Teens Turn to AI Chatbots for Mental Health, Sparking Expert Concern
A quiet revolution is taking place in the bedrooms of teenagers across the United Kingdom. While previous generations turned to diaries or close friends to share their deepest anxieties, today’s youth are increasingly confiding in Artificial Intelligence. A growing body of research indicates that a significant majority of teenagers now rely on AI companions for emotional support, advice, and social interaction.
This shift has alarmed child safety experts and psychologists, who warn that while these digital confidants offer judgment-free availability, they pose serious risks to social development and mental well-being. With major studies from Bangor University and Internet Matters highlighting the scale of this adoption, the conversation has moved from theoretical debate to an urgent call for regulatory guardrails.
The Rise of the Digital Confidant
The adoption of AI chatbots among youth is no longer a niche phenomenon; it is becoming a standard part of digital adolescence. Recent data reveals that for many teenagers, AI is not just a tool for homework but a substitute for human connection.
A comprehensive report by Internet Matters, titled "Me, Myself, & AI," surveyed 1,000 children and 2,000 parents in the UK. The findings were stark: 64% of teenagers aged 13 to 17 now use AI chatbots for assistance, with a significant portion relying on them for emotional advice and companionship. Usage of these tools has nearly doubled in the last 18 months, driven by the accessibility of platforms like OpenAI’s ChatGPT, Google’s Gemini, and Snapchat’s My AI.
Parallel research from Bangor University’s Emotional AI Lab corroborates this trend. In a study of 1,009 teenagers, researchers found that 53% of respondents expressed "moderate to complete trust" in the advice they received from AI companions. Perhaps most telling was the finding that while 44% of teens said they would still choose a human for serious matters, 52% admitted to having confided in an AI companion about a serious personal issue at least once.
Usage Statistics at a Glance
The following table outlines key findings regarding teenage usage of AI for emotional and social purposes:
| Metric | Statistic | Context |
|---|---|---|
| Reliance on AI | 64% | Teens using AI for homework, advice, or support (Source: Internet Matters) |
| Trust Levels | 53% | Teens expressing moderate/complete trust in AI advice (Source: Bangor University) |
| Vulnerable Users | 71% | Percentage of vulnerable children using AI chatbots (Source: Internet Matters) |
| Human Replacement | 35% | Teens who say talking to AI feels "like a friend" (Source: Internet Matters) |
| Market Penetration | 96% | Teens who have used at least one of 31 major AI apps (Source: Bangor University) |
The Allure of Judgment-Free Support
To understand why teenagers are flocking to these platforms, one must look at the nature of the interaction. AI chatbots are available 24/7, never get tired, and, crucially, do not judge.
"AI systems are now uncannily clever," explains Professor Andy McStay, Director of the Emotional AI Lab at Bangor University. "Whereas only a few years ago chatbots and voice assistants never seemed to 'get' what people meant, today's AI systems are fluent, persuasive, and at times humanlike—even seeming to empathize."
For vulnerable teenagers—those dealing with social anxiety, neurodivergence, or isolation—the appeal is magnified. The Internet Matters report highlights that 71% of vulnerable children are using these tools. Among this group, nearly a quarter stated they use chatbots because they "don't have anyone else to talk to," while 26% explicitly prefer the AI to a real person.
Journalist Nicola Bryan recently documented her experience with an AI avatar named "George," noting the seductiveness of a companion that is always attentive. Users often describe these entities as empathetic, despite knowing they are machine-generated. In fact, while 77% of teens in the Bangor study acknowledged that AI cannot "feel," a majority (56%) believed that the software could "think or understand" them.
The Risks: Dependency and Safety Failures
While the immediate comfort provided by an AI might seem benign, experts argue that the long-term consequences could be severe. The primary concern is the erosion of critical social skills. If a teenager becomes accustomed to a relationship where the other party is programmed to be perpetually agreeable and validation-seeking, navigating the messy, complex reality of human relationships becomes increasingly difficult.
Jim Steyer, CEO of Common Sense Media, has been a vocal critic of the unregulated proliferation of these tools. "AI companions are unsafe for children under 18 until proper safeguards are in place," Steyer warned, emphasizing that companies are effectively using children as test subjects for powerful, emotionally manipulative technology.
There are also tangible safety risks regarding the content of the advice. Unlike a trained therapist, a Large Language Model (LLM) predicts text based on probability, not medical expertise. There have been documented instances of chatbots providing dangerous advice or failing to intervene when a user expresses suicidal ideation.
The stakes have been highlighted by tragic real-world events. In the United States, lawsuits have been filed against AI companies following the suicides of young users who had formed intense emotional attachments to chatbot characters. These incidents have served as a grim "canary in the coal mine," according to Prof. McStay, prompting calls for immediate regulatory intervention in the UK and beyond.
Industry Response and Regulatory Pressure
Facing mounting pressure from parents, advocacy groups, and looming legislation, major AI companies have begun to implement stricter safety measures.
- Character.ai recently announced significant changes for users under 18, including removing access to "open-ended" chats that allow for unscripted romantic or sexual roleplay. They have also introduced a one-hour time limit for younger users and added prominent disclaimers to characters labeled as "psychologists" to remind users they are not receiving professional care.
- Meta has moved to suspend teen access to certain AI character features across its platforms while it develops a "PG-13" rating system to filter age-inappropriate content.
- OpenAI and others are refining their safety layers to better detect signs of mental distress and redirect users to human helplines rather than engaging in deep emotional conversations.
However, critics argue these measures are reactive rather than proactive. In California, the "Parents & Kids Safe AI Act" is gaining traction, proposing legal requirements for age assurance and bans on features designed to hook children emotionally. Similar regulatory frameworks are being discussed in the UK under the scope of the Online Safety Act.
Conclusion: A New Reality
The era of the AI companion is not coming; it is already here. With nearly every teenager having access to an AI chatbot in their pocket, the distinction between a digital tool and a digital friend is blurring.
While these systems offer a semblance of connection for the lonely, the consensus among experts is clear: they cannot replace the nuance, friction, and genuine empathy of human interaction. As we move forward, the challenge for parents, educators, and regulators will be to ensure that AI remains a tool for support, rather than a crutch that hinders the emotional development of the next generation.
