From Digital Victims to Cyber Offenders Children, Artificial Intelligence and the Emerging Challenges of Criminal Accountability in India

Abstract Artificial intelligence intensifies children's dual position in digital environments. A child may be impersonated, groomed, profiled, extorted or depicted in synthetic sexual content; the same child may experiment with generative systems, automated deception or unauthorised-access tools and thereby become a child in conflict with law. This paper asks whether Indian law protects the first child without adultifying the second. Doctrinal analysis shows a substantial but fragmented framework. POCSO and Section 67B of the Information Technology Act address child sexual exploitation material, while the 2024 Supreme Court decision in Just Rights for Children Alliance v. S. Harish adopts a purpose-sensitive approach to possession, viewing and reporting. Yet entirely synthetic depictions that do not involve an identifiable or actual child raise unresolved questions of statutory language and harm. The 2026 amendments to the Intermediary Rules now regulate synthetically generated information through user declarations, labelling and due diligence, but do not independently define every deepfake as a crime. The Digital Personal Data Protection Act's child-specific Section 9 and the corresponding final 2025 Rules were enacted but, under the notified phased commencement, were not yet operational on 19 August 2026. For children accused of AI-assisted conduct, the Juvenile Justice Act remains adequate in principle because it demands individualised, rehabilitative inquiry; technical ability is evidence, not a proxy for maturity or mens rea. The paper recommends a child-specific, risk-based AI framework joining safety-by-design, precise synthetic-CSAM coverage, data minimisation, platform duties, evidence standards, diversion and specialised rehabilitation.