AI in Catalonia's classrooms, question by question

According to the Government's official summary, the guidelines set three bands. Under 14 there is no direct use: the teacher gives an introductory AI literacy lesson. From 14 to 18, use is allowed with the authorisation of the family or legal guardian, guided by default and with teacher supervision. From 18, use is autonomous and responsible.

Some press reports mentioned authorisation from the teacher; the Government document says family or legal guardian. The age of 14 has a legal precedent: Spain's data protection act (LOPDGDD, article 7) sets 14 as the age from which a minor can consent to the processing of their data; below that, the consent of parents or guardians is required.

The bands go by age, not by year group. In the simulator on this page you can see, for example, that in ESO year 3 (age 14-15), in September, students born from October to December are still 13.

Those with a data protection impact assessment (DPIA) validated by the Department. The guidance for teacher training published on the CRP network in June 2026 says that the Department is reviewing the assessments for the tools it makes available to teachers: Gemini and NotebookLM, from Google, and Copilot, from Microsoft.

The other tools are not made available to teachers, have not been analysed and therefore cannot be recommended in the training activities certified by the Department. A tool being free or popular does not make it suitable for the classroom.

No personal data, neither of students nor of teachers: names, grades, reports, photographs or emails. The same guidance asks to anonymise data and, whenever possible, to switch off training on conversations and persistent memory.

The Department's framework of teacher digital competence in AI includes this as an indicator of the basic stage (1.A.3.3): applying security measures such as data anonymisation, switching off memory and restricting training.

Yes, if it is declared. The Generalitat's official guide to the baccalaureate research project says that, whenever AI has played a relevant role in the process or in the writing, this must be stated in the written report, for transparency and traceability. Among the possible annexes it includes a responsible-use declaration on AI, with the detail of the requests made (the prompts).

At the start-of-year press conference, as reported by El Món, the Department added that AI contributions must be reviewed, checked and reworded with one's own judgement, and that the oral defence is the moment to check the student's authorship and understanding. The official guide already describes the defence as a dialogue in which the student explains the process, justifies the decisions and answers the committee's questions.

In baccalaureate year 2, part of the class turns 18 during the school year: for these students, according to the bands, use is already autonomous.

As support, yes; as the only reference for giving a grade, no: the Department's guidance asks to avoid this. The EU AI Act classes as high risk the systems that evaluate learning outcomes, decide admission to an institution, assign the educational level or monitor exams.

The same Act prohibits, since 2 February 2025, systems that infer people's emotions in educational institutions, unless it is for medical or safety reasons.

There is also no reliable way to know whether a piece of work was done with AI: detectors are wrong too often to accuse anyone.

The Department recommends setting up a working group in the school to draw up a proposal for incorporating AI into educational work, and plans training for the teaching staff.

The Government has also announced a common framework for the use of AI in schools (2026 to 2028) and instructions for the use of digital education platforms, which were modified after the opinion of the Catalan Data Protection Authority and are pending signature.

And the EU AI Act asks, since 2 February 2025, that those who use AI systems take AI literacy measures for their staff (article 4), without requiring a specific level. For a school, this means training those who use them and being able to show it.

They are described in the Department's document “Competència digital docent en intel·ligència artificial” (December 2025, CC BY 4.0 licence), which organises them into four dimensions: understanding and safe use of AI, ethics and a humanist view, pedagogy with AI and professional development.

Each dimension has three stages, basic, intermediate and advanced, which correspond to acquire, adopt and create. The framework builds on UNESCO's AI competency framework for teachers (2024) and on the AI supplement to the European DigCompEdu framework.

The XTEC AI space lists two certifiable 30-hour courses, “IA pràctica per a docents” and “Cocreació de situacions d'aprenentatge amb IAG”, within the shifts of the training offer (on 5 October, enrolment was closed). There are also two open courses on Odissea, without certification: “La intel·ligència artificial i l'educació: un binomi de futur” (15 hours) and “La intel·ligència artificial a l'aula” (30 hours).

For anyone, the same space recommends CiutadanIA, from UPC, IDEIA and the Generalitat.

That until age 14 the school will not have their children use AI directly, and that between 14 and 18 their authorisation will be needed for them to use it, in a guided way and with teacher supervision.

Outside school, the Government wants to extend the age-14 threshold to generative AI in libraries, youth facilities, educational leisure and public sports activities, and to ask providers to verify age and offer modes adapted to minors.