Canada has announced a national AI literacy initiative with an ambition to reach more than 50,000 K–12 educators and up to one million post-secondary students. The federal government and the Alberta Machine Intelligence Institute (Amii) are behind the partnership, announced on 9 September with a reported $13 million investment.
The student offer is a free three-hour course that participating colleges and universities can provide. The educator offer begins with a chapter of practical training and classroom resources, with five more planned through the autumn. The programme's organising sequence is simple: understand AI, use AI, and, if you choose, build with AI.
Those are substantial numbers. They also describe reach, not what changes in a classroom.
The teacher in the middle
Picture a Year 8 geography teacher who finishes an introductory AI course. They now know that generated text may contain bias and error. Next Tuesday's lesson is on migration data. What, precisely, should students do differently because the teacher took the course?
That is the step national programmes struggle to cross. The broad ideas are teachable at scale. The subject application depends on the material in front of the teacher, the misconceptions their pupils already have, and the time available in a lesson.
For that geography class, an activity might ask pupils to compare an AI summary of a migration chart with the chart itself. Which categories did the model collapse? Which figures did it omit? What claim can the data support, and what claim would require another source? The teacher is teaching geography and AI literacy in the same ten minutes.
Canada's first educator materials were scheduled for release on 21 September. Their value will become clearer when teachers can show that kind of classroom move, rather than only a certificate or a completed module.
Sequence matters
The programme's “understand, use, build” sequence is a useful corrective to training that starts with prompting tips. Students who can generate a polished answer before they understand what the system is doing may become faster at producing work without becoming better at judging it.
Understanding need not begin on a screen. A class can sort examples of good and poor evidence, argue about a missing perspective in a data set, or decide what information is too personal to put into a tool. Later, when students do use AI, those habits have somewhere to attach.
Building with AI is a further step. It should not become the only destination. Many pupils need the civic skill of recognising where a system is unreliable or unfair even if they never design one.
What the announcement does not settle
The Canadian announcement does not mandate a K–12 curriculum or specify how schools will assess AI literacy. It offers free resources and sets an ambitious reach target. It remains to be seen how many educators take up the offer and whether the learning transfers into lessons.
For UK schools, the question is familiar. A national course can give staff a common vocabulary. It cannot tell a history teacher which source comparison to run on Thursday, or a science teacher how to test a generated explanation of osmosis. That last step has to be built into subject planning.
Count the teachers trained. Then ask to see the lesson they taught differently. Both numbers matter.
Source: Government of Canada, National AI Literacy Initiative, 9 September 2026. Reach figures are programme targets, not completed participation.