Rights, Risks, and Obligations of Schools in the Use of Generative Artificial Intelligence in Education

The use of generative artificial intelligence (Gen-AI) in education brings new opportuniti­es for learning, teaching, and administrative support to schools and other stakeholders, while also raising important ethical and legal questions. This article systematically exa­mines the key legal risks, obligations of educational institutions, and recommendations for developing lawful and responsible practices for the use of AI in the school enviro­nment. As an “AI deployer”, schools must operate within the legal framework set by the General Data Protection Regulation (GDPR), the EU AI Act, and ZVOP-2. The EU AI Act classifies systems according to risk, with schools explicitly prohibiting high-risk systems such as the analysis of students‘ emotional responses or social scoring. For high-risk ap­plications, such as automatic assessment or profiling, the school is introducing strict sa­feguards, such as mandatory risk assessment (DPIA/FRIA), documentation, and ensuring human intervention in the decision-making process. When integrating AI into education, we find that there are serious risks, including „hallucinations“ (generating inaccurate information) and excessive dependence by students, which reduces self-confidence and critical thinking. Therefore, the teacher takes on the role of a mentor and guide who adjusts the level of control and independence according to the age of the students— from complete control in lower grades to teaching academic integrity in secondary and higher education. For the balanced and safe integration of AI, it makes sense to develop transparent internal policies. These policies should clearly define permitted uses (e.g., proofreading, idea generation) and prohibitions (e.g., original creative work) and require the disclosure of AI use in students‘ work. In this way, the school ensures transparency, appropriate staff training, and informs parents about the rules for safe use at home. Only with a comprehensive understanding of legal and pedagogical limitations will the use of AI contribute to more successful and critical learning.