Quebec Courts Issue Binding AI Rules, Barring Generative AI from Judicial Reasoning

Quebec's courts jointly issued governance guidelines on September 6, 2026, restricting judges' use of generative AI and placing judicial reasoning firmly off-limits for AI systems.

The guidelines were adopted by the Quebec Court of Appeal, Superior Court, Court of Quebec, and related bodies. The move follows a March 2026 report by La Presse that a judicial decision contained references to non-existent case law, raising concerns that a judge may have used generative AI without proper oversight.

Key provisions:

• Judicial reasoning is exclusively human: "Judging cannot be reduced to a technical exercise," the guidelines state. "It is an intrinsically human activity."
• Prohibited uses: AI cannot be used for substantive legal reasoning, analyzing or characterizing facts, assessing witness credibility, or any function that determines a legal outcome.
• Permitted uses: Judges may use AI for limited editorial tasks such as correcting, revising, rephrasing, or translating text. All output must remain under human control and be independently verified.
• Confidentiality requirement: Judges must not input draft judgments, case notes, or information about ongoing proceedings into AI tools.
• Full accountability: "No error can be attributed to the tool used." Judges remain fully responsible for all content in their decisions.

The courts note that no approved, secure institutional generative AI tool currently exists for judicial use. Until one is available, any AI use must be disclosed. The ruling sets a precedent for how regulated professions are drawing firm boundaries around AI's role in high-stakes decision-making.

Read the full article on Castanet Kamloops