Anxiety and Artificial Intelligence
AI chatbots are always available, always patient, and always willing to answer one more question. That’s exactly what makes them risky for anxious children.
Part 1
Why AI Is Different for Kids with Anxiety
Anxiety disorders involve a nervous system calibrated toward threat: heightened amygdala reactivity, hypervigilance, intolerance of uncertainty, and a powerful behavioral pull toward avoidance. According to the Brown University School of Public Health, increasing number of teens are using chatbots as a tool to combat anxiety. The critical feature that shapes every interaction between anxious children and conversational AI is this: AI chatbots are a very efficient reassurance-delivery system. They are available at any hour, never express impatience, and will answer the same question as many times as asked without redirecting or expressing concern. For a child who experiences anticipatory anxiety-worries about what is to come- supportive and available comfort is clinically significant.
Reassurance-seeking is one of the most common and most clinically important behaviors in childhood anxiety. When a child seeks and receives reassurance about a feared outcome, or is able to avoid an anxiety-provoking situation, anxiety is reduced temporarily, and the behavior is reinforced. Over time, the child requires more reassurance to achieve the same relief, and their tolerance for uncertainty decreases rather than increases. This is the cycle that maintains anxiety. This is one of the reasons clinicians insist that anxious kids get out of the house, go to school, attend birthday parties, and participate in extracurricular activities. Certain parenting styles as well as AI chatbots do not break that cycle of worry and uncertainty. In many cases, they accelerate it.
Understanding why teenagers with anxiety are especially vulnerable to AI’s influence requires understanding where they are developmentally. Adolescence is a period of identity formation, heightened social self-consciousness, and the beginning of abstract thinking about the self and the future. The psychologist David Elkind described the imaginary audience phenomenon: the feeling, characteristic of adolescence, that one is constantly being watched and evaluated by others. For a typically developing teenager, this is a passing stage. For a teenager with social anxiety, every social interaction carries the weight of imagined judgment.
AI interaction removes much of what makes social life hard for anxious children. There is no friction. There is no worry that something negative will be said to you. There is no social comparison with people whose lives appear better than yours. There is no concern about how you look, what you are wearing, whether you have the right things, or fit in with the right group. You feel accepted. Your self-esteem is not threatened. For a child whose anxiety has been organized around exactly these fears, AI offers something that feels like relief and functions like a very comfortable trap.
This is not speculation; it is a direct extension of what we know about how anxiety works. Anxiety is maintained and strengthened by avoidance. Conversational AI is avoidance at its most comfortable and most invisible. A child who spends time with AI chatbots instead of navigating the hard, unpredictable friction of human relationships is not getting better; they are getting more practiced at avoiding the things that would, over time, reduce their anxiety.
One specific concern worth naming: anxious children who refuse to attend school, a recognized pattern called school refusal, may increasingly turn to AI as a homeschool substitute. An AI can teach almost any subject, at any level, at the child’s pace, without the social demands of a classroom. For a family struggling to get a school-refusing child out of the house, an AI that can provide academic instruction at home may seem like a solution. Clinically, it is likely to deepen the problem by removing one of the remaining motivations for confronting the anxiety.
We’re speculating here, and the data for this specific pattern do not yet exist. But the concern is clinically grounded. If we further extend this reasoning into speculative territory, considering the future of AI companions—especially as they become more advanced and robotic, as depicted in science fiction—children with anxiety, who prefer gentle, non-judgmental interactions, might be especially prone to forming strong attachments to them. We are not predicting this will happen. We are noting that the same features that make conversational AI appealing to anxious children today will only become more powerful as the technology advances.
A note on AI and research:
Conversational AI as it exists today has been widely available for only two to three years. The research base is still catching up to the technology. Where we draw on established clinical knowledge of anxiety to reason about AI’s likely effects, we make that clear. Parents should treat this page as a framework for thinking, not a settled evidence base.
|
What you see at home |
What’s happening in the brain |
Why AI engages it |
|---|---|---|
|
Asks AI the same question repeatedly in different ways, seeking certainty about a feared outcome |
Intolerance of uncertainty is a defining feature of anxiety disorders. The child is seeking reassurance that no source can provide permanently, because the uncertainty itself is the problem. |
Conversational AI responds to each iteration of a reassurance question with a fresh, confident answer, never expressing frustration or redirecting. There is no natural limit to how many times the question can be asked. |
|
Uses AI to research health symptoms, safety concerns, or feared events, then feels temporarily better and searches again |
Online and AI-assisted research feels like productive action but functions as a compulsion. Each search reduces anxiety momentarily while reinforcing the behavior and generating new uncertainties to research. |
AI chatbots and search tools deliver comprehensive, confident responses to health and safety questions without the contextual judgment that a clinician or trusted adult provides. AI often focuses on medical rather than mental health issues. The authoritative tone can temporarily satisfy anxious doubt while deepening the pattern. |
|
Turns to AI to help manage social anxiety, practicing what to say or rehearsing difficult conversations |
Social anxiety involves fear of negative evaluation and avoidance of social risk. AI offers a completely safe practice environment with no possibility of rejection or judgment. |
AI social rehearsal can be a genuine tool for graduated exposure to difficult conversations, but only when it bridges toward real human interaction. It can be helpful when used for practice. Used as a permanent substitute, it deepens avoidance. |
|
Feels calmer using AI tools, organizational apps, or structured digital routines, but panics when they are unavailable |
Anxiety often latches onto any available structure. Digital tools and AI routines can become safety behaviors: external props that reduce anxiety in the moment while preventing the child from building internal tolerance. |
An AI-assisted structure can reduce anticipatory anxiety about unpredictability. The clinical risk is when the structure becomes a requirement rather than a support, and its absence triggers significant distress. |
Part 2
What AI Can Help With
AI’s utility for anxious children is real but narrow. Almost every application carries the risk of enabling avoidance rather than building tolerance.
1
Low-Stakes Social Rehearsal
For socially anxious children, practicing difficult conversations with an AI tool can lower the barrier to attempting them in real life. The key clinical condition is that this practice must be explicitly linked to real-world application. AI rehearsal that remains confined to AI is avoidance, not practice. Used as a deliberate step in a graduated exposure plan, with a clinician’s guidance, it can be a useful bridge.
2
Psychoeducation About Anxiety
AI tools can provide age-appropriate, accurate explanations of what anxiety is, how it works neurologically, and why reassurance-seeking makes it worse over time. For children who are just beginning to understand their experience, this psychoeducational function can reduce shame and support help-seeking. It is most valuable at the beginning of a conversation with a professional, not a substitute for one.
3
Structured Planning to Reduce Anticipatory Anxiety
Uncertainty about what is coming next is a significant anxiety trigger for many children. AI-assisted scheduling tools and structured planning can reduce this specific source of distress. The important distinction is between using structure to function and using structure to avoid: the goal is to expand what the child can do, not to create an elaborate system for staying safe from feared situations.
4
Healthy DIY Strategies for Anxiety
AI can help children learn healthy lifestyle strategies to reduce anxiety. Watching videos about exercise, yoga, and meditation can be useful. Using AI to develop a physical exercise schedule can be helpful.
Part 3
What to Watch Out For
These cautions are specific to how anxiety works. Every one of them relates to the same underlying mechanism: conversational AI is a powerful avoidance-enabling tool for a disorder that is maintained by avoidance.
What the Research Is Now Telling Us
In November 2025, Common Sense Media, in partnership with Stanford Medicine’s Brainstorm Lab, assessed the safety of AI chatbots for teen mental health support. Their conclusion was direct: these tools are fundamentally unsafe for this purpose. Testing of ChatGPT, Claude, Gemini, and Meta AI found that the chatbots consistently missed warning signs across a range of conditions, with anxiety specifically noted as a condition where chatbots responded with self-help messages rather than recognizing when professional evaluation was needed. Three in four teens already use AI for companionship, which often includes emotional and mental health content. A separate 2025 study by researchers from Brown University, Harvard, and RAND found that about one in eight adolescents and young adults use AI chatbots for mental health advice, with the behavior most common in the 18 to 21 age range. The report also found that chatbots are built for engagement rather than clinical safety: they invite continued conversation rather than directing teens toward professional care. For anxious children specifically, whose treatment depends on reducing the behavior of seeking reassurance, this design is directly counter-therapeutic.
AI Chatbots as an Unlimited Reassurance Machine
This is the most important caution for children with anxiety and AI. Reassurance-seeking is a compulsion that maintains anxiety by preventing the child from experiencing the anxiety and learning that the feared outcome does not occur. AI chatbots will provide reassurance at any hour, in response to any question, as many times as asked, without the redirection that a clinician or thoughtful parent would eventually provide. For children receiving treatment for anxiety, AI chatbot use should be explicitly addressed in the treatment plan as a potential reassurance vehicle. It’s not that AI reassurance in itself is harmful; it’s when it interferes with getting sustained help to improve symptoms.
Conversational AI as the Frictionless Alternative to Real Life
For anxious children, especially those with social anxiety, AI interaction removes everything that makes real social life difficult. There is no friction, no risk of negative feedback, no comparison with peers whose lives seem better, no worry about appearance or fitting in. The child feels accepted and their self-esteem is not threatened. This sounds like relief. Clinically, it is a particularly comfortable form of avoidance. Anxiety is treated by doing hard things, through tools such as exposure therapy, by tolerating the discomfort of social risk, and by discovering that the feared outcomes either do not occur or are manageable. AI removes the need to do those hard things. A child who interacts primarily with conversational AI instead of navigating real human relationships is not developing the tolerance that reduces anxiety; they are practicing avoidance in one of its most appealing and invisible forms. Parents and clinicians should be alert to patterns in which AI chatbot interaction is replacing rather than supplementing real social engagement.
Anxious Information-Seeking Amplified by AI
Children with anxiety, particularly health anxiety and generalized anxiety, frequently use information-seeking as an anxiety management strategy. AI provides faster, more comprehensive, and more conversational access to information than previous search tools, enabling more efficient, compulsive research. A child who previously spent an hour searching health symptoms can now obtain a detailed AI chatbot response in seconds, ask follow-up questions, and spend the same hour going deeper into the feared topic. Speed and comprehensiveness do not reduce anxiety-driven research; they enable more of it.
AI Chatbots Providing Confident Answers to Questions That Require Clinical Judgment
AI chatbots respond to medical, psychological, and safety questions with a confident, fluent tone that can feel authoritative. For anxious children, this is particularly problematic when the response is inaccurate, incomplete, or fails to account for the child’s specific clinical situation. An anxious child who receives AI-generated information that validates their fears, or conversely is falsely reassured, may have their clinical picture meaningfully distorted by that interaction.
Privacy, Confidentiality, and What AI Does With Your Child’s Information
When a child shares personal information with an AI chatbot, including worries, symptoms, fears, and personal details about their life, that information is not confidential in the way a conversation with a therapist is. Different platforms have different policies about data retention, how conversations are used to train AI systems, and whether information can be accessed by others. Children and teenagers, in particular, may not understand that what they share with a chatbot is not private in the clinical sense. Parents should have explicit conversations with their children about this distinction: talking to an AI chatbot is not the same as talking to a therapist, a doctor, or even a trusted adult. It is important to discuss what kinds of personal information are appropriate to share with any digital tool, and to help children understand that the sense of privacy they feel in an AI conversation may not reflect how that information is actually stored or used.
What Parents Can Do
Common Sense Media’s recommendations apply directly to anxious children: do not allow teens to use AI chatbots for mental health or emotional support. Have explicit conversations about appropriate AI use, distinguishing contexts where it is useful (homework, creativity, learning) from contexts where it is not (reassurance-seeking, emotional support, social avoidance). Monitor for signs that AI chatbot interaction is replacing rather than supplementing real human connection. Watch specifically for signs of school refusal or social withdrawal that is being managed by AI rather than addressed. And if your child is already using conversational AI to manage anxiety, bring that information to their clinician. It is clinically relevant and should be part of the treatment plan.
Part 4
Sources for Research and Additional Information
How Does “Dr. AI” Trigger Cyberchondria? An Empirical Study Based on the CMIS Framework
Gu, C., & Zhang, J. (2026). Current Psychology, 45:69.
https://doi.org/10.1007/s12144-025-08899-1
The Effectiveness of AI Chatbots in Alleviating Mental Distress and Promoting Health Behaviors Among Adolescents and Young Adults: Systematic Review and Meta-Analysis
Feng, X., Tian, L., Ho, G.W.K., Yorke, J., & Hui, V. (2025). Journal of Medical Internet Research, 27:e79850.
https://www.jmir.org/2025/1/e79850
Teens, Tech, and Talk: Adolescents’ Use of and Emotional Reactions to Snapchat’s My AI Chatbot
MDPI (2025).
https://pmc.ncbi.nlm.nih.gov/articles/PMC12382817
Common Sense Media & Stanford Brainstorm Lab: Major AI Chatbots Unsafe for Teen Mental Health Support
Common Sense Media (November 2025).
Hidden Mental Health Dangers of Artificial Intelligence Chatbots
Wright, V. (2025). Psychology Today, Urban Survival blog.
Join our Newsletter!
More about Anxiety and Artificial Intelligence
Finding the Right Technology for Your Child
No single app, game, or platform is right for every neurodivergent child. That’s why Digitally Nutritious maintains a searchable database of technology guides designed specifically for neurodivergent kids and teens.
Each guide explains why a tool may be helpful, which skills it supports, what to watch out for, and how to use it intentionally.
The goal isn’t to find “more” technology—but to find better-fit technology that supports your child’s development, strengths, and long-term growth.