As the use of AI across industries continues to expand, so does the controversy surrounding the technology. The legal industry is no exception, and the consequences of treating AI output as a finished product instead of a first draft have become much clearer for Canadian lawyers.
This summer, the Law Society Tribunal handed down two significant AI-related discipline decisions. They suspended a Toronto-based lawyer and another in Sault Ste. Marie.
Both situations trace back to the same tech-driven problem: AI hallucinations.
While previous misconduct involving AI had resulted in reprimands, this was the first time the Law Society Tribunal suspended licences over the matter. The disciplinary actions weren't due to the use of AI. Rather, the tribunal found they failed to fact-check filings, which would ensure they don't mislead the court, cause delays or violate professional standards.
For Canadian law firms, the suspension issued to the two Ontario lawyers isn't just a cautionary tale. It's a preview of what happens when firms trust AI-generated work to be factual and don't verify the cases they cite, which are often used in support of their arguments.
But it's a lesson that applies well beyond the courtroom, including how firms market themselves online and build their professional practice.
What the Law Society Investigated
What was found by the Law Society Tribunal came down to the same core issue: failing to fact-check AI-generated court submissions. As a result, they decided to suspend each lawyer's licence. Here's what happened.
Mary Hyun-Sook Lee (also known as Jisuh Lee), is a Toronto-based sole practitioner with roughly 30 years in the industry. She was representing a client in what began as a family law matter that turned into a complex estate dispute after the client's husband died.
When the Justice presiding over the case couldn't find several of the cases she'd cited in her factum, he asked if she'd used ChatGPT, to which she responded "No".
She then claimed her staff had used the tool without her knowledge, a response given when the judge ordered her to explain why she shouldn't be held in contempt. Lee later admitted she used ChatGPT to prepare the factum and filed it without checking the case citations.
Timothy Harmar was also suspended by the Law Society Tribunal, but his case was a little different. He failed to cooperate with three separate tribunal investigations between June 2025 and February 2026.
Two of the complaints alleged he had used AI tools to draft factums for court. In at least one case, a judge found the legal principles in the case cited weren't accurate. Harmar admitted he used AI software to prepare it and hadn't read the case himself.
The investigation also revealed that Harmar practised law while placed under administrative suspension, which added to their disciplinary measures.
Details of the Ontario Lawyers' Suspensions
The Law Society tribunal suspended Lee's licence for six months and ordered her to pay the LSO $10,000. Harmar also had to pay around $4,200 but faced a much harsher discipline: Harmar's licence was suspended immediately, continuing indefinitely unless or until he satisfies certain requirements.
Harmar must provide a "complete response to the outstanding requests" to investigators. Should he do this, an additional two-month suspension will occur. If he doesn't comply with the Law Society's Guidelines for Lawyers Who Are Suspended or Who Have Given an Undertaking Not to Practise, his licence will remain revoked.
The tribunal has emphasized that the help of AI in preparing court documents isn't a form of professional misconduct. The issue is that neither Ontario lawyer verified what ChatGPT and other AI software put in their filings. For Lee, she then repeatedly lied about it to both the court and the Law Society.
While it's the first time Ontario lawyers were suspended over misuse of AI, what happened were not isolated incidents.
One such notable case just happened in June. Ontario lawyer Shahryar Mazaheri was ordered to pay the Law Society $31,150 in costs after he relied on AI-generated materials that contained hallucinations while trying to challenge his own licence suspension.
Canadian law societies are not the only country responding to these types of issues in the legal industry. Last year, a U.S. federal judge disqualified three lawyers at a large, well-regarded firm from a case after filing AI-hallucinated materials. In that scenario, the firm had internal policies warning staff of this exact risk.
The Ethics and Dangers of Law Firms Using AI
There's no doubt that AI can be leveraged as a valuable support tool by law firms in Ontario and across Canada. Many practices are already doing it.
But the recent suspensions of two Ontario lawyers have put a spotlight on a clear gap in understanding the limitations and risks associated with the use of assistants like ChatGPT, Claude and Perplexity. While these tools were built to sound confident and fluent, AI should be an aid, not replace a lawyer's work.
