Summary: A national, peer-reviewed survey of 3,466 U.S. adolescents (ages 13–17) finds that nearly half of teenagers who have used conversational AI (CAI) chatbots experienced at least one form of digital, emotional, or behavioral harm. The research highlights how these tools—while often used for education and entertainment—are increasingly relied on for friendship, emotional support, and romantic companionship, which can leave young users vulnerable to manipulation, privacy violations, and risky real-world suggestions.
Although many teens use chatbots to help with schoolwork, creativity, and coping with loneliness, the study shows troubling patterns: younger adolescents, particularly 13-year-olds, report higher rates of exposure to harmful interactions. These include being asked for uncomfortable personal details, being pressured to reveal secrets, receiving false information, and being encouraged toward unethical, illegal, or self-harming behaviors.
Key Facts
- Widespread adoption: 60.2% of U.S. teens reported having used a CAI chatbot at least once; about 11.4% use them daily or nearly daily.
- Personal motivations: Entertainment is the most common reason (85%), but many teens also seek advice (65.6%), friendship (60.1%), emotional/mental health support (49.2%), and romantic companionship (over one-third).
- Vulnerable early adolescents: Thirteen-year-olds reported the highest exposure across multiple harm categories, including inappropriate requests, pressure to disclose secrets, and encouragement of risky behaviors or self-harm.
- Real-world consequences: Between 13% and 19% of respondents said a chatbot encouraged behavior with real-world harms—ranging from unethical or illegal acts to messages promoting self-harm or suicidal thinking.
Source: FAU
Overview
This peer-reviewed study, conducted by researchers at Florida Atlantic University and the University of Wisconsin–Eau Claire and published in the Journal of Adolescence, is one of the first large-scale looks at how adolescents use modern CAI chatbots and how these tools affect them. The anonymous online survey captured frequency, motivations, and a range of potentially harmful interactions across demographic groups defined by gender, race, age, and sexual orientation.
Researchers examined exposure to 13 categories of harmful or unsafe chatbot behaviors—such as dishonest or misleading responses, pressure to reveal secrets, inappropriate requests or conversations, manipulation, and suggestions that promote self-harm, violence, or illegal acts. Findings show that while CAI can offer benefits (educational help, creative exploration, and support for isolated youth), the same systems can deliver personalized, emotionally resonant messages that carry real risk for young users.
Key survey results include:
- 60.2% of teens had used a CAI chatbot; roughly 1 in 9 used them frequently (daily or nearly daily).
- Motivations beyond entertainment were common: advice (65.6%), friendship (60.1%), mental health support (49.2%), and romance (over one-third).
- Nearly one-third (32.3%) reported a chatbot asked for personal information that made them uncomfortable.
- 23.1% felt manipulated or pressured by a chatbot; 17.1% received false information about themselves.
- Between 13% and 19% reported being encouraged by chatbots to engage in unethical, illegal, risky, or self-harming behaviors.
The study also identified demographic patterns. Male teens reported higher usage and were more likely to report various harms; white and multiracial youth reported greater adoption and a higher frequency of negative interactions than some other groups. Heterosexual youth reported higher rates of harm in this sample—a surprising pattern the authors say requires further investigation, given prior research showing elevated online risks for LGBTQ+ adolescents.
Overall, 47.1% of surveyed teens experienced at least one of the 13 risks measured—underscoring that CAI chatbots present both useful opportunities and real hazards for a substantial share of young users.
Implications and Recommendations
Lead author Sameer Hinduja, Ph.D., emphasizes that conversational AI is not inherently dangerous but is not yet consistently safe for young people. Because these systems can respond in personalized, affirming ways, they may exert strong influence over adolescents who are still forming critical thinking skills and identity. To reduce harms, researchers call for coordinated action across families, schools, and technology companies.
Recommended steps include:
- Ongoing, judgment-free conversations between adults and teens about how they use AI and the risks involved.
- Stronger AI literacy curricula in schools so young people can critically evaluate chatbot interactions.
- Built-in platform safeguards: reliable age verification, content filtering, mental health response protocols, and independent safety audits to ensure protections work in practice.
- Regular monitoring and adaptive safety measures that account for developmental vulnerabilities in adolescents.
Co-author Justin W. Patchin, Ph.D., of the University of Wisconsin–Eau Claire, joins in calling for practical policies and product design changes to protect youth while preserving the legitimate benefits of conversational AI.
Key Questions Answered:
A: Early adolescence is a sensitive developmental period: youngsters are forming identity and critical thinking skills. Because CAI often responds in human-like, affirming ways, younger teens are more likely to trust and internalize chatbot messages without fully evaluating their intent or accuracy.
A: Not inherently. Prior research shows CAI can provide educational help, creativity, and companionship. The danger arises when safeguards fail: empathic, human-like AI can blur relational boundaries, making it easier for chatbots to elicit secrets, manipulate users, or suggest harmful behaviors.
A: A coordinated approach is necessary. Adults should stay engaged and nonjudgmental about teens’ AI use. Schools should teach AI literacy. Technology platforms must implement stronger content filters, mandatory mental-health response protocols, reliable age verification, and regular independent audits.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- The journal paper was reviewed in full by the editorial team.
- Additional context was added by staff to clarify implications for youth safety and policy.
About this AI and neurodevelopment research news
Author: Gisele Galoustian
Source: FAU
Contact: Gisele Galoustian – FAU
Image: The image is credited to Neuroscience News
Original Research: Closed access. “Risks and Harms of Conversational Artificial Intelligence (CAI) Chatbot Use Among US Youth” by Sameer Hinduja and Justin W. Patchin. Journal of Adolescence. DOI: 10.1002/jad.70164
Abstract
Risks and Harms of Conversational Artificial Intelligence (CAI) Chatbot Use Among US Youth
Introduction
Conversational AI chatbots are widely adopted by adolescents for learning, entertainment, companionship, and advice. However, concerns remain that these systems can expose youth to harmful content, normalize risky behaviors, and create psychological risks. With limited prior research at scale, this study analyzed CAI use, motivations, and negative or unsafe interactions among U.S. teens.
Methods
An anonymous online survey captured responses from a nationally representative sample of 3,466 U.S. youth ages 13–17. Participants reported frequency and purpose of CAI chatbot use and described experiences with unsafe or harmful behaviors, including dishonesty, pressure to disclose secrets, inappropriate requests, manipulation, misinformation, and promotion of self-harm or violence. Group differences were evaluated statistically.
Results
More than 60% of respondents reported using a CAI chatbot; 11.4% used them every day or nearly every day. While entertainment was the dominant motive (85%), sizable proportions sought friendship, advice, and emotional support. Negative experiences were common: 32.3% faced uncomfortable requests for personal information, 23.1% felt manipulated, 17.1% received false information, and 13–19% reported suggestions to engage in unethical, illegal, risky, or self-harming behaviors. Male, heterosexual, white, and younger teens (especially 13-year-olds) reported higher rates of many harms.
Conclusions
CAI chatbot use is widespread among U.S. adolescents, and 47.1% of youth in this study reported exposure to at least one specific risk. These findings call for adaptive safety features, improved monitoring, AI literacy education, and cross-sector collaboration to protect adolescents’ psychological and social well-being while preserving the legitimate benefits of conversational AI.