The court's analysis rested on Quebec civil procedure law and the principle that decision-makers cannot delegate core authority to computer programs. The specific mechanism by which the hallucinated citations undermined the award—and whether the court would have reached the same conclusion absent such obvious defects—remains unclear from available reporting.
The ruling arrives as arbitrators and counsel across Canada are actively experimenting with generative AI for drafting, legal research, and administrative tasks. Practitioners should expect heightened scrutiny of AI use in awards going forward. The immediate practical question is whether courts can reliably detect improper AI reliance when an award contains no obvious hallucinations or other glaring errors. Arbitrators should document the scope and limits of any AI assistance and ensure that reasoning and conclusions remain demonstrably their own.