The Law Society Tribunal said the lawyer has no discipline history across her 30-year career
Explaining its decision to suspend a Toronto lawyer’s license for six months after she was dishonest with a court about her ChatGPT use, the Law Society Tribunal said its penalty is appropriate because the lawyer is genuinely remorseful, will be seriously impacted by the suspension as a sole practitioner nearing the end of her career, and is facing other consequences for her misconduct, including contempt of court proceedings later this year.
The Law Society Tribunal, which hears misconduct, licensing, and fitness to practice cases for legal professionals in Ontario, first announced that it would temporarily suspend Mary Hyun-Sook Lee’s license on July 16. The tribunal also ordered the lawyer, known as Jisuh Lee, to pay the Law Society of Ontario $10,000. The LSO had informed Lee last summer that it was investigating her conduct.
The case may represent the first time a lawyer in Canada has had their license suspended partly due to their reliance on AI-hallucinated case citations.
The tribunal has since released its formal reasons for sanctioning Lee, explaining that the lawyer failed to serve her client in a family law matter by using ChatGPT to prepare a factum containing AI-hallucinated case citations and arguments. The tribunal clarified that it is not, per se, professional misconduct to use AI tools to assist in preparing court submissions; Lee’s mistake was that she failed to verify the accuracy of her factum.
The tribunal said it took greater issue with Lee's repeated lies to the Ontario Superior Court of Justice and the LSO about her use of AI, telling both that her staff had used ChatGPT to prepare the factum without her knowledge. She later admitted that she had not employed an associate since 2020 or a law student since the previous summer.
“While we are concerned by the lawyer’s reckless reliance on the non-existent cases produced by AI, we are extremely troubled by the lawyer’s subsequent repeated lies to the court and to the law society, which are serious integrity failures,” the tribunal said.
However, the tribunal also listed a series of mitigating factors that guided their assessment of penalties. These included the fact that Lee had initially taken on the family law case “in the best traditions of professional service by a lawyer”: the client had been referred to her by the Korean Canadian Women’s Association, which serves women in need. Lee began representing the client through a Legal Aid certificate. When the client’s husband died, effectively converting the case into a complex estate matter, which prompted Legal Aid to withdraw funding, Lee continued to represent her client, who had few funds, on a pro bono basis.
The tribunal also noted that Lee had little to no prior experience with ChatGPT, had no disciplinary history throughout her 30-year career, and had repeatedly admitted to her misconduct.
“The evidence suggests that her integrity lapses in this case, while serious, are out of character,” the tribunal said.
The hallucinated material in Lee’s court submissions was first discovered last spring, when Justice Frederick Myers of the Ontario Superior Court of Justice was unable to locate several cases Lee had cited. When Myers asked Lee whether she had used ChatGPT to write her factum, Lee said she had not. The lawyer later told an LSO investigator that she had been pressed for time while preparing the factum and was unable to file it with the court until the last possible minute.
Myers ordered Lee to explain why she should not be cited for contempt of court for failing to verify the accuracy of her submissions. Lee then submitted a letter to Myers, explaining that her staff had used ChatGPT to prepare the factum without her knowledge. She apologized for the mistake and promised to complete at least six hours of continuing professional education in legal ethics and technology. Myers accepted her apology and noted that due to her mistake, Lee had already “suffered a public shaming near the end of an unblemished career,” the tribunal said.
Around the same time, the LSO began investigating the incident. When an investigator asked Lee to identify the staff member who had used ChatGPT to prepare the factum, the lawyer declined to cooperate. Later, she admitted that she had prepared the factum herself and did not know hallucinations existed until Myers confronted her about her submissions. She told the LSO that her dishonesty stemmed from her embarrassment and shock.
Lee then admitted her dishonesty to Myers, who referred the carriage of a contempt of court proceeding to Ontario’s attorney general. That case remains ongoing.
In a statement on Monday, Nadia Liva, a partner at Liva Freeman Dent LLP who represented Lee in the tribunal proceedings, noted that Lee and the LSO jointly agreed that a six-month suspension was appropriate.
“In determining whether the joint submission was appropriate, the [tribunal’s] hearing panel found Ms. Lee to be genuinely remorseful,” Liva said. “The panel recognized that Ms. Lee’s conduct ‘arose in unique circumstances, rooted in a failure to understand an emerging technology, of which she had little to no experience.’”
Liva added that the tribunal also took into account Lee’s lack of a disciplinary record across her decades-long career, “concluding that her lapses of integrity were out of character.”
Lee is the latest Ontario lawyer to face consequences from the tribunal for conduct related to the use of AI. In June, the tribunal ordered another lawyer, Shahryar Mazaheri, to pay the LSO $31,150 for relying on AI-hallucinated materials to challenge his license suspension.
In its reasons in Lee’s case, the tribunal noted that counsel representing the parties “were unable to provide us with any Ontario cases that address penalty when a lawyer or a paralegal has relied on an AI document containing hallucinations” because the issue is still relatively new.
Spokespeople for the LSO did not immediately respond to a request for comment.