For lawyers, using AI comes with several risks, and the ethics surrounding it are important to consider:
- AI hallucinations sound real. Fabricated case law these systems create reads exactly like a real case. They have the proper citation format, plausible party names and confident legal reasoning. Oftentimes, there's no obvious "tell" that it's fake, so everything should be double-checked.
- The duty to verify sits with the lawyer. Canadian courts and law societies have made it clear that a lawyer can't place their duty of competence on AI or any other type of software. Lawyers are responsible for everything filed under their name.
- Covering it up is worse than making the original mistake. In the Law Society Tribunal's reasoning for suspending two Ontario lawyers for misuse of AI, dishonesty was treated more seriously than the reliance on false cases or flawed details.
- Reputational damage can be greater than legal consequences. Lawyers named in the news and/or tribunal decisions face long-term damage to their reputations. The record of what happened never goes away, regardless of whether the reprimand is overturned.
AI can do many things, but it will never function like the human brain. It won't gain experience in the courtroom or be able to understand the nuances of how real people affect cases in real-time. It doesn't know how a specific judge may rule on certain issues or understand what clients worry about. And if it can't find the answer, there's a good chance AI will make one up.
For more guidance, check out the Canadian Bar Association's toolkit, Ethics of AI for the Legal Practitioner or access your provincial law society's resources.
How Canadian Law Firms Can Safely Use AI
Law firms knowing how to properly use AI has become an issue of professional competence.
The Law Society Tribunal's own comments when issuing their decisions to suspend Ontario lawyers Lee and Harmar point toward a practical standard of how law firms use AI. It's something most firms can adopt, and in doing so, avoid professional reprimands and reputational damage, while increasing efficiency.
The following are tips for law firms who use AI (and those who are considering it):
- Treat AI output as a first draft, never a final submission. Every citation, quote and factual claim needs to be verified against a primary source, such as CanLII or provincial court case portals, before it goes anywhere near an official filing.
- Integrate verification into your workflow. A written AI policy is good for all Canadian law firms to have, but it does nothing when someone is under pressure. In addition to a policy, build a verification step into the drafting process of all documentation to confirm citations are real and that they say what you claim they say.
- Disclose use. If a judge or regulator asks whether AI was used to create a court document, be honest about it. The recent licence suspensions by the Law Society show that the cover-up carried more weight in the reprimand than the mistake itself.
- Know what tools are appropriate. General-purpose chatbots are not the same or as helpful as legal-research platforms because they weren't built with verified case law databases and citation-checking safeguards. Understanding and recognizing these differences should help determine which tool is best based on the task.
- Commit to training. Everyone who goes near a filing should be trained on how to ethically use and leverage AI. The technology isn't going anywhere, and training those in the practice is much cheaper and safer than the repercussions of misusing it.
The reality is that no lawyer can afford to have their integrity and the quality of their services questioned.
The Connection Between AI for Case Management and Law Firm Marketing
LawOnline.ca doesn't usually cover courtroom news or Law Society investigations. It's outside our role as a legal marketing agency. But the two Ontario lawyers having their licences suspended lines up closely with something we constantly talk about with the lawyers we work with.
Our client conversations are frequently centered around how AI is a genuinely useful tool, but it's very easy to misuse and can cause serious damage. Unfortunately, that damage often isn't obvious until it's too late, and fixing it is far more complicated.
A good example is why generic, AI-generated legal content costs firms clients and search rankings. The pattern is strikingly similar to what led to Lee and Harmar's suspensions.
Content that's fully AI-written can look polished on the surface, but what it's missing is the same thing Lee's factum was missing: someone who checked it for accuracy and gave it an experience-based point of view.
That's why the content we create and publish for law firms isn't written by AI. We follow the principle that AI-assisted is far more effective than AI-created. After all, there are benefits; some of these tools are extremely helpful with outlines and research and can speed up certain parts of our work (just like they do for law firms).
But our insight, knowledge of Law Society guidelines and decades of experience providing legal marketing services to law firms can't be replicated by a machine.
Law firm AI visibility is another area we see an overlap of issues. A law firm website that's created by AI, including its content, is not going to appear in front of a lot of potential clients. It may look nice, but it won't have the technical health, structure or type of quality, people-first content that gets firms cited or recommended in Google's AI Overviews or by chatbots.
If anything, the human experience and critical thinking matter more than ever before.