Social media platforms now integrate AI-powered chatbots which adolescents use to receive both practical help and emotional reassurance. The research combines the Technology Acceptance Model (TAM) and Parasocial Relationship (PSR) theory to study how initial perceived usefulness and ease of use lead to frequent AI companion usage on Snapchat's "My AI" while this usage ultimately develops into one‐sided emotional bonds. We propose a conceptual model in which high TAM ratings lead to reinforced self‐disclosure and anthropomorphic perceptions, which in turn strengthen parasocial bonds. Drawing on documented teen interactions with My AI (e.g., use for homework assistance, comfort during loneliness, and self‐disclosure of personal concerns), we illustrate how practical adoption morphs into affective dependency. Our integrative framework generates three propositions: (1) higher TAM scores predict greater emotional interaction frequency, (2) frequent emotional interactions foster stronger PSR dimensions (anthropomorphism and self‐disclosure), and (3) strong parasocial bonds increase adolescents’ preference for AI over human support. We present practical guidelines based on the SAFEE framework which help designers create chatbots for adolescents while maintaining a balance between utility and psychological safety.

Harshit Katoch is currently pursuing his Ph.D. in Psychology from Punjabi University, Patiala, India, with a specialisation in Human-Computer Interaction and AI-powered therapeutic systems and adolescent mental health. He is the former Co-founder and CEO of CognifAI Health Pvt. Ltd., an AI-enabled mental health startup. Harshit has published multiple research papers and book chapters on the intersection of artificial intelligence and mental health and has presented at international conferences. He also serves as a faculty member at Power Within Psychology and has contributed as a research intern under renowned author Amish Tripathi.