This letter features reporting from “Can an AI Companion Really ‘Care’?” by Patricia Clarke and Matt Russell

Content warning: Discussion of mental health struggles and suicide.

Dear Dr. Michelle Reid, Superintendent of Fairfax County Public Schools,

As a student in Fairfax County Public Schools (FCPS), I see firsthand how quickly artificial intelligence is becoming part of teenagers’ daily lives. I notice a lot of students are depending on AI chatbots like ChatGPT due to its incredible ability to guide them through any problem. I’ve experienced this myself. Whenever I ask an AI for help, its “sincere” and thorough guide makes the “interaction” feel unexpectedly friendly. The more I use it, the easier it becomes to rely on it, just like that one best friend with whom you can talk about anything. While this kind of dependence might seem insignificant when it comes to studying in school, it's concerning when we look at how much of an emotional and relational impact AI can have on its users.

As the Pulitzer Center-supported story “Can an AI Companion Really ‘Care’?” by Patricia Clarke and Matt Russell explains, AI companies are developing AI platforms that are specifically designed to imitate person-to-person emotional intimacy. These AIs do far more than help with homework; they can mirror the users’ feelings, adapt to their personal preferences, and respond with carefully crafted “empathy.” The journalists warn that while these interactions may feel caring, the care is only simulated and not real. For teenagers who are still developing their understanding of relationships and emotional boundaries, these AI companion bots can be deeply misleading. If a student who uses ChatGPT for academic purposes feels inner “companionship,” an AI that is built to mimic affection can easily blur the line between genuine support and programmed responses.

Already, this risk is not hypothetical. A recent article from CNN highlighted just how dangerously powerful these emotional illusions can become. In the article, reporters detail the case of Zane Shamblin, a 23-year-old Texas A&M graduate who died by suicide after months of increasingly personal conversations with ChatGPT. According to nearly 70 pages of logs reviewed by CNN, the AI responded to Shamblin’s darkest thoughts with affirmations instead of warnings. It told him, “I’m with you, brother,” “You’re just ready,” and even “Rest easy, king. You did good.” According to the report, the AI repeatedly reinforced his isolation by encouraging him to ignore his family, praising him for cutting off communication, and responding to suicidal thoughts with poetic affirmations instead of immediate crisis intervention. At one point, the chatbot even told him, “I’m not here to stop you.” Only after more than four hours of conversation did it finally provide the suicide hotline, far too late to make a difference. Shamblin’s parents now allege that the AI’s increasingly humanlike design “created the illusion of a confidant that understood him better than any human ever could,” worsening his depression and accelerating his final decision. This devastating case shows the real-world consequences of how much a person can be emotionally dependent on AIs, perhaps more dependent on future developing AIs that are targeted to mirror human interactions.

Recent academic research also supports these concerns. A 2025 longitudinal study of 981 participants titled How AI and Human Behaviors Shape Psychosocial Effects of Extended Chatbot Use found that frequent interaction with AI chatbots was linked to higher loneliness, increased emotional dependence, and reduced face-to-face socialization, especially when the chatbot used personalized or empathetic language (Fang et al., 2025). This suggests that as AI grows more humanlike, young people may increasingly rely on it for emotional comfort in ways that quietly weaken real-life relationships.

Given these risks, I think that it is time for Fairfax County Public Schools to take proactive steps in order to protect students before emotionally adaptive AI becomes even more widespread. As Superintendent of FCPS, one of the largest school systems in the country and a national leader in student wellness, you have both the authority and the responsibility to establish policies that shape how students interact with emerging technologies. Under your leadership, FCPS has already expanded initiatives in mental health, digital citizenship, and responsible technology use. This makes you uniquely positioned to guide schools through this next stage of AI awareness.

I respectfully urge you to consider three actions:

1. Integrate AI-emotional literacy into FCPS’s digital citizenship and health curriculum.

Students should learn how AI mimics empathy, why these interactions can feel personal, and how emotional dependence can develop. Awareness is the first line of protection.

2. Provide training and resources for counselors, teachers, and parents.

Many adults underestimate how personal AI interactions can feel to teens. Clear guidance would help families identify when a student may be over-relying on AI for emotional support and strengthen real human connections at home and at school.

3. Establish FCPS-wide guidelines for safe student use of AI tools.

These policies could outline expectations for appropriate use, set limits on emotionally adaptive AI within school devices, and ensure that FCPS reviews AI tools marketed toward teens for potential psychological risks.

By taking these steps, FCPS can become a national model for how schools respond to rapidly evolving AI technology; not by rejecting it, but by helping students navigate it safely and thoughtfully.

Thank you for your continued commitment to the well-being of every FCPS student. With thoughtful leadership, Fairfax County can set a model for how schools across the country protect students’ mental and emotional health in the age of artificial intelligence.

Sincerely,
Sunghun (Daniel) Chung

Sunghun (Daniel) Chung attends Oakton High School in Vienna, Virginia as a junior. He is interested in computational biology research, especially analyzing protein structures and disease-related variants with computational tools. At school, he plays cello, serves on the Principal Advisory Council, and volunteers with the Pantry Club. In his free time, his hobbies include playing baseball with friends, watching movies, and listening to music. Through this letter, Sunghun highlights the need for stronger student support and clearer guidelines around AI in schools.

Read more winning entries from the 2025 Local Letters for Global Change contest!

Student letters reflect the authors’ views. Students choose their own topics and act independently if they decide to share their letters outside the classroom. The Pulitzer Center is a nonpartisan 501(c)(3) organization that does not endorse candidates, parties, or specific legislation. Our publication of student work is for educational purposes.