The inclusion of Artificial Intelligence (AI) into the field of early childhood education has led to the shift in the paradigm of great proportions. Thus, compelling to re-evaluate the role of an adult in the regulation of technologies. The paper aims to perform a critical examination of adult management and ethical mediation in terms of children using AI, placing the theory of Vygotsky and his Zone of Proximal Development (ZPD) as the main theoretical perspective. The employed methodology was a Systematic Literature Review (SLR) that uses the PRISMA protocol; therefore, eleven articles published since 2024-2025 were identified and underwent qualitative analysis. The research findings denote that the roles played by adults have changed to dynamic mediators of ethics and scaffolds as opposed to the traditional supervisory roles. The discussion shows that successful cognitive management requires the high procedural control, including limiting the screen time to less than fifteen minutes and using the data-driven content customization to reduce the cognitive load. The research also found out that the effectiveness of AI interventions largely depends on the level of self-efficacy of teachers and AI-TPACK literacy. The results led to the development of an Adaptive Mediated Scaffolding (AMS) Model, which has highlighted the necessity to balance digital automation and human intervention in order to prevent social isolation. The research provides important policy implications to policymakers and educators to create the guidelines that facilitate ethical and safe use of AI in enhancing the holistic growth of children during the digital age.