International Centre for Missing and Exploited Children research finds 19 per cent of young Australian victims tell an AI chatbot, versus 13 per cent who tell a teacher, doctor, police officer.
Children tell AI chatbots about sexual abuse more often than they tell teachers, doctors, counsellors or police, the first national study of online child sexual abuse in Australia has found.
The International Centre for Missing and Exploited Children, an Australian child-safety research organisation, released the study on July 7. It found that 19 per cent of young victims told an AI chatbot about the abuse, compared with 13 per cent who told a teacher, doctor, counsellor or police officer. More than a third told no human at all. The figures were independently reported by ABC News Australia.
At least one in 25 Australians has either experienced, or has a close friend who has experienced, online sexual victimisation involving AI before turning 18, roughly one young person in every Year 12 cohort.
AI's footprint is also growing inside the abuse itself. Non-consensual sexual image sharing involving children has AI as a driver in more than a quarter of cases, the study found. That share is rising fastest where the material is AI-generated, including "nudify" services that strip clothing from ordinary photographs and image-synthesis models tuned to school-age subjects.
The pattern is gender-asymmetric in two directions. Boys are more likely to be targeted by AI-generated abuse material; girls face higher rates of sexual image abuse overall. Colm Gannon, who leads work on the issue for ICMEC Australia, has warned that young men are particularly exposed to sexual extortion (coercion using intimate images) because they engage more freely with their own sexuality on camera, leaving more material that can be weaponised against them.
Gannon's explanation for the disclosure gap is blunt. "A 15-year-old will not disclose sexual extortion on Facebook when they are not supposed to be on Facebook," he told the Guardian. The adults positioned to receive disclosure are simply not on the channels where the abuse and the disclosure both happen. The chatbot, by contrast, is always on, anonymous and already in the same app where the abuse surfaced.
The researchers, led by University of Adelaide Professor Tim Cubitt, frame the finding as a readiness test for the institutions around children. If chatbots are absorbing the first disclosure, the question is what the next step looks like, and whether schools, helplines and the platforms where abuse circulates are equipped to meet children where they are.
The federal government has moved to ban AI "nudify" apps, services that generate undressed images of real people from ordinary photographs. Independent MP Kate Chaney has separately pushed to criminalise the underlying technologies used to generate child sexual abuse material.
Neither move touches what the ICMEC data actually exposes: that the first responder in many Australian cases of online child sexual abuse is now a piece of software rather than a person. Closing that gap means training parents and teachers for a conversation increasingly likely to start with a chatbot screenshot rather than a face-to-face disclosure, reconfiguring helplines to receive what those chatbots are absorbing, and pressing platforms to route disclosures to humans rather than the in-product assistants that currently catch them.
The next test of that readiness comes when Australian schools resume and the cohort measured in this study moves through another year of largely unsupervised time online.