The article substantiates and describes a complex of educational practices aimed at fostering the socially responsible use of Artificial Intelligence (AI) technologies among students. The relevance of the study is driven by the need to educate schoolchildren who are capable not only of using AI as a tool for solving academic tasks, but also of independently distinguishing the benefits of digital solutions, assessing the risks of their implementation (deepfakes, plagiarism, data leakage), and anticipating social consequences. The theoretical basis for the development was provided by classifications of methods from Russian pedagogy and the three-level model of upbringing outcomes by D.V. Grigoryev and P.V. Stepanov (ranging from acquiring knowledge to gaining experience in independent public action). The authors have designed four interconnected scenarios of educational practices for high school students: the awareness campaign “Digital Footprint: Camouflage or Exposure?”, the intellectual game “The Similarity Trap: Ethical Boundaries of Text Clustering,” as well as socio-simulation games “The Emotional Censor: How AI Evaluates Our Feelings” and “Mem-Laboratory: AI Erases Culture.” The proposed practices transform the student’s role from a passive user into a conscious participant in the process (“a human-in-the-loop”), capable of identifying neural network errors, protecting cultural identity, and applying technologies for creative purposes and solving real-world social problems. The materials presented in the article can be used in the development of upbringing programs, computer science courses, and digital literacy modules within the general education system.