An AI policy for your school, in one page
Schools that use AI tools are “deployers” under the EU AI Act. Since 2 February 2025, emotion recognition in education institutions is banned (with narrow medical or safety exceptions), and schools must support the AI literacy of their staff. Stricter rules for high-risk uses in education, such as admissions, grading and exam proctoring, now apply from 2 December 2027. A one-page policy should cover purpose, allowed uses, data protection, assessment, transparency and training.
What the EU AI Act asks of schools today
| Since | What applies | What it means for a school |
|---|---|---|
| 2 Feb 2025 | Ban on certain practices, including emotion recognition in workplaces and education institutions | Do not use tools that infer students’ or staff emotions from faces, voices or other biometric data. The exception for medical or safety reasons is narrow. |
| 2 Feb 2025 | AI literacy (Article 4) | Take measures to support the AI literacy of staff who use AI tools. No specific test level is required. |
| 2 Aug 2026 | General application, including transparency duties | Be open when content is AI-generated, and expect labels from tool providers. |
| 2 Dec 2027 | High-risk rules for education uses (Annex III) | Extra duties for AI used in admissions, evaluating learning outcomes, deciding the level of education, or monitoring students during tests. |
The one-page template
Copy these headings and fill each with two or three sentences that fit your school.
- Purpose. Why the school uses AI, and the values it follows: learning first, fairness, privacy.
- Allowed uses for staff. For example: lesson planning, differentiation, feedback drafts that a teacher checks.
- Allowed uses for students, by age. UNESCO suggests 13 as a minimum age for using generative AI tools in the classroom. Check each tool’s own terms.
- Not allowed. Emotion recognition, entering personal data into public tools, AI-only grading of high-stakes work.
- Data protection. Use only tools approved by the school. Run a data protection impact assessment where the risk is high. Respect the national age of digital consent (15 in Greece).
- Assessment and honesty. When AI may be used in assignments, and how students show it.
- Transparency. Label AI-generated materials for students and parents.
- Training. How the school builds AI literacy for staff each year.
- Review. Who updates the policy, and when (at least yearly).
Good guidance to lean on
- European Commission: Ethical guidelines on the use of AI and data in teaching and learning for educators (updated March 2026).
- OECD and European Commission: AI Literacy Framework for primary and secondary education (June 2026), with four domains: engage with AI, create with AI, manage AI, shape AI.
- DigComp 3.0 (November 2025), where AI is part of every competence area.
- UNESCO: Guidance for generative AI in education and research (2023).
Questions coordinators ask us
Is ChatGPT “high-risk” under the AI Act?
Not in itself. High-risk depends on the purpose. Using any AI system to grade exams or decide admissions falls into the high-risk education uses.
Do teachers need a certificate in AI?
No. The AI Act asks schools to support AI literacy. It does not require a specific test or certificate.
Sources
- Regulation (EU) 2026/1744 (Digital Omnibus on AI)
- Council of the EU: artificial intelligence timeline
- European Commission: guidelines on prohibited AI practices
- European Commission: ethical guidelines on AI for educators
- OECD and European Commission: AI Literacy Framework
- European Commission: when is a DPIA required?
- UNESCO guidance on generative AI in education (ESEP summary)
Facts checked on 16 October 2026. Found something out of date? Tell us and we will fix it.
