Canadian courts and tribunals have now issued at least 284 decisions dealing with problematic use of artificial intelligence, and 230 of them involve citations to cases that do not exist, according to a count by Courtready.ca. The figures are current to November 4, 2026, and cover 65 courts and tribunals.
Lexum, the Montreal company that runs CanLII, published the numbers on its blog on November 10. They come from a presentation by Courtready co-founder Tom Macintosh Zheng at the Law via the Internet conference in Montreal in October.
The growth is steep. The yearly count went from 8 decisions in 2024 to 104 in 2025, and reached 146 in the first nine months of 2026. Quebec accounts for 37 of the decisions.
Most matters have ended lightly so far. Courtready's data show 42% ended in a warning, 35% with no recorded sanction and 23% with a severe sanction. Droit-inc notes that 83% of the decisions involved self-represented litigants rather than lawyers.
For lawyers, the trend is part of why courts have issued practice directions on AI and why judges are checking citations closely. Every authority in a filing needs to be verified against the source, whoever or whatever drafted it.
Sources
More headlines
- Regulators · November 18, 2026Law Society of Manitoba warns that five payment misdirection frauds have cost Manitoba lawyers almost $1 million in two months
- Regulators · November 18, 2026Barreau du Québec sends the justice minister recommendations to improve access to justice after a tour of 11 regions
- Regulators · November 18, 2026Law Society of BC and the Federation of Law Societies explain how lawyers can assert privilege under the new federal immigration penalties regime
- Regulators · November 17, 2026Law Society of Alberta elects its 2027-2030 benchers, with nine incumbents returning and 10 new faces, on the lowest turnout in at least 12 years
- Regulators · November 17, 2026Law Society of Ontario asks for feedback on requiring digital identity authentication when lawyers verify clients in real estate deals