In a context where the AI is changing the word of education, of work and of global and social powers, the Europe with the recent AI Act is trying to mark the boundaries of the use of a such disruptive and divisive technology. This quantitative and qualitative analysis from a specific pedagogical point of view tries to verify if and how the education issues have been considered. The first results show that educational issues are poor present in the text and that, despite an explicit appeal to the “anthropocentric perspective”, real innovative AI educational paradigms are not yet implemented. The main focus remains the risk-benefit worry. But in the future the framework of the AI Act could lead to a real innovation in the methodology and in the practices of educational use of AI.