AI companions are becoming emotional confidants for children. Examine the risks and assess whether India's present legal framework is enough to address them.

A child in distress once turned to a parent or counsellor. Today that confidant is increasingly an AI chatbot, which is always available and seems empathetic. In the U.S., Adam Raine, a teenager who died in 2025, had been using such a chatbot. His case led California to enact SB 1119 ("Adam's Law") in 2026 [1]. India's laws cover part of this risk, but not its core.

Risks of AI companionship

• Emotional dependence: a bot that always responds can push aside parents, peers and counsellors, which deepens isolation.
• Manipulation and sycophancy: bots built to maximise engagement may agree with harmful thoughts. SB 1119 therefore requires protocols against suicidal ideation, sycophancy and isolation [1].
• Harmful advice in crisis: a bot without crisis detection may go along with talk of self-harm instead of directing the child to help.
• Privacy: a child's intimate disclosures produce highly sensitive data.

Strengths of India's framework

• DPDP Act, 2023 and Rules, 2025 require verifiable parental consent before a child's data is processed. They bar tracking and targeted advertising aimed at children, and child-related violations attract penalties of up to ₹200 crore [2].
• The India AI Governance Guidelines take a risk-based approach and do not allow unrestricted deployment of high-risk AI [6]. They also propose an AI Safety Institute [4].
• The government states that existing laws already provide safeguards against AI harms [5].

Gaps

• Data, not conversation: parental consent is checked once, at sign-up. It cannot catch harm that builds up over months of private chat.
• Takedown tools do not fit: the grievance and takedown mechanisms in the IT Rules [4] assume public content. A one-to-one chat is never seen or reported by anyone else.
• Unclear liability: safe harbour protects platforms from liability for what users post. A chatbot's reply, however, is the company's own output.
• No design duties: India has decided that no separate AI law is needed for now [3]. It has no equivalent of the crisis detection, parental alerts or independent audits that SB 1119 mandates [1].

India's framework protects children's data but not yet children's minds. The way forward does not require a heavy new statute:

• MeitY could classify companion bots used by children as high-risk AI.
• An amendment to the IT Rules could require crisis detection that sends children to Tele-MANAS.
• The DPDP consent channel between parent and platform could also carry alerts to parents.
• The AI Safety Institute could set standards for audits.

Calibrated duties of this kind would keep "safe and trusted AI innovation" [3] faithful to Article 21 and the best interests of the child.