WARNING: This article was entirely written by ChatGPT
BENMOR’S PROMPT to ChatGPT: Please summarize this case by quoting critical parts: https://www.canlii.org/en/on/onsc/doc/2025/2025onsc6785/2025onsc6785.html
RESPONSE from ChatGPT: Below is a clear, structured summary of the case, with key quotations from the decision highlighting why it matters. I focus on the legal significance, the AI-hallucinated citations issue, and why the court treated this as an escalating contempt problem.
Overview of the Case
Ko v. Li, 2025 ONSC 6785 is an Ontario Superior Court decision addressing criminal contempt of court arising from a lawyer’s use of ChatGPT-generated “hallucinated” case law and subsequent dishonesty to the court. The decision is a continuation of earlier contempt proceedings and escalates the matter because the lawyer later admitted she deliberately misled the court during the first contempt proceeding.
The court treats this as an exceptional and serious threat to the administration of justice, ultimately referring the matter to the Attorney General, appointing amicus curiae, and signalling that a full criminal contempt hearing is required.
The Initial Contempt: Fake AI Case Law
The problem began when lawyer Jisuh Lee filed a factum containing non-existent legal authorities generated by ChatGPT.
The judge discovered the issue during oral argument:
“I could not find any reference online to the cases cited and relied upon by Ms. Lee.” (para. 5)
When directly asked whether she used ChatGPT, Ms. Lee initially denied it:
“I asked Ms. Lee directly if she had used ChatGPT to write the factum. She responded, ‘I did not.’” (para. 6)
She also claimed uncertainty about how the false cases appeared:
“She was unable to advise whether her factum had been prepared using generative artificial intelligence or whether the cases… were ‘hallucinations’ fabricated by an AI platform.” (para. 7)
First Contempt Proceeding Was Purged — Based on Falsehoods
After a show-cause order, Ms. Lee later told the court that:
- A student prepared the factum
- She learned of AI use from her assistant
- Staff had been involved
In a letter dated May 9, 2025, she wrote:
“I made necessary inquiries with my staff and confirmed that the factum in question was, in part, drafted with the assistance of an AI-based tool (ChatGPT).” (para. 9)
On that basis, the judge concluded the contempt was purged:
“The expressions of accountability and regret, coupled with the withdrawal of the offending factum, purge any contempt of court that might have been found.” (para. 62 of earlier decision, quoted at para. 12)
The Turning Point: September 30, 2025 Admission
Months later, Ms. Lee sent an unsolicited letter admitting that everything she previously told the court about staff involvement was false.
She wrote:
“In truth, I prepared the factum in question solely and entirely by myself… I personally used ChatGPT… I failed to verify the citations independently.” (para. 15)
She admitted the earlier misrepresentations were deliberate:
“The misrepresentation… was made out of fear of the potential consequences and sheer embarrassment… I recognize that this attempt to deflect accountability only compounded my initial error.” (para. 15, emphasis in original)
Why This Triggered a New Contempt Proceeding
The court treated this admission as a new and separate act of contempt, not merely a continuation of the original issue:
“Your admission of dishonesty in the contempt proceeding is a fresh act.” (para. 16)
The judge emphasized the distinction:
“The initial contempt proceeding concerned principally your failure to review the citations… The new disclosure raises issues of deception of the court in a contempt of court proceeding.” (para. 16)
Specific Allegations of Criminal Contempt
Ms. Lee was ordered to show cause for four deliberate deceptions, including:
- Lying in open court that she did not write the factum
- Misleading the court about staff inquiries
- Claiming a student prepared the factum
- Implying delegation in her written oral statement
“Your untruthful oral statement in open court… Your misleading statement… Your untruthful oral statements in the Case Conference…” (para. 17)
Legal Framework: Criminal Contempt in the Face of the Court
The court classified this as criminal contempt in the face of the court, citing appellate authority:
“Conduct which seriously interferes with or obstructs the administration of justice… with deliberate or intentional conduct, or conduct which demonstrates indifference… akin to recklessness.” (para. 19, citing R v. DaFonte)
The judge stressed the seriousness of submitting fake authorities:
“Delivering a factum containing non-existent legal submissions creates a serious risk of a miscarriage of justice.” (para. 29)
Crown’s Position
The Attorney General took carriage and argued:
1. Filing fake AI-generated cases was reckless
2. Lying to purge the first contempt compounded the offence
“Ms. Lee’s act of attempting to purge her first criminal contempt by deliberately submitting an untruthful fact narrative compounds her wrongdoing.” (para. 30)
The Court Rejects Informal Resolution
Despite Ms. Lee’s apologies and remedial steps, the court refused to resolve the matter summarily:
“A finding of contempt is made on proof beyond a reasonable doubt… any evidence led… should be subject to cross-examination.” (para. 50)
The judge highlighted the unprecedented nature of the case:
“I have not seen any case law in which a lawyer… admits to deliberately misleading a court in a criminal contempt of court proceeding about herself.” (para. 51)
Appointment of Amicus Curiae
Because Ms. Lee refused to retain counsel and continued making unsworn confessions, the court appointed amicus curiae to protect fairness and guide the process:
“This is a very unusual case and one that may resonate throughout the administration of justice.” (para. 51)
Amicus was tasked with advising on:
- Fair process
- Whether contempt has been purged
- Sentencing principles if contempt is found (paras. 56–57)
Why This Case Matters
This decision is significant because it:
- Is one of the first Canadian cases directly confronting ChatGPT “hallucinated” citations
- Establishes that lying to the court about AI use is worse than the AI error itself
- Treats candour failures by lawyers as criminal contempt
- Signals that apologies and remediation do not automatically “purge” contempt
- Demonstrates judicial concern about AI misuse undermining public trust
If you want, I can also:
- Extract key legal principles for citation
- Compare this case to other AI-citation cases internationally
- Help draft a case note, memo, or law-review style summary
BENMOR’S REACTION: So cool!
Case link: https://www.canlii.org/en/on/onsc/doc/2025/2025onsc6785/2025onsc6785.html
Steve Benmor, B.Sc., LL.B., LL.M. (Family Law), C.S., Cert.F.Med., C.Arb., FDRP PC, Acc.D.C., is a full-time Family Mediator/Arbitrator and principal lawyer of Benmor Family Law Group, a boutique matrimonial law firm in downtown Toronto. He is a Certified Specialist in Family Law, a Certified Specialist in Parenting Coordination and was admitted as a Fellow to the prestigious International Academy of Family Lawyers. Steve is regularly retained as a Divorce Mediator/Arbitrator and Parenting Coordinator. Steve uses his 30 years of in-depth knowledge of family law, court-room experience and expert problem-solving skills in Divorce Mediation/Arbitration to help spouses reach fair, fast and cooperative divorce settlements without the financial losses, emotional costs and lengthy delays from divorce court.
Share this article on: