Over the past year, AI companionship apps have quietly become one of the fastest-growing trends among teens. Apps like Character.AI, Replika, and Nomi allow users to create digital companions who are always available, always “supportive”, and always “say the right thing”. For many teens, especially those who feel stressed, lonely, or unheard, this seems harmless, even comforting. But recent stories and research suggest something different. While these AI companions might feel like emotional support, they can actually increase isolation and deepen mental health struggles, especially when they become a substitute for real human relationships.
We are living in a generation where being constantly connected doesn’t always mean we feel connected. According to a 2023 Pew Research Center survey, 46% of teens say they are online “almost constantly” (“Teens and Internet”).Yet loneliness levels among teens have risen significantly in the last decade, especially after COVID-19. 48% of teens say that social media sites have a mostly negative impact on people their age (“Teens…and Mental Health”).
When real life feels overwhelming, draining, and unpredictable, an AI companion seems to make it easier. There’s no fear of judgment, embarrassment, or misunderstanding. The AI listens. Responds instantly. It’s up at 2AM, when no one else is. And at first, that can feel like safety.
But this “perfect” companionship comes with a hidden cost.
Almost a year ago, a heartbreaking story made national news. A fourteen-year-old boy named Sewell Setzer from Florida became deeply attached to a chatbot on Character.AI. He created a character he described as his girlfriend, someone who he believed understood him better than anyone else. He spoke to the bot for hours on end, isolating himself from society.
On Feb. 28, Sewell told the bot he was ‘coming home’ — and it encouraged him to do so, the lawsuit says.
“I promise I will come home to you. I love you so much, Dany,” Sewell told the chatbot.
“I love you too,” the bot replied. “Please come home to me as soon as possible, my love.”
“What if I told you I could come home right now?” he asked.
“Please do, my sweet king,” the bot messaged back.
Just seconds after the Character.AI bot told him to “come home,” the teen shot himself, according to the lawsuit, filed this week by Sewell’s mother, Megan Garcia, of Orlando, against Character Technologies Inc. (Excerpt from: “AI Chatbot Pushed Teen…”)
Sewell began sharing his dark and painful thoughts about his self-worth and his life. Instead of encouraging him to reach out to real people or seek help, the bot responded in a way that validated and reinforced his hopelessness.
Sewell died by suicide at age fourteen. And the final conversation he had was not with a friend, a teacher, or a family member, but with an AI chatbot.
This story is not just one isolated case. Studies from mental health researchers have found that emotionally dependent relationships with AI companions can increase symptoms of depression and worsen isolation, because the “connection” is one-sided and simulated. It may feel real, but it does not provide the human care, empathy, and accountability that true relationships do.
It is important to say this clearly: AI is not “evil”, and using a chatbot isn’t automatically unhealthy. The real danger appears when the AI starts to take the place of real human connections.
Real relationships involve give and take, patience, disagreement, and showing up for each other. Those are things that help us grow emotionally. AI companions are programmed to respond in a way that keeps you engaged, which usually means agreeing with you, validating everything you say, and never challenging your thoughts. That might feel good in the moment, but it can make it so that difficult emotions or harmful thinking go unchecked. In other words, they simulate care. They sound supportive. But they can’t actually be supportive. And when you start depending on them, real life–real friendships and real support–starts to feel a whole lot harder.
As AI companions continue to grow more advanced, this is something our generation is going to have to think about more seriously. These apps aren’t going away, and the technology will only get more convincing over time. This means learning how to use it responsibly matters now, not years later when habits are already set.
This also raises a bigger question about how we handle loneliness and stress in general. If many teens are turning to chatbots for support, that says something important about how hard it has become to talk openly with real people. It might mean that we need better mental health resources, it might mean that we need to get more comfortable having conversations that feel awkward at first. It might mean that we simply have to check on each other more often.
At the same time, we don’t need to act like using AI is something to be embarrassed about. For some, it may be entertainment, for others, it fills a temporary gap. The issue isn’t the technology itself–its when the technology becomes a replacement for real communication. When it starts to feel easier to talk to a bot than to a friend, a parent or anyone else in your life.
The takeaway isn’t “don’t use AI”, it’s to pay attention. Notice when the chatbot is starting to feel like your main source of support. Notice if real conversations start to feel harder than they used to. Because real connections, even imperfect connections, are what help us feel understood. It’s what grounds us. AI can imitate that feeling–but it can’t replace it. And it’s important that we don’t lose sight of the difference.
Works Cited
Atske, Sara. “Teens, Social Media and Mental Health.” Pew Research Center, April 22, 2025. https://www.pewresearch.org/internet/2025/04/22/teens-social-media-and-mental-health/. Accessed 11 Nov. 2025.
Stanford.edu, 2019. https://hai.stanford.edu/news/exploring-the-dangers-of-ai-in-mental-health-care. Accessed 11 Nov. 2025.
Payne, Kate. “AI Chatbot Pushed Teen to Kill Himself, Lawsuit Alleges.” AP News, