Anne Maheux, PhD, discusses trends in youth adoption of generative AI

A recent discussion with Anne Maheux, PhD, assistant professor of psychology at the University of North Carolina at Chapel Hill, highlights emerging concerns about youths’ rapidly increasing use of generative artificial intelligence (AI), particularly chatbot applications.

Although these technologies have only recently become widely available, adoption among children and adolescents is already substantial. Maheux notes that roughly half of mid- to late adolescents are using generative AI tools, raising questions about potential risks and benefits that remain largely unstudied. Areas of concern include social development, mental health, sexual development, and exposure to harmful or age-inappropriate content.

Maheux’s team conducted a descriptive study using passive sensing data collected directly from youths’ smartphones and tablets. Unlike self-report surveys, which can be limited by memory errors or social desirability bias, passive sensing provides objective data on actual app use.

The findings reveal that approximately half of teens aged 15 to 17 years use generative AI applications. Usage was reported among 42% of young teens aged 13 to 14 years, 20% of preteens aged 10 to 12 years, and 10% of children aged 8 to 9 years.

On average, most youths spend relatively little time on these applications—approximately 2 to 3 minutes per day. However, usage patterns vary widely. A small subset of children, particularly preteens and young teens, spend significantly more time on generative AI tools, sometimes multiple hours per day on weekends.

Maheux suggests that reduced parental oversight and increased unstructured time on weekends may contribute to heavier use. While most engagement appears minimal, the high usage among a small group may present greater risk if negative consequences are associated with prolonged exposure.

The study also examined which types of applications youths are using. While the majority use mainstream tools such as ChatGPT, a notable proportion—41% of the top apps identified—are marketed for companionship, including simulated romantic partners or sexual role-play experiences.

Some of these apps are designed for adults and may contain explicit or inappropriate content. A small number of children in the sample were found to be using such platforms, raising additional concerns about social and sexual development.

Maheux emphasizes that the long-term implications of generative AI use are not yet clear. Potential risks include displacement of real-world social interactions, development of unrealistic relationship expectations because of agreeable AI responses, and exposure to harmful content.

She calls for longitudinal research to determine whether vulnerable youths are more likely to seek out AI tools or whether AI use itself contributes to later mental health challenges. Future work should combine behavioral data and self-reported experiences to better understand how generative AI shapes children’s development over time.

No relevant disclosures.

Reference

Maheux AJ, Akre-Bhide S, Boeldt D, et al. Generative artificial intelligence applications use among US youth. JAMA Netw Open. 2026;9(2):e2556631. doi:10.1001/jamanetworkopen.2025.56631