Study finds more than one in three sexually victimised Australian teens either tell no one or turn to AI chatbots, AFP warns

More than one in three Australian adolescents who reported being sexually victimised online either did not tell anyone or sought help from an artificial intelligence chatbot, according to a new national study led by the Australian Cybercrime Observatory at the University of Adelaide. The finding has alarmed Australia’s federal policing chief, who said law enforcement cannot be the only backstop for children exposed to online sexual harm.

What the research found

The study surveyed a nationally representative sample of older teenagers and measured both conventional forms of online sexual victimisation and the emerging role of generative AI in perpetration and disclosure. Researchers found that 35 percent of young people who experienced online sexual victimisation either told no one or confided in an AI chatbot rather than a person. The research also identified measurable rates of AI involvement in image based abuse and other forms of exploitation.

Associate Professor Tim Cubitt, deputy director of the Australian Cybercrime Observatory and a lead author of the study, described the results as a turning point in how child sexual harm occurs and how victims seek support. The study also reported that boys were disproportionately likely to disclose to AI compared with girls, reversing some conventional gender patterns in help seeking.

AFP response and policy implications

Australian Federal Police Commissioner Krissy Barrett, speaking in media interviews marking her first year in the role, said the research was ‘‘sober reading’’ and reinforced the limits of relying on policing alone to protect children. She urged the development of better safety nets outside law enforcement, including improved frontline support systems and services for young people who disclose harm to automated systems.

Commissioner Barrett cautioned against seeing AI disclosure as a sufficient substitute for human intervention. ‘‘Young people are turning to AI more and more for advice and for support and for guidance, and that should be concerning to all of us, not just parents,’’ she said in her interview. Barrett said the AFP is investing in technology that can triage and escalate reports where human assessment is needed, but also noted that by the time police become involved options are often limited.

Why adolescents are turning to AI

Researchers and child welfare specialists point to several reasons why teenagers may prefer AI chatbots as a first port of call. AI is available 24 hours a day and can feel less judgemental or embarrassing than talking to a parent, teacher, counsellor or police. For some young people, especially those worried about privacy or repercussions, an automated interaction may feel safer than disclosing to a person.

At the same time, the study found that AI is sometimes a direct tool of harm. Generative tools have been used to produce altered nude images and to coerce or groom young people. That dual role makes the policy response more complicated: AI can both exacerbate victimisation and absorb disclosures that would previously have reached human support services.

What experts and advocates recommend

• Improve accessible, youth friendly human support channels so children have low barrier, trusted alternatives to automated help.
• Require clearer safety and reporting pathways from AI service providers, including obligations to signpost human support and crisis services when minors disclose harm.
• Strengthen digital literacy and consent education so young people understand the limits of automated advice and where to get help.
• Increase investment in technology that triages disclosures to humans while protecting privacy and reducing false positives.

Child protection advocates said a mix of design changes, regulation and funded services will be needed. They emphasised that any reporting pathway must be culturally appropriate and accessible to marginalised groups, including First Nations communities and young people in regional Australia.

How this ties into existing Australian reforms

The findings arrive as Australia continues to implement sweeping online safety reforms enacted since 2024, including a minimum age regime for social media and expanded enforcement powers for the eSafety regulator. Those laws targeted platform design and age assurance, but they did not directly regulate conversational AI in the same way. The new research refocuses the policy debate on whether and how AI chat services should carry obligations to identify and respond to disclosures of harm made by children.

Policy makers will now face practical questions about measuring risk, protecting privacy, and avoiding unintended consequences. For example, design requirements that force AI services to report disclosures could deter other users from seeking help or create privacy and surveillance risks. Balancing these concerns will be central to debates that follow.

What happens next

Researchers said the report should prompt coordinated action across government, industry and community services. The Australian Federal Police signalled it will continue to develop technical tools to triage automated disclosures and improve investigative responses. Meanwhile researchers urged that regulators and technology companies urgently design clearer pathways so a teenager who tells a chatbot that they are being exploited is rapidly connected to an appropriate human service.

For parents, teachers and practitioners the immediate takeaway is the importance of open, nonjudgmental conversations with young people about online harms, and of teaching where to go for help. For policy makers, the study makes clear that the age of AI as a default confidant for vulnerable adolescents is already here, and public systems need to catch up.