No more hiding: Putting AI to work for Education

"No more hiding"

Used in the title of this report, the term “hiding” highlights a paradox: although AI has already become part of educational practice, its use remains poorly documented and supervised.

The expression is therefore intended to encourage the education system to acknowledge existing practices and to recognise, collectively and transparently, the growing presence of AI in education – among students, teachers and the wider education community.

The objective must be explicitly stated: this is not a case for advocating the widespread or indiscriminate use of AI in schools or in teaching and learning. Rather, the question is no longer whether AI has a place in education – it is already there – but how to stop hiding while ensuring the quality of its contribution. This requires the education system to engage fully with the issue, committing now to embrace the changes and make the necessary investments to respond appropriately to this new reality.
 

Read the executive summary

Read the full report (French)

See the see the appendices (French)
 

A Guiding Principle: Learning with and without AI

The integration of AI into primary and secondary education should be guided by a fundamental principle that runs throughout this report: a dual approach, combining teaching "with AI" and "without AI". Distinct educational time slots would be organised:

• With AI, where it demonstrably adds educational value and its use is subject to clear and rigorous safeguards;
• Without AI, to preserve dedicated periods of learning from technology, where its use is neither necessary nor relevant.

This dual approach should be co-developed with teachers, who are best placed to determine when and how AI can deliver genuine educational value. It also requires assessment to place greater emphasis on human capabilities and collective intelligence, while curricula should be progressively revised to incorporate AI-related knowledge and skills. On 19th June, 2026, Prime Minister Sébastien Lecornu announced that, from the start of the 2027 school year, Year 11 pupils (seconde) would receive one hour of dedicated AI teaching per week.

The underlying challenge is naturally to safeguard students’ intellectual autonomy and their cognitive, social and emotional development, while preparing them to evolve in an environment in which AI plays an increasingly prominent role.

The outcome of a collective work

Given the fundamentally collective nature of this topic, the French AI and Digital Council (CIANum) established a multidisciplinary working group bringing together experts with a wide range of professional backgrounds:

• Étienne GRASS, Global Chief AI Officer of Capgemini Invent, member of the CIANum
• Guillemette PICARD, Founder at Cyan, member of the CIANum
• Antonin BERGEAUD, Economist, Professor at HEC, member of the CIANum
• Yann ALGAN, Economist Associate, Dean of Pre-Experience Programmes and Professor of Economics at HEC
• Litzie MAAREK, Partner and co-founder of Educapital
• Luis VASSY, Director of Sciences Po
• With the participation of Étienne CHAMPION, Rector of the Versailles Education Authority (académie)

This report is the outcome of a programme of hearings involving around 50 experts and key stakeholders from the Ministry of National Education, the EdTech ecosystem, social partners, the health sector, the research community, experts from the education system and teachers, education-related networks and organisation, as well as local and regional authorities.

Towards a managed integration of AI in schools: findings, action guidelines and recommendations

The report is structured around three successive sections.
First, the report examines the role AI actually plays within the French education system, both in and outside the classroom. It provides an overview of current uses of AI in education, while assessing the balance of risks and benefits associated with these uses. Several key questions frame this section, including the implications of “shadow AI” in education, the potential benefits for teachers, the need to adapt assessment methods, and the types of AI tools that should be prioritised in primary and secondary education.
Second, it sets out a number of action guidelines, considered essential to ensuring the responsible and effective integration of AI within the French education system:

• Reaffirm the purpose and vocation of school;
• Bring AI into the classroom;
• Put the principle of a dual approach into practice, combining teaching with AI and teaching without AI;
• Build a policy founded on monitoring and transparency;
• Train teachers on AI and with AI;
• Provide high-quality solutions, whether by procuring or by developing them;

Enable EdTech providers to achieve genuine economic viability.

Lastly, the third part outlines operational recommendations to concretise these action guidelines. It puts forward eleven recommendations, highlighting a number of urgent and demanding challenges:

(1) Teacher training is a fundamental priority in a context where initial teacher education has yet to keep pace with these new tools, while continuing professional development remains insufficiently structured. Another key priority is the sharing of good practice among peers, complementing traditional top-down training approaches.
With regard to teaching practice, the report also stresses the need to:

• Adapt teaching practices, particularly assessment methods;
• Protect and encourage experimentation in the classroom and within schools.

(2) Students must be supported and protected. This requires:

• Devoting a strong educational programme to AI;
• Promoting “Socratic” tools, which guide students through continuous questioning and reasoning, whilst regulating the tendency of models towards sycophancy and flattery;
• Addressing the mental health and cognitive risks associated with exposure to AI;
• Organise before the end of 2026 a conference on age-related safeguards and restrictions to define an acceptable level of exposure to AI;

(3) Where dedicated periods for AI use are introduced in schools, it is essential that the deployed tools are rigorously tested, properly governed and demonstrably deliver tangible benefits. In this regard, the report recommends:

• Systematic evaluation of the AI solutions to which student are exposed;
• Co-development of a charter for the use of AI in schools, setting out the appropriate framework and guiding principles;
• Accreditation of AI tools to ensure that they meet appropriate ethical and pedagogical standards;
• Testing of these tools by teachers, in order to preserve professional autonomy and pedagogical freedom

The hearings also highlighted a recurring difficulty for teachers in navigating the wide range of tools available. To address this, the report proposes the creation of a national platform (kiosk) of sovereign AI solutions. This would give teachers access to validated tools, selected according to their subject-specific needs.

(4) These recommendations form part of a broader, interministerial and public-private strategy aimed at supporting the creation of a sovereign EdTech ecosystem accessible to all teachers.

Ensuring the availability of a structured ecosystem of French and European providers is a priority, both from an educational perspective and in terms of digital autonomy, at a time when our dependence and vulnerabilities on foreign technology providers has continued to grow over the past years. These issues naturally raise the question of the funding that will need to be mobilised, as the cost of AI is likely to be significantly underestimated at present.