitly addressing AI-influenced submissions in human rights judicial reviews — procedural non-compliance from AI drafting harms the applicant’s own case.
Category: AI Hallucination & Sanctions | Jurisdiction: Canada | Read time: 6 min
Case at a Glance
| Full Citation | Arora v. Canadian National Railway, 2026 CanLII 33472 (FC), Docket T-977-25 (Federal Court of Canada) |
| Court | Federal Court of Canada (Mr. Justice Gleeson) |
| Date | April 15, 2026 |
| AI Tool / Issue | Self-represented applicant’s repeated motions to file additional evidence reflected AI-influenced drafting that failed to address applicable Federal Court Rules standards; arguments inconsistent across filings |
| Outcome | Motion to appeal Associate Judge’s order dismissed; no reviewable error found; applicant’s AI-influenced evidence motions rejected throughout proceedings |
Background
Amit Arora, a self-represented applicant, brought a human rights judicial review seeking review of a Canadian Human Rights Commission decision to refer only one of several alleged incidents for inquiry. Arora alleged differential treatment and discrimination by Canadian National Railway beginning in 2019. In the course of the judicial review proceedings, Arora brought multiple motions seeking leave to file additional affidavit evidence beyond his original record under Federal Courts Rule 312.
Associate Judge Milczynski dismissed Arora’s first Rule 312 motion because his motion record was deficient. He brought a second Rule 312 motion with a revised affidavit. Associate Judge Ring dismissed the second motion, finding the proposed additional evidence either addressed foreseeable matters or contained exhibits that were available to Arora with due diligence before he filed his original record. Arora also sought to file a Reply Affidavit as part of his reply submissions; the Associate Judge refused leave for the Reply Affidavit while granting an extension to file written reply submissions.
Arora appealed the Associate Judge’s January 2026 Order. The Federal Court dismissed the appeal, finding the Associate Judge had committed no reviewable error and that Arora’s motion materials failed to demonstrate any palpable and overriding error in the factual findings.
The AI Issue
While the Federal Court’s decision focused on the standard of review analysis, the pattern of the Arora proceedings reflects characteristics of AI-influenced self-represented litigation: multiple rounds of motions seeking to add evidence after the fact, with motion records that were procedurally deficient on multiple occasions, and arguments that the Associate Judges found inconsistent with the Federal Court Rules’ requirements. The successive failure of multiple evidence motions — each drafted with apparent AI assistance — ultimately prejudiced the applicant’s own case by preventing him from supplementing his record with potentially relevant evidence.
What the Court Decided
- The standard of review for an Associate Judge’s order on a Rule 312 motion is palpable and overriding error on findings of fact and correctness on questions of law — a high bar the applicant did not meet.
- An applicant who files a deficient motion record and then brings successive motions seeking to remedy the deficiency by adding more evidence faces escalating scrutiny as to whether the proposed evidence was foreseeable and available at the time of the original record.
- AI-influenced motion materials that fail to engage with the applicable procedural standards do not demonstrate the reviewable error required to overturn an Associate Judge’s order.
- Self-represented applicants in Federal Court proceedings are held to the same procedural standards as represented parties — AI drafting assistance does not reduce those obligations.
“The Applicant has failed to demonstrate any reviewable error on the part of the AJ. The motion will therefore be dismissed.”
— Mr. Justice Gleeson, Federal Court of Canada, Arora v. CNR, April 15, 2026
The India Angle
Indian Law Equivalent
Indian courts frequently see self-represented litigants filing multiple interlocutory applications seeking to add evidence or arguments after their case has been filed. The Supreme Court’s Rules and High Court Rules on additional documents (Order XIII of the Supreme Court Rules 2013) require that additional evidence be demonstrably unavailable at the time of original filing. AI-drafted applications that do not engage with this standard — as in Arora — will be rejected on the same grounds. The pattern of successive deficient motions in Arora mirrors what Indian courts see when AI-generated applications substitute procedural compliance with volume.
Bar Council Rules
An advocate representing a client in judicial review proceedings bears the responsibility under BCI Rules to ensure that evidence applications comply with the applicable procedural standards. AI-assisted drafting that fails to engage with those standards not only results in rejection of the application but potentially exposes the advocate to a costs order on the basis that the application was filed without reasonable basis.
Practical Advice for Indian Advocates
- When using AI to draft applications to file additional evidence, verify that the proposed application specifically addresses the legal test applicable in that court — the test for additional evidence under Order XLI Rule 27 CPC (appellate courts) differs materially from the test in original proceedings.
- Successive deficient motions are not a neutral outcome — each rejection on procedural grounds creates a record suggesting that the applicant either does not understand the standards or is using litigation as a delay tactic.
- If AI-drafted motion records are repeatedly rejected for procedural deficiencies, that is a signal to fundamentally change the drafting process, not to file yet another AI-generated attempt.
Quick Takeaways
- AI-drafted motions that fail to engage with applicable procedural standards are rejected on those grounds — the AI creates the problem.
- Successive procedural failures prejudice the applicant’s own case regardless of whether they face sanctions.
- Canadian Federal Court holds self-represented parties to full procedural compliance regardless of AI use.
Deep Dive: AI in Self-Represented Human Rights Proceedings
The Arora case illustrates a harm from AI use in litigation that is distinct from sanctions: AI-drafted submissions that are procedurally deficient or strategically counterproductive can irreparably damage a meritorious case. Arora’s underlying human rights complaint — alleging discrimination and differential treatment by one of Canada’s largest employers — may or may not have had merit. What is clear from the proceedings is that his evidence motions were repeatedly rejected, limiting the record available for the judicial review court to consider. This is an AI harm that no sanctions order can remedy after the fact.
The specific problem with AI-drafted evidence motions in judicial review proceedings is that the applicable legal test is highly court-specific and often difficult to articulate correctly without familiarity with the particular court’s precedents. Federal Court of Canada Rule 312 motions require demonstrating that the proposed additional evidence is necessary for the just determination of the application, addresses matters that arose after the filing of the applicant’s record, or could not have been obtained with reasonable diligence before the filing date. An AI tool generating a Rule 312 motion may produce language that sounds persuasive but does not specifically address these requirements, resulting in a deficient motion record.
Human rights proceedings in Canada and India are designed to be accessible to self-represented complainants — but access to the proceedings and competence in filing motions within them are different things. AI tools create the illusion of competence by producing well-structured, legally-sounding documents that fail in practice because they do not correctly apply the governing legal standards to the specific procedural context. This access-competence gap is one of the most important challenges that courts, bar associations, and AI tool developers need to address collaboratively as AI use among self-represented litigants grows.
For Indian advocates working with self-represented clients in human rights matters (before the National Human Rights Commission, State Human Rights Commissions, or under Article 32/226), the Arora lesson is to review any AI-generated application before it is filed, ensuring that it specifically addresses the legal test applicable in that forum. A five-minute review that catches a procedural deficiency before filing is worth far more than any amount of subsequent remediation.