Asey v. Association of Justice Counsel: AI Hallucinations Before Ontario Human Rights Tribunal — Access to Justice Impact Flagged

Background: A Human Rights Complaint, a Deferral, and a Reconsideration Request

Farid Asey, a self-represented complainant, filed a human rights application against the Association of Justice Counsel — the union representing federal government lawyers. His application had been deferred by the Tribunal pending the outcome of union grievance proceedings (2026 HRTO 94). He filed a Request for Reconsideration under Rule 26 of the Tribunal’s Rules of Procedure, seeking to have the deferral decision revisited.

The reconsideration request was lengthy, repetitive, and, in Adjudicator Ashton’s characterization, “difficult to penetrate.” Both the respondent and the Tribunal itself flagged concerns about the accuracy of the legal submissions and the quality of the cases cited.

The AI Issue: A Troubling Trend Before This Tribunal

Adjudicator Ashton found that the applicant’s submissions “appear to be the product of artificial intelligence.” The problems were multiple: dramatically incorrect articulations of the applicable Rules and the Tribunal’s Practice Direction on Reconsiderations; arguments beyond the appropriate scope of reconsideration; and cases that were “either non-existent or, do not stand for the principles proposed.”

The Adjudicator noted this was “a troubling trend seen more and more frequently before this Tribunal” — confirming that AI hallucinations had appeared in multiple prior HRTO proceedings. The decision invoked Tribunals Ontario’s formal Practice Direction on the Use of Artificial Intelligence (AI) in Tribunal Proceedings — a province-wide directive that applies to all Ontario administrative tribunals and requires parties to be mindful of AI risks, responsibilities, and consequences.

The Adjudicator also made an important systemic point: “While Tribunal resources are unnecessarily expended to confirm the veracity of a party’s submissions, others are deprived of those resources and their own timely access to justice. This is unacceptable.” This access-to-justice framing is distinctive — AI hallucinations impose resource costs on tribunals that delay justice for other litigants.

Holdings

1. Reconsideration denied. No basis existed under Rule 26 to reconsider the deferral decision.
2. AI submissions identified and condemned. The submissions were found to be the product of AI, containing non-existent cases and incorrect statements of Tribunal Rules.
3. Tribunals Ontario AI Practice Direction invoked. The Adjudicator specifically cited this Practice Direction, reminding the applicant of obligations regarding AI use in tribunal proceedings.
4. Access to justice impact flagged as “unacceptable.” The systemic impact of AI hallucinations — consuming tribunal resources and delaying other applicants — was expressly noted as an independent reason for concern beyond the individual party’s conduct.

“While Tribunal resources are unnecessarily expended to confirm the veracity of a party’s submissions, others are deprived of those resources and their own timely access to justice. This is unacceptable. I refer to Tribunals Ontario’s Practice Direction on the Use of Artificial Intelligence (AI) in Tribunal Proceedings.”

— Adjudicator Sally Ashton, HRTO, 2026 HRTO 367

India Angle: AI Hallucinations in Indian Administrative Tribunals

India’s administrative tribunal ecosystem — National Green Tribunal (NGT), Central Administrative Tribunal (CAT), Armed Forces Tribunal (AFT), Debt Recovery Tribunals (DRT), National Company Law Tribunal (NCLT), Consumer Forums, and Labour Courts — handles millions of proceedings annually. Many are initiated by self-represented complainants who now have access to AI tools. The HRTO pattern will recur in Indian tribunals.

Relevant Indian Law

• Tribunals Reforms Act 2021 / Administrative Tribunals Act 1985: Administrative tribunals in India are bound by principles of natural justice but are not strictly bound by CPC or Evidence Act provisions. This flexibility can make AI hallucination errors harder to identify — tribunals may not scrutinise every citation as rigorously as civil courts.
• Consumer Protection Act 2019 / National Consumer Disputes Redressal Commission: Consumer forums are accessible without lawyers and handle large volumes of self-represented complaints. AI tools for consumer complaints are already available; hallucinated citations in consumer forum submissions will become increasingly common.
• Right to Information Act 2005 / Central Information Commission: RTI appellate proceedings before the CIC also involve self-represented appellants who may use AI to draft second appeals. AI hallucinations about disclosure exemptions under Section 8 of the RTI Act would be particularly problematic.

Three Practical Tips

1. Read the tribunal’s own Rules before citing them. The HRTO found that the applicant’s submissions “dramatically incorrectly” stated the Tribunal’s own Rules. This is an easily avoided error: download the applicable Rules or Practice Directions from the tribunal’s website and read them directly before filing. AI tools frequently misstate procedural rules.
2. Tribunal proceedings are on the record and affect other litigants’ access to justice. The HRTO’s access-to-justice framing is a reminder that AI-impacted tribunal filings have systemic consequences — not just personal ones. This applies equally to Indian tribunals with heavy dockets (NGT, NCLT, consumer forums).
3. Self-represented complainants in Indian tribunals should seek at least a one-time legal aid consultation before filing. NALSA-funded legal aid clinics and bar association legal aid committees can provide a one-time review of AI-drafted submissions to catch hallucinations before they reach the tribunal record.

Quick Takeaways

• AI hallucinations have now been documented in the Human Rights Tribunal of Ontario — an administrative forum, not just a civil court — confirming the tribunal ecosystem is equally affected.
• Ontario’s formal Tribunals Ontario AI Practice Direction provides a province-wide framework applicable to all Ontario administrative proceedings.
• The access-to-justice framing — AI hallucinations waste tribunal resources and delay other litigants — is a compelling systemic argument for stricter AI use policies.
• Incorrectly stating a tribunal’s own procedural Rules is a distinctive AI hallucination failure mode; always read the current Rules directly from the source.
• In India: no equivalent of Ontario’s Tribunals AI Practice Direction exists yet; individual tribunal presidents have inherent authority to impose AI disclosure and verification requirements.