Law360 Canada (June 1, 2026, 11:25 AM EDT) --

Sara Blake

The Quebec Superior Court has set aside a decision of a labour arbitrator because the arbitrator’s reasons were written with the assistance of artificial intelligence: Assn des ressources intermédiaires d’hébergement du Québec (ARIHQ) c. Santé Québec - Centre intégré universitaire de santé et de services sociaux du Centre-Sud-de-l’Île-de-Montréal, 2026 QCCS 1360.

The court noted several risks of using AI to write reasons for decision. First, there is the risk of hallucination — AI makes things up, including false references. Second, the rigidity of AI’s programmed logic is inconsistent with an exercise of discretion that must reflect human and community values and allow for changing circumstances. Third, AI may have biases due to the unknown data on which it was trained and programmer biases inherent in its undisclosed algorithms. Fourth, deliberative secrecy is lost because all uses of AI feed back into the AI program for use by others. Fifth, the use of AI by a decision-maker may lead to a loss of public trust, which is based on a belief that decisions are fair, humane and reasoned. If the public come to believe that decisions are made by a disembodied or opaque machine rather than a human being, they may lose confidence in the justice system.

The court noted that the use of AI also violates the requirement that decision-makers write their reasons for decision themselves. They may not delegate the writing of reasons to anyone. Case law has established that this task may not be delegated to a party, a lawyer or tribunal staff. The court adds AI to this list and ultimately found that the arbitrator unlawfully delegated the task of writing reasons to AI.

As with all allegations of procedural unfairness, the court requires proof. How do we know AI was used by the arbitrator to write these reasons? Material parts of the arbitrator’s reasons relied on false references to authorities. After review, the court characterized them as non-existent and “delusional.”

The court suggested that the use of AI for insignificant parts of the reasons might not violate procedural fairness but left this possibility to be decided in future cases.

Sara Blake is the author of Administrative Law in Canada, 7th edition, LexisNexis Canada. Her practice is restricted to clients who exercise statutory and regulatory powers.

The opinions expressed are those of the author(s) and do not necessarily reflect the views of the author’s firm, its clients, Law360 Canada, LexisNexis Canada or any of its or their respective affiliates. This article is for general information purposes and is not intended to be and should not be taken as legal advice.

Interested in writing for us? To learn more about how you can add your voice to Law360 Canada contact Analysis Editor Peter Carter at peter.carter@lexisnexis.ca or call 647-776-6740.