8Declare the use of AI and what has been checked

In a resource created with vibe coding the code is written by the artificial intelligence (AI), and usually nobody has reviewed it line by line. The people who reuse it need to know this in order to decide how far they can trust it, just as they need to know which part its author has checked. The statement on the use of AI answers those two questions in a couple of sentences.

Transparency is one of the ethical principles of the «Guía sobre el uso de la inteligencia artificial en el ámbito educativo» (Guide on the use of artificial intelligence in education) of Spain’s National Institute of Educational Technologies and Teacher Training (INTEF), according to which students and teachers must know when AI is used, how it works and what its limitations are. The National Centre for Curriculum Development in Non-Proprietary Systems (CEDEC), in its article on the Conference on Free Software and Open Artificial Intelligence (seLIA), proposes sharing with each resource which tools were used, what was reviewed by hand and what was generated automatically. It sums it up by saying that this documentation «is not a layer of bureaucracy, it is a layer of honesty» towards the community that is going to reuse the resource.

In the European Union, the AI Regulation establishes in its article 50 some transparency obligations that apply from 2 August 2026. Most of them fall on the providers of AI systems, who must mark generated content so that its origin can be detected. Providers whose systems were already on the market before that date have until 2 December 2026 to do so, according to the amendment introduced by Regulation (EU) 2026/1744. For the people who use those systems, it requires disclosure in two cases: when they disseminate generated or manipulated images, audio or video that resemble real people, objects, places, entities or events and could be taken as authentic, what the Regulation calls a deep fake, and when they publish texts to inform about matters of public interest without these having gone through human review or editorial control for which someone is responsible.

An educational resource reviewed by its author does not usually fall into those cases, unless it includes realistic images, audio or video of that kind, such as a recreation of a historical event that could be taken as authentic. Outside that case, declaring the use of AI is good practice and not a legal requirement. Other countries have different rules, and it is advisable to consult them.

The statement must be brief and specific. It should indicate three things:

• The tool used. The name of the chatbot, the platform or the agent with which the material was created.
• What the person has checked. The correctness of the content, how it works, how data are handled or accessibility. This is the most informative part, because it varies from one resource to another.
• The date. Tools change quickly, and the date makes it possible to place the material in time.

A statement of this kind could be «Created with [name of the tool] in September 2026. The content has been reviewed and the operation tested. It collects no data». The minimum is for it to appear inside the material itself, next to the licence, so that it accompanies it when it is downloaded or embedded in another page. José Luis Miralles Bono does this in applications such as ScoreSpacer, whose footer states that it was created with the help of a specific AI and that all processing takes place in the browser. When the material is a project with several files, the same statement goes in its documentation, with a link to the decision log of recommendation 7, which is what explains how it was made.

In code, the involvement of the AI is always the same, so assigning a level to it adds nothing. In teaching content the opposite happens. The questions of a quiz may have been written entirely by the teacher, or the AI may have generated them and the teacher merely validated them, and that difference matters to the person who is going to use them in their classroom.

To indicate it, the Framework for integrating generative AI into educational tasks (MIAE) can be used, which describes AI involvement at six levels, from the completely human work of level 0 to human supervision of an autonomous AI at level 5. To do so it is enough to add the level to the statement, for example «Content at MIAE level 2, reviewed by the author». This guide applies it to itself, and its credits page states that it was produced at level 4, that of advanced collaboration between the person and the AI.