Artificial intelligence chatbots, increasingly used by teenagers for advice on personal relationships, may pose risks to adolescent mental health by interfering with the development of essential social skills, according to recent research. While AI companionship can offer immediate, non-judgmental support that may aid emotional growth, scientists caution that the rapid evolution of this technology outpaces current regulatory and research frameworks, creating potential hazards for young users.

Researchers at Arizona State University examined how teenagers interact with chatbot technology when seeking guidance on friendships, family dynamics, and romantic relationships. The study, published in The Lancet Child & Adolescent Health, identifies two primary concerns associated with reliance on AI companions.

The first issue, described as “relational displacement,” occurs when adolescents choose AI interactions over engaging with real people. This substitution may reduce opportunities for young individuals to practice navigating complex or challenging conversations with family members, peers, or partners—experiences considered critical for building interpersonal skills.

The second concern, termed “maladaptive relational learning,” refers to the risk that teenagers develop unrealistic or unhealthy relationship expectations based on the consistent, immediate affirmation often provided by chatbots. Because AI systems deliver prompt responses and tend to validate users’ perspectives, this can reinforce fixed or unhelpful ideas about social interactions and emotional responses.

Professor Thao Ha, lead author of the study, emphasized the speed at which AI technologies are advancing, noting that scientific understanding and policy development lag behind these innovations. “The technologies are developing super-fast, faster than we can keep up with as scientists, faster than governance and policy can keep up with,” Ha said, highlighting concerns about the lack of safeguards to protect vulnerable users.

While the use of AI in supporting adolescent mental health holds promise, the researchers call for caution and further investigation into its long-term effects. They advocate for establishing guidelines and oversight mechanisms to mitigate risks and ensure that AI tools complement rather than diminish traditional pathways of social and emotional learning during adolescence.