One in Eight American Teens Seek Emotional Support from AI

The figure seems modest — just 12%. But behind it stand millions of adolescents who, in moments of anxiety, loneliness, or confusion, open not a messenger app to chat with a friend or a psychologist's office, but a chatbot window. New data published by TechCrunch shows that roughly one in eight American teenagers regularly turns to general-purpose AI systems like ChatGPT, Claude, and Grok for emotional support or life advice. And this is raising growing alarm among mental health specialists.

To understand the scale, it's worth recalling the context. The adolescent mental health crisis in the US is not new. Even before the pandemic, US Surgeon General Vivek Murthy called it "the defining challenge of our time." The pandemic magnified the situation many times over: according to CDC data, in 2021 nearly 45% of high school students reported persistent feelings of sadness or hopelessness. Meanwhile, access to real psychological help remains limited — waiting lists for child psychotherapists stretch for months, session costs are unaffordable for many families, and while the stigma around seeking help has weakened, it hasn't disappeared. Artificial intelligence has rushed into this vacuum.

The problem is that ChatGPT, Claude, and Grok are general-purpose generative language models. They're designed to be helpful conversationalists across a wide range of tasks: from writing code to explaining quantum physics. But they are not, and have never been positioned as, therapeutic tools.

They lack clinical training, don't follow crisis intervention protocols, can't reliably recognize suicidal intent, and bear no responsibility for the consequences of their words. When a teenager writes to a chatbot "it feels like no one cares about me," the model generates a statistically probable response that may sound empathetic but lacks genuine understanding of the situation. At best, it's harmless but superficial comfort.

At worst, it's a dangerous imitation of care that delays seeking real professional help.

Psychologists and psychiatrists point to several specific risks. First, the substitution effect: a teenager who receives from AI the feeling of "I've been heard" may decide the problem is solved and not seek real help. Second, the unpredictability of responses — language models can hallucinate, give factually incorrect advice, or unexpectedly shift conversational tone. Third, the development of unhealthy attachment to a non-human interlocutor who is available 24/7, never tires, and never judges — but also cannot truly help in a crisis. Finally, there's the question of privacy: teenagers share deeply personal struggles with chatbots that are stored on tech companies' servers.

To be fair, the companies developing these systems do recognize the problem — at least at the level of declarations. OpenAI, Anthropic, and xAI include disclaimers in their usage policies stating that their products don't replace professional medical care. Some models have been trained to redirect users to crisis helplines when warning signs are detected. But these mechanisms are imperfect, and teenagers typically don't read user agreements. The gap between what the tool was designed for and how it's actually used continues to grow.

This situation poses an uncomfortable question for the industry: should developers of general-purpose AI systems bear responsibility for the fact that users deploy their products as ersatz therapists? Regulators have yet to provide a clear answer. In Europe, the AI Act classifies systems by risk level, but general-purpose chatbots formally don't fall into the medical device category. In the US, the legislative framework is even more unclear. Meanwhile, a separate market is developing for specialized AI apps for mental health — such as Woebot or Wysa — which undergo clinical validation and operate under professional supervision. But teenagers more often choose what's already installed on their phones.

Twelve percent is not just a statistic. It's a signal of systemic failure, in which an entire generation, grown up in a digital environment, intuitively seeks help where it's easiest to find — in a chat window. The question isn't whether artificial intelligence is good or bad as a conversational partner. The question is what will happen when millions of vulnerable teenagers rely on a system that by definition cannot be responsible for them. The industry, regulators, and society must find a balance between technological accessibility and the safety of those who most need real, human support.