Chatting with AI to combat online grooming

Virginia Tech researchers are developing virtual tools to help teens protect themselves against cybergrooming. 

Using survey data from parents and teens and leveraging artificial intelligence, the research team created a chatbot that gradually advances the interactions from innocent to predatory, giving teens first-hand experience of cybergrooming tactics without the risk of actual harm.

“This will be an online program where youth can sign up to take it,” said Sang Won Lee, associate professor in the Department of Computer Science and project team member. “The AI chatbot is going to simulate the scenario so teens can experience and learn perpetrators’ behavior as it emerges over the multiple stages of cybergrooming. They can learn strategies that help them protect themselves. They learn what kind of responses will make them more vulnerable than others.”

Supported by the U.S. National Science Foundation, the International Computer Science Institute, the Commonwealth Cyber Initiative, the William T. Grant Foundation, and the Center for Human-Computer Interaction at Virginia Tech, the multi-institution project aims to not only educate young people about online dangers but empower them to mitigate potential threats. A beta version of the program is being tested and modified via a codesign session with teens, bringing fresh data to aging teen-influenced data sets.

“This research shifts our perspective from simply detecting cybergrooming to proactively preventing it,” said Jin-Hee Cho, principal investigator and associate professor for the Department of Computer Science at Virginia Tech. "Most existing cybergrooming research focuses on detecting predatory behavior. Our approach instead emphasizes empowering potential teen victims by equipping them with the awareness and tools to recognize grooming attempts early and make safer decisions. Rather than focusing only on the offender, we intervene on the human side to help protect those who are most at risk."

The online program builds on the research that produced the paper, “From Vulnerable to Resilient: Examining Parent and Teen Perceptions on How to Respond to Unwanted Cybergrooming Advances,” which was published at the May 2026 conference for the Association for Computing Machinery Conference on Human Factors in Computing Systems. According to the paper, 60% of teens surveyed reported experiencing some form of online sexual solicitation, while the parents estimated a far lower number.

Cybergrooming is a long-term process in which an offender builds trust and emotional connection with a minor online to enable sexual exploitation, coercion, or abuse. The progression occurs in six stages: initiating a friendship, gaining trust, weighing the risks of engaging, establishing exclusivity, making sexual solicitations, and reaching the conclusion where offenders attempt to move the interaction toward physical meetings. The process can take place over a series of days or even years and often starts with predators presenting as seemingly innocuous individuals or people the teens already know.

Researchers divided into two teams to create the chatbot. One developed the technical components and one focused on the human side, validating the program with volunteer teenagers. The design allows the chatbot to progress through the different stages of escalation, depending on the teen’s responses. The researchers believe allowing the teens to experience cybergrooming at its various stages sets this work apart from other efforts. 

“What makes our approach novel is instead of treating any stranger online like they are a predator from the beginning, we’re using a more realistic scenario where teens meet other people online,” said Pamela J. Wisniewski, principal research scientist from the International Computer Science Institute, an independent nonprofit affiliated with University of California Berkeley. “It’s about teaching [teens] the skills to regulate their boundaries in the earlier stages and then getting more protective of themselves if things escalate.”

Developing credible educational scenarios also presented ethical challenges. Researchers needed to create realistic scenarios and messages without pushing the boundaries of sensitive topics too far. But they believe striking the right balance has tremendous potential for helping countless teens and inspiring other researchers.

“I'm also a parent,” Cho said. “Through this research, I hope to inspire more AI and computer scientists to develop technologies that empower and protect vulnerable youth, and to pursue research that makes our society safer, stronger, and more compassionate.”

Original study: doi.org/10.1145/3772318.3790949