Generative AI has complicated a basic assumption behind child sexual exploitation and abuse material investigations: an abusive-looking image may no longer be a direct record of something that happened exactly as depicted. It can involve existing CSEAM altered using AI, an ordinary photograph of a real child transformed into fabricated sexual imagery, a completely synthetic child who never existed, or fake material used to threaten or blackmail an identifiable child.
The scale is already significant internationally. NCMEC received more than 400,000 CyberTipline reports with a generative-AI nexus in 2025, including over 182,000 involving possession, generation or attempted generation of AI-linked CSEAM. Since tracking began in 2023, more than 158,000 images and videos have been categorised as generative-AI CSEAM and over 275 victims identified. Comparable India-specific figures are not publicly available.
Indian law, however, anticipated computer-generated material before the recent AI boom. POCSO’s definition, amended in 2019, expressly includes computer-generated images indistinguishable from an actual child and modified images appearing to depict children. Section 11 also covers threats involving fabricated depictions, while Section 67B of the IT Act includes creating digital images depicting children in sexually explicit material.
The legal position becomes more complicated when the depicted child is entirely synthetic. Several provisions refer to “a child”, “uses a child” or material “involving a child”. While the law clearly recognises computer-generated imagery, Indian courts may eventually have to determine how these provisions apply when no real child corresponding to the image ever existed.
India’s February 2026 IT Rules added another layer by expressly covering synthetically generated information. Services capable of generating or modifying synthetic content must deploy technical measures to prevent their systems from being used to create unlawful material, including CSEAM. Regulation therefore begins not only when material is uploaded, but potentially before it is generated.
For investigators, AI creates new attribution and classification problems. Police may need to determine who generated an image, which service was used, whether a real child’s photograph was supplied, whether existing CSEAM contributed to the output, and whether the material was stored, circulated or used for coercion. The image may be synthetic, but the investigation and evidence cannot be.
Read the full article on Chapter One Magazine.
