Ontario Lawyer Suspension Moves AI-Citation Failures Into Professional Discipline

A six-month licence suspension in Ontario shows how quickly an AI-assisted filing problem can become a professional-discipline case when false authorities are followed by misleading statements and dishonesty toward the regulator.

On July 16, 2026, the Law Society Tribunal found that Toronto lawyer Mary Hyun-Sook Lee committed professional misconduct arising from a factum prepared with an artificial intelligence tool. The Tribunal ordered a six-month suspension beginning July 17 and required Lee to pay C$10,000 in costs.

The published order is brief, but its structure matters. The panel did not treat the matter as a single failure to verify citations. It found three separate forms of misconduct tied to competence, candour to the court, and honesty with the Law Society of Ontario.

Three failures found, not just a citation mistake

First, the panel found that Lee failed to serve her client competently by relying on a factum that contained only nonexistent or irrelevant case law generated by an AI tool. The Tribunal identified this as a violation of Rule 3.1-2 of Ontario's Rules of Professional Conduct.

Second, the panel found that Lee deliberately misled a court while relying on the AI-assisted factum. That finding fell under Rule 5.1-2, which governs advocacy.

Third, the panel found that Lee was dishonest with the Law Society about her use of an AI tool while drafting the filed factum. The Tribunal treated that conduct as a failure to act with honour and integrity under Rule 2.1-1.

Those findings explain why the decision belongs in the professional-responsibility category rather than the growing list of case-specific AI citation sanctions. The false authorities started the problem. The panel's findings about deliberate conduct and dishonesty expanded it.

The order reaches beyond hallucinated cases

Many court orders involving AI-generated authorities focus on verification: whether a lawyer checked that a cited case existed, confirmed that a quotation was accurate, or reviewed the final filing before signing it.

The Ontario order adds another layer. A lawyer's response after an AI-related error can create separate professional exposure. Statements to the court and to the regulator are not secondary details. They can become independent grounds for discipline.

That distinction is important for firms designing AI-use policies. A policy that says lawyers must verify citations addresses only the first point of failure. A workable response plan also needs to address escalation, correction, preservation of the drafting record, and candour when a court or regulator asks how a document was prepared.

What the published order establishes

The Tribunal's July 16 publication establishes the misconduct findings, the six-month suspension, and the costs award. It states that the suspension began July 17, 2026. The costs are payable in two C$5,000 instalments, due by December 31, 2026, and December 31, 2027, with interest on overdue amounts.

The public document does not provide the panel's full reasoning. The Tribunal's case-document guidance explains that written or oral reasons may be released after a hearing, sometimes months later. Until fuller reasons appear, the safest reading is the narrow one: the panel found the three listed violations and imposed the stated sanction. The short order does not support broader conclusions about how every use of generative AI will be treated in Ontario discipline cases.

Practical implications for legal teams

The decision supports several immediate controls for lawyers and law firms.

Verify the filed version

Citation checking should cover the exact document filed with the court. Every authority should exist, support the stated proposition, and match any quotation or pinpoint citation. Reviewing an earlier draft does not establish that the final filing is accurate.

Preserve enough information to investigate

When AI assists with research or drafting, the firm should be able to determine which tool was used, what material it produced, who reviewed the output, and what was changed before filing. That record can be essential if the court identifies a problem.

Correct quickly and accurately

An AI-related error should trigger a defined escalation path. Counsel should verify the scope of the problem, notify responsible lawyers, preserve relevant records, and determine what correction or disclosure the forum requires. Guessing about the drafting history can make the situation worse.

Treat regulatory responses as a separate risk point

Answers to a law society or disciplinary authority require the same care as representations to a court. The Ontario panel's dishonesty finding shows that the investigation response can become part of the misconduct case.

The broader professional-responsibility shift

AI-citation cases are no longer confined to warnings that lawyers must check machine-generated work. Courts have imposed monetary sanctions, filing restrictions, referrals, and other remedies in several jurisdictions. The Ontario matter shows the issue moving squarely into professional discipline, with a licence suspension based on a combination of AI-generated authorities, misleading advocacy, and dishonesty during the regulatory process.

The practical lesson is direct. Verification is the first control, not the entire control system. Legal teams also need supervision, reliable records, a correction process, and a plan for accurate responses when a court or regulator starts asking questions.

Sources

This article summarizes a published professional-discipline order. It does not provide legal advice. The Tribunal publication reviewed for this article states the findings and sanction but does not contain full reasons.