<p>Artificial intelligence (AI)-based conversational agents are increasingly integrated into digital mental health applications designed to provide guidance, emotional support, and coping strategies for individuals experiencing psychological distress. Their rapid expansion has been driven by advances in large language models and the growing demand for accessible mental health resources in contexts where professional services remain limited. While existing research has largely focused on the effectiveness, usability, and ethical considerations of these technologies, less attention has been given to their potential implications for public health systems and help-seeking behaviors. This article examines AI-mediated emotional support from the standpoint of digital public health. It analyzes the limits of algorithmic empathy, the potential effects of AI-mediated interactions on help-seeking pathways, concerns related to algorithmic bias and data privacy, and the institutional implications that arise when these technologies begin to perform functions of emotional support in contexts of psychological vulnerability. Drawing on recent literature on digital psychiatry, AI ethics, and public health governance, the article proposes a conceptual governance framework based on differentiated levels of clinical and relational risk to guide the responsible integration of conversational agents into digital mental health ecosystems. From a public health perspective, developing regulatory and institutional safeguards will be essential to ensure that AI-mediated emotional support expands access to resources without displacing human care or transferring ethical and clinical risks to vulnerable users.</p>