TraspAgente: A Scenario-Based Educational Framework for AI Transparency and Responsible Decision-Making

Generative Artificial Intelligence (GenAI) is increasingly being used by students for writing, information seeking, translation, image generation, and interaction with conversational agents. This diffusion poses a major challenge for Computer Science education: students must understand not only how AI systems work, but also when their outputs should be accepted, verified, revised, or rejected. Several existing studies share the belief that AI has necessitated a radical rethinking of educational models related to the use and relationship of technologies. The authors propose an AI literacy approach grounded in the premise that the diverse applications of artificial intelligence are not merely technological innovations to be learned and used, but also new models and forms of social change inextricably intertwined with technological evolution.Understanding and governing such changes requires multidisciplinary, interdisciplinary, and transdisciplinary theories, tools, and approaches. Following this perspective, the paper presents TraspAgente (TranspAgent), a scenario-based educational framework for teaching AI transparency as responsible and permission-aware decision-making in lower secondary education. This framework combines open-ended AI-use scenarios, a structured transparency sheet, collaborative debate, and iterative revision. It translates key concepts, including reliability, bias, authorship, privacy, accountability, transparency, and permission, into observable learning artefacts and a reproducible evaluation protocol. The paper describes the framework's theoretical foundations, instructional workflow, implementation materials, and assessment design. As a design-stage contribution, it does not claim empirical learning gains, but offers a replicable basis for future classroom studies on AI literacy and responsible AI use.