University AI Policies in 2026 – What’s Actually in Place Across Canada
Working in academic integrity at UBC I have a cross-institutional view that might be useful. Here’s what I’m seeing across Canadian universities in 2026.
The institutions that moved fast in 2023-2024: most have now walked back blanket bans and landed somewhere in the “disclose and demonstrate” space. They require students to document AI use and to demonstrate they understand their own work through oral components or process submission.
The institutions that moved slowly: some still have vague policies that say “AI use is prohibited” without defining what AI use means, without any detection methodology, and without any clear process for investigation. These are the ones generating the most complaints and the most inconsistent outcomes.
What nobody has figured out: how to handle the discipline process when detection evidence is inherently probabilistic. Academic integrity processes were designed around documentary evidence. Probabilistic scores break those processes in fundamental ways.
Academic integrity in the AI era requires us to be more precise about what we’re actually assessing.
3 Replies
Join the discussion.
Log In to Reply
exactly matches what Hao et al. (2025) found on institutional variation. the slowest-moving institutions are generating the most problematic outcomes. sharing with my department.
what's missing from most university AI policies I've seen is any acknowledgment that faculty use AI too. the policies are entirely student-facing. until institutions address faculty AI use in research and teaching, the ethical foundation of student restrictions is on shaky ground.
The probabilistic evidence problem is the key insight here. University disciplinary processes require a standard of proof. What standard does a 78% Turnitin AI score meet? None that's legally defensible. Institutions that understand this have moved toward conversation-based investigation and process documentation. Those that haven't are sitting on liability.