Instructive examples of “things not to do” continue to germinate in Canada

This newsletter tries to report on developments related to the presence, use and abuse of AI and large-language models (LLMs) in Canadian legal affairs. The editors suspect that these developments will soon increase from a trickle to a deluge. For this period, we offer a couple of examples of odd and disturbing cases. In the British Columbia case of Zhang v. Chen the parties were involved in high-conflict family law proceedings and the father’s counsel filed an application seeking permission for the children to go to China with the father. In support of the application the father’s counsel pleaded two cases, of which citations and summaries were provided in her brief. When the mother’s counsel could not find the cases and demanded copies, the father’s counsel responded that she had realized the cases were not real. She had obtained the citations and summaries from ChatGPT, and due to her unfamiliarity with the platform she had not realized there was any chance they could be fictitious. She had contacted the bar society upon realization, and made apologies to both the court and opposing counsel. Nonetheless, the mother’s counsel sought special costs against the father’s counsel personally, based on the extra time and expense required to “expose the AI ‘hallucination’” and the father’s counsel’s alleged delay in dealing with the matter.

Read more here