You trust your lawyer to know the law, or at least know how to look it up. But when a Toronto practitioner decided to copy and paste artificial intelligence hallucinations directly into her court filings, the judicial system fired back with a devastating reality check.
A Canadian lawyer faces a 6-month suspension for citing ChatGPT cases in a court hearing, making this a historic penalty that completely upends how the legal industry views automated text tools. The Law Society Tribunal handed down the suspension to Mary Hyun-Sook Lee, who also goes by Jisuh Lee. Her six-month suspension took effect in July 2026, forcing a major reckoning across the legal profession.
The core issue isn't just that she used a chatbot to do her homework. It's that she tried to pass off fake information as legitimate law, panicked when caught, and actively lied to a judge. If you think automation can completely replace professional oversight, this case proves exactly why that line of thinking will destroy your career.
How a Simple Copypaste Became a Career Disaster
The trouble started during an estate and family law matter in front of Ontario Superior Court Justice Fred Myers. Lee submitted written arguments, known as a factum, filled with citations meant to support her client’s position.
When Justice Myers sat down to review the paperwork, things felt off. He ran the case names through standard legal databases like CanLII and Westlaw. Nothing came up. Some of the cases literally did not exist. Others were real cases, but they meant the exact opposite of what Lee claimed they did. In one bizarre instance, a case she cited to support removing an estate trustee actually concluded that the trustee should stay.
Justice Myers didn't let it slide. He asked Lee to provide physical or digital copies of the full decisions she relied on. She couldn't do it. The judge openly questioned whether she used artificial intelligence to build her brief. Instead of admitting the mistake immediately, Lee claimed her office didn't usually use those tools and said she needed to check with her clerk.
That excuse blew up in her face.
The Cover Up is Always Worse Than the Crime
Using an AI chatbot to generate research is incompetent, but lying about it to a judge is what turns a mistake into professional suicide. Lee later admitted to the court that she lied about her use of ChatGPT because she was embarrassed and terrified of the professional consequences.
To make matters worse, she initially tried to shift the blame to a non-existent law student. When you are an officer of the court, your word is your entire currency. Once you lie to a judge, the system stops looking at your technological ignorance and starts looking at your ethical fitness to practice law.
The Law Society Tribunal found that Lee violated several fundamental professional conduct rules. She failed to serve her client competently, deliberately misled the court, and failed to act with basic honor and integrity. On top of the six-month license suspension, the tribunal hit her with a $10,000 order to pay the Law Society of Ontario's costs.
Her nightmare isn't over either. The Ontario Superior Court referred her to the provincial attorney general for criminal contempt proceedings, which are still ongoing.
This is Not an Isolated Incident
If you follow legal tech, you know this keeps happening. Lawyers keep treating large language models like search engines when they are actually text predictors. They guess the most probable next word based on vast datasets, meaning they will gladly manufacture a highly believable legal citation from thin air if it fits the linguistic pattern of your prompt.
We saw this with Mata v. Avianca in New York, where attorney Steven Schwartz submitted half a dozen fake citations to a federal judge and admitted he didn't realize a chatbot could just make things up. We saw it in British Columbia with lawyer Chong Ke, who filed fabricated precedents in a child custody dispute.
Even elite corporate operations aren't safe. In early 2026, Wall Street powerhouse Sullivan & Cromwell had to issue an embarrassing public apology to a bankruptcy judge after an emergency motion they filed turned out to be riddled with fake, AI-generated legal authorities. If a firm with massive resources and strict internal review processes can let a hallucination slip through the cracks, an independent lawyer working solo has zero margin for error.
Legal tech tracking specialist Tom Macintosh Zheng noted that the Lee case represents a massive shift in how regulators handle these screw-ups. Previously, regulators handed out warnings or financial reprimands. A six-month suspension means the hammer is officially coming down.
Why Modern AI Tools Hallucinate and Fool Smart Professionals
People wonder how smart professionals fall for this over and over. The answer lies in how these models operate. They don't have a database of real world facts or court registries that they check for accuracy. They are trained to write sentences that sound exactly like what a human lawyer would write.
When you ask a generic chatbot for a case that supports a specific legal argument, it won't tell you that no such case exists. Instead, it creates a perfectly formatted citation complete with a plausible style of cause, volume number, series indicator, and page number. It looks identical to a real precedent. Unless you log into a verified legal repository and look up that exact volume, you can't tell it's completely fabricated.
What You Must Do to Protect Your Business and Practice
Technology can make you faster, but it won't cover your back when things go wrong. If you use automated platforms for research, drafting, or content generation, you need a hard protocol to avoid career-ending errors.
Never Trust a Direct Citation Without a Secondary Source
If any automated system gives you a case name, a statute number, a financial figure, or a specific quote, consider it fake until you find it somewhere else. Open an official registry or database to read the source yourself. If it doesn't show up in an official repository, delete it instantly.
Maintain Total Transparency with Your Clients and the Court
If you use automated assistance to build documents, understand your local rules. Many jurisdictions now require specific disclosures if AI helped draft your submissions. More importantly, if a judge ever questions your work, do not defensive-wall your way out of it. Own the mistake immediately, apologize, ask for a brief adjournment to fix it, and tell the absolute truth. The court can forgive an operational error; it won't forgive a lie.
Audit Every Output Line by Line
Do not skim for style and grammar. You must check the underlying logic and factual basis of every single sentence you publish under your name. If you hand the drafting off to an assistant, an associate, or an automated program, you retain 100% of the professional liability if that document misleads someone.
The legal world is drawing a clear line in the sand. Innovation is fine, but leaving your professional brain at the door will get you cleared out of the business faster than you can hit undo.
Verify everything. Trust nothing.
