Quebec's Ministère de l'Éducation published a 22-page 2026 update of its teacher guide to generative AI, which sets three criteria for any classroom use: pedagogical relevance, ethical principles, and legal obligations. The guide is written for teachers from preschool through secondary school, vocational training, and adult general education, and it leaves the choice of approved tools to each school organization. It presents itself as a support tool built around reflection questions, not as a rule book.
Key takeaways
- The 2026 text updates a teacher guide first published in 2024 and runs 22 pages.
- Teachers are asked to reflect before, during, and after each use, starting with whether the school organization has authorized the tool.
- Five ethical principles frame the guide: digital sobriety, quality, equity and inclusion, transparency and explainability, and agency.
- Staff should treat anything typed into a generative AI system as public unless the terms say otherwise, and should not enter personal information without approval from the organization's responsible officer.
- The guide sets criteria and questions. It does not measure whether AI improves learning.
Where the guide comes from and who it covers
The Guide pour une utilisation pédagogique, éthique et légale de l'intelligence artificielle générative is a Quebec government publication dated 2026. It describes itself as an update of the teacher guide released in 2024 and says its content will change as laws, regulations, government frameworks, research, and school practice evolve. Two Quebec research groups, Obvia and GRIIPTIC, are thanked for contributing to the writing.
The guide connects to existing Quebec documents rather than standing alone. Its ethical principles come from a public-sector statement of principles on responsible AI issued by the Ministère de la Cybersécurité et du Numérique and from the Ministry of Education's own "Recommandations éthiques de l'intelligence artificielle en éducation," a 55-page document dated 2026. It also ties AI use to the twelfth professional competency for teachers, which concerns using digital tools, and to the province's digital competency framework.
Criterion one: pedagogical relevance
The first criterion asks teachers to reason in three steps. Before using generative AI, a teacher names the goal, decides whether the tool adds anything, checks that the system is authorized by the school organization, and considers whether students have the digital skills to use it critically. The guide adds that systems designed for education are preferable, so that use does not turn into dependence. During use, the teacher checks whether the output meets the objectives and holds students' interest, and adjusts the request or switches tools if it does not. Afterward, the teacher reviews what worked and considers whether a solution without AI would serve better.
A separate part covers assessment. Generative AI may help build rubrics and exercises or draft feedback, and it can save time, but it may not replace the teacher's professional judgment. Output must be reviewed for accuracy. A teacher who wants a system to evaluate student work must confirm that the organization has approved it and that the work contains no personal information. The guide also reminds readers that the intellectual property in a student's work belongs to its author, and it advises teachers to present acceptable and unacceptable uses to students as clear rules tied to each activity.
Criterion two: five ethical principles
| Principle | What the guide asks of teachers |
|---|---|
| Digital sobriety | Weigh environmental cost, write precise requests, limit their number, and prefer smaller models where possible; check whether a non-AI resource would do the job |
| Quality | Verify generated content, since systems can produce confident falsehoods and are not search engines |
| Equity and inclusion | Watch for bias and for groups, such as Indigenous communities and linguistic minorities, that training data may represent poorly |
| Transparency and explainability | Say when AI was used, naming the system and how it was used, because AI cannot be an author and its reasoning is hard to inspect |
| Agency | Keep human judgment central, avoid technological dependence, and avoid describing the system with human qualities |
Each principle comes with questions a teacher can ask, for example whether students can validate generated content themselves. The guide also sets out a literacy grid with four verbs applied to teachers and to students: understand, use, analyze, and create.
Criterion three: legal obligations
The legal section names three areas: information security, personal information, and copyright. On security, the guide warns that content entered in a generative AI system is generally not confidential by default and may be used for other purposes or shared with a third party. Unless the terms of use are thoroughly understood, the information should be treated as public, and sensitive, confidential, or personal information should not be entered. Where such use is necessary, the system must be analyzed and approved by the school organization's information-security officer.
On personal information, the guide points to Quebec's act on access to documents held by public bodies and the protection of personal information, which makes a public body responsible for the personal information it holds. It notes that a student's answers on an assessment can identify the student indirectly when combined with other details, and recommends checking with the organization's privacy officer. On copyright, ordinary rules still apply, and a person who enters a protected work must hold the rights to reproduce it. The guide's summary box states that tool selection and permitted uses are decisions for each school organization, and that teachers must follow its rules.
What the guide does not do
The guide gives criteria, not outcome data. It cites OECD and UNESCO documents as background for why AI might help students with special needs or struggling students, but it does not test those claims. Nor does it list approved products, which the text assigns to each school organization. Readers outside Quebec should treat it as an example of how one Canadian ministry frames teacher decisions, since education is a provincial responsibility in Canada and other provinces publish their own guidance.
This article summarizes a government document in general terms. It is not legal advice, and a school's privacy officer or counsel decides how the rules apply to a specific tool.
FAQ
Is the Quebec guide a binding rule?
The guide describes itself as a support tool and a set of reference points. It points to binding laws, such as the provincial privacy act, but the guide itself offers reflection questions and recommendations.
Who decides which AI tools Quebec teachers may use?
Each school organization. The guide states that tool selection and permitted uses fall to the organization, and teachers are responsible for following its directions.
Can a teacher paste student work into a chatbot to get feedback?
Only if the organization has approved the system and the work holds no personal information. The guide notes that answers can identify a student indirectly, so the privacy officer should be consulted when in doubt.
Does the guide show that AI improves learning?
No. It frames decisions and asks teachers to check whether the goal was met, and it even suggests asking whether a solution without AI would serve better.
Related reading
- The impact and potential of AI in education
- AI chatbots in schools
- The role of an education department
- AI and the learning experience
More in AI in education.
Last verified: October 2026
Sources
- Guide pour une utilisation pédagogique, éthique et légale de l'intelligence artificielle générative, Ministère de l'Éducation du Québec, 2026
- Recommandations éthiques de l'intelligence artificielle en éducation (cited in the guide), Ministère de l'Éducation du Québec, 2026
- Énoncé de principes pour une utilisation responsable de l'intelligence artificielle par les organismes publics (cited in the guide), Ministère de la Cybersécurité et du Numérique du Québec, 2025