Define an AI Usage Policy for Your School

From Individual Guidelines to School-Wide Policy

Having a personal stance on AI isn’t enough. As long as every teacher sets their own rules, students move from class to class shifting frameworks without understanding why, every challenge gets resolved case by case, and the most permissive colleague becomes the default reference. For integration to be consistent, you need a school-wide policy supported by all stakeholders: administration, teachers, parents, and students. This lesson guides you in building a pragmatic policy—precise enough to resolve concrete cases, yet flexible enough to survive the next wave of tool updates.

Stakeholders to Involve

In a school, a policy doesn’t start from scratch: it implements the national framework for AI use in education published by the Ministry of National Education and aligns with the school’s strategic plan. Whenever a tool processes student data, usage falls under the responsibility of the school principal—or the DASEN (Departmental Director of National Education Services) in primary schools—and the CNIL (France’s data protection authority) recommends systematically involving the academy’s Data Protection Officer (DPO); the board of directors or the pedagogical council may also be consulted.

Administration must validate and champion the policy; otherwise, individual initiatives will remain experiments without authority. To secure this support, present three things. First, concrete benefits: show time saved on creating differentiated assignments, quantified based on a sequence you actually prepared rather than a promise. Second, the risks of inaction, which are often the most convincing: students are already using these tools without a framework, and inequality in access to paid tools is already visible in their submissions. Finally, a phased implementation plan with clear milestones, which reassures leadership that the timeline is manageable.

With your colleagues, a collaborative workshop works better than a memo. A two-hour session allows you to map existing uses, including those no one admits to in the staff room, and then surface fears and resistance: loss of professional purpose, feelings of technical incompetence, or fear of widespread cheating. These objections don’t disappear because you contradict them; they lose their power once the team has voiced them. Only then can a small set of shared principles be adopted, and only then does designating AI leads by department or discipline make sense: these leads become the local point of contact that administration cannot be.

With parents, transparency defuses most concerns. It is also a legal obligation: if a tool uses student data, the CNIL reminds us that families and students must be informed, even if their consent is not strictly required in principle. Explicitly state what AI does and does not do in your school, describe the protections in place regarding data, ethics, and supervision, and indicate how they can support usage at home, which is often far less regulated than in the classroom. As for students, involve them in co-constructing the rules: a framework they helped write will be respected more than an imposed ban, and the discussions it sparks constitute a lesson in applied ethics in themselves.

Structure of an Effective Policy

Start with three to five founding principles, no more: beyond that, no one remembers them, and they cease to serve their purpose. They can be stated as follows. AI is a tool serving learning, not a substitute for thinking. Transparency about AI use is a core value. Each teacher adapts rules to their discipline and pedagogical goals. Protecting student data is non-negotiable. These short sentences describe no specific situation, and that is their strength: they serve as arbitration when an unexpected case arises and no detailed rule applies.

Next come context-specific rules, which translate these principles into concrete permissions. The table below revisits the spectrum levels defined in the previous lesson and associates them with the four most common situations.

This breakdown has the advantage of being immediately enforceable. A student challenging a remark on their homework is referred to a specific line written before the fact; a teacher hesitating on workshop instructions finds their answer there without calling a meeting. The “Justification” column is not decorative: it allows you to resolve a similar unforeseen case by reasoning through intent rather than the letter of the law. In primary schools, these levels apply within the limits of the national framework: no generative AI manipulation by students in elementary school, limited and teacher-supervised classroom use starting in grade 7 (4ème), and autonomous use in high school within an explicitly defined framework. Since the start of the 2026 school year, the Pix IA pathway is mandatory for all students in grade 7 (4ème), grade 10 (2de générale), and the first year of CAP (vocational certificate).

Finally, plan an update procedure. AI evolves quickly, and a policy written once becomes false within months: schedule a review every semester, fueled by feedback from teachers and students. Two questions are enough to prepare it—which rules proved unworkable, and what new uses have appeared that no line covers?

Your Turn

The preamble to your policy.

1. Draft about ten lines: founding principles and scope—which classes, which levels, which assignments
2. Verify consistency with the Ministry’s usage framework and the school’s strategic plan
3. Submit the text to ChatGPT asking for phrasing that a reader might interpret opposite to your intent
4. Have the Data Protection Officer review it if tools process student data
5. Present it to the pedagogical council as a working document

Official sources: Artificial Intelligences and Their Uses in Education (éduscol) · Data Controller: How to Implement AI Systems in Education? · Teacher: How to Use an AI System in the Course of Your Duties? (CNIL)

Key Takeaways

• A school-wide policy is more effective than an individual charter
• Involve all stakeholders, including students
• Adapt rules to the pedagogical context (exam, project, workshop)
• Three to five founding principles are enough to arbitrate unforeseen cases
• Plan for regular policy reviews
• In schools, the policy implements the Ministry’s usage framework and involves the Data Protection Officer