Research covering nearly 1,000 teenagers reveals gaps in parental awareness, growing emotional engagement with chatbots and new questions for companies designing conversational AI.
Artificial intelligence companions are becoming part of teenagers’ digital lives, but a new European study suggests the design of these systems may be outpacing families’ ability to understand their potential effects.
A report published October 22 by European policy research organization Interface found that 32% of surveyed French teenagers aged 11–17 had used AI companions, while 50% had used AI embedded in social media platforms.
The research, authored by Interface senior policy researcher Jessica Galissaire, draws on a YouGov survey conducted between June 24 and 30, 2026, involving 970 teenagers and 1,138 parents.
Its findings raise questions about the responsibilities of developers whose products increasingly occupy spaces traditionally associated with friendship, emotional support and personal advice.
AI Adoption Begins Before Adolescence
Among teenagers surveyed, 78% reported using general-purpose generative AI tools, while 55% had used voice assistants.
The research findings also show that 56% of children aged 11–12 used at least one generative AI category weekly. Within this age group, 16% reported regularly using AI companions.
Parents appeared less aware of their children’s activity. Among 11- and 12-year-olds, 40% reported regular use of general-purpose AI, compared with the 26% estimated by parents.
The gap complicates the effectiveness of parental supervision, especially when conversational interfaces are embedded in services children already access.
Emotional Engagement Becomes a Product Design Question
The report identified a notable pattern in how young people experience time spent with AI.
Among teenagers who used generative AI, 92% felt they spent longer than intended on at least one surveyed category. Specifically, 59% of AI companion users and 57% of users of social-media-integrated AI reported spending more time than they wanted.
Researchers also found that 50% of regular AI companion users reported negative emotions at least weekly, compared with 38% of surveyed teenagers overall.
These findings do not establish that AI companionship causes emotional distress. Teenagers experiencing difficulties may also be more likely to seek conversational support.
Separate research published in JAMA Pediatrics in June found that 19.2% of surveyed US adolescents and young adults aged 12–21 had sought mental health advice from AI chatbots. Among those users, 63.3% had not disclosed the activity to anyone.
The studies used different populations and measures, so their percentages are not directly comparable. Together, however, they underline why emotional AI interactions deserve closer examination.
The Challenge for AI Developers
For AI companies, these findings shift attention beyond conventional content moderation.
Product teams must consider how conversational memory, personalized responses, engagement incentives and age restrictions influence behavior, particularly when users treat software as a confidant.
The French study found that only 32% of parents felt sufficiently supported in helping their children navigate generative AI. When asked who should provide assistance, 51% identified schools and teaching staff, 41% AI providers and 30% public authorities.
This creates an emerging design and governance challenge: systems intended to be engaging must also provide meaningful safeguards, transparent boundaries and appropriate opportunities for human support.
The findings do not suggest teenagers should be excluded from every beneficial AI application. Instead, they indicate that the next phase of AI safety will depend increasingly on how products shape relationships—not simply what answers their models generate.
