Since June 2025, France has a national framework for artificial intelligence in schools. This ministerial text redefines what teachers can offer in class, distinguishing between students’ maturity levels in relation to generative tools. Educational technology is entering a phase where regulation now precedes usage, rather than the other way around.
National Framework AI and Education: What the June 2025 Text Requires
The framework published on June 14, 2025, draws a clear line. Awareness of AI is allowed from primary school, but the direct use of public generative AI by students is only regulated starting from the 8th grade. This distinction did not exist before: each institution was making its own rules.
The text states that AI must assist staff without replacing them. It recommends prioritizing simple, free, or sovereign solutions, which directs the purchasing choices of local authorities towards tools developed in Europe or open source.
Students in 8th grade, 10th grade, and first year of vocational training must now complete two dedicated AI courses via the Pix platform. These courses cover algorithmic biases, deepfakes, personal data protection, and the environmental impact of digital technology. The educational aspect takes precedence over the technical aspect: the goal is not to learn how to code a model, but to understand how systems that produce text or images work.
Several specialized publications, including analyses published on the Tic et Net website, provide insights into the evolution of these regulations and the educational resources that stem from them.
Protection of School Data: Strengthened Partnership between CNIL and National Education
On December 16, 2025, the Ministry of National Education and CNIL renewed their partnership. The agreement focuses on three areas: enhanced support for data protection officers in academies, compliance with GDPR for tools deployed in classrooms, and the production of specific resources on the use of AI in schools.
This renewal is significant. The massive arrival of generative AI tools in institutions raises concrete questions about the handling of student productions. When a student submits a text to an AI assistant hosted outside the European Union, the data passes through servers subject to foreign jurisdictions. The joint CNIL-National Education framework aims precisely to clarify these gray areas.
Field feedback varies on this point. Some teachers report that the GDPR-compliant tools offered by the institution are less effective than American commercial solutions. Others believe that the technical simplicity recommended by the national framework encourages more thoughtful usage, where AI is used for specific tasks rather than as a permanent assistant.
Édu-Up Program: Free and Sovereign Digital Resources
The ministry supports the production of educational digital solutions through the Édu-Up program. These resources are made available to teachers and their students for free or after voluntary registration. They meet the requirements of inclusive education and take into account the diversity of needs.
Among the projects supported in 2025, Écrivor, designed by a French teacher, uses AI to enhance student interactions with their own texts by providing targeted feedback on written production.
The common point of these projects is that they operate without massively collecting student data. This approach aligns with the recommendation of the national framework in favor of sovereign solutions.
Teacher Training in Digital Technology: The Gap Between Expectations and Reality
The regulatory framework exists. The tools are multiplying. However, teacher training remains the weak link in the chain. Several testimonies collected by the national press show that teachers report moving forward without clear guidelines, sometimes without initial or ongoing training on AI tools.
The problem is not limited to technical mastery. Assessing an assignment written with the help of generative AI requires new pedagogical skills. Distinguishing what belongs to the student’s personal production and what has been reformulated by a language model requires rethinking evaluation criteria.
The available data does not allow for conclusions about the exact extent of the training deficit. Academies are progressing at different rates, with some offering dedicated AI training sessions as early as the 2025 school year, while others are limited to occasional webinars.
Digital Skills and Student Learning
The issue goes beyond the question of AI. Educational technologies (edtech) cover a broad spectrum: adaptive learning platforms, virtual classrooms, administrative management tools. For these technologies to genuinely improve student outcomes, pedagogical support must precede technical deployment.
The edtech sector in France produces a variety of solutions, but adoption remains uneven between urban and rural institutions, between general education and vocational tracks. The hybridization of learning pathways (in-person and online) is progressing in higher education and vocational training centers, while primary and middle schools remain largely anchored in sporadic use of digital technology.
French educational technology is structured around a regulatory triptych: AI framework, CNIL partnership, Édu-Up program. This institutional architecture lays the foundations, but the gap between the texts and classroom practices remains measurable. The next step will depend less on the available tools than on the ability of academies to train teachers en masse on concrete and regulated uses.
