AI-powered chatbots significant potential for delivering scalable, personalized, and cost-effective mental health support at any time. However, they face ethical concerns and technical challenges. As techno-social systems, the technical and social dimensions of chatbots are deeply intertwined. Based on a comprehensive literature review and evaluation of state-of-the-art chatbot systems, this study identifies critical gaps in their design and deployment for mental health applications. Our analysis underscores the need for robust regulatory frameworks, improved safety measures, and rigorous empirical evaluation. We further identify key questions and directions for future research.