AI companions are increasingly becoming emotional confidants for children. In the context of constitutional child protection and India’s existing digital laws, suggest a balanced regulatory framework for AI-mediated psychological risks. (150 words)

AI companions are always available and apparently empathetic, but prolonged interaction can create emotional dependence, manipulation and harmful advice. This is a child-safety challenge beyond conventional cybercrime.

• Article 21 protects life and personal liberty; Article 39(f) requires conditions enabling children’s development in freedom and dignity. • The IT Act, 2000, DPDP Act, 2023 and POCSO Act, 2012 address important digital-safety dimensions, but do not comprehensively regulate psychological harms caused by conversational AI. • Harm may emerge gradually through thousands of seemingly innocuous exchanges, complicating detection, attribution and investigation.

• Mandatory age assurance, child-specific defaults, parental controls and clear notification when safety settings are disabled. • Crisis-detection protocols with safe escalation to guardians, counsellors and emergency services; prohibit responses that encourage self-harm or substitute for professional care. • Independent safety audits, red-team testing, incident reporting, explainable logs and rapid grievance redress. • Privacy-by-design: data minimisation, purpose limitation and restrictions on profiling children for engagement or commercial manipulation.

• Adopt risk-based, proportionate rules rather than blanket prohibition, preserving beneficial innovation and access. • Place duties on developers, deployers and app stores, with stronger obligations for systems designed or likely to be used by minors. • Combine regulation with digital literacy, school counselling and accessible human support; review safeguards as capabilities evolve.

Article 21; Article 39(f); child safety; AI companions; age assurance; crisis response; safety audits; DPDP Act; risk-based regulation