Trustworthy and Open Educational Practices with AI: Small Language Models for Digital Sovereignty in Higher Education

The emergence of generative AI has radically transformed the educational landscape, offering unprecedented opportunities for personalization and openness in learning. However, these advances pose critical challenges related to the accuracy of information, the trust of the educational community, and technological dependence on large corporations. This chapter addresses how Small Language Models (SLMs), integrated into multi-agent architectures and connected to learning management environments through open standards, can contribute to safer, more reliable, and sustainable open educational practices. Based on experimental work at the University of Alicante, a Smart Learning Applications (SLApp) model is presented that enables the creation of intelligent assistants for teachers and students, ensuring compliance with the European AI Regulation and privacy protection. The chapter also reflects on the need to link pedagogical innovation and institutional digital sovereignty, placing trust and openness as core principles of education in the age of AI. Thus, a path towards higher education that harnesses AI’s potential while preserving transparency, autonomy, and the common good is proposed.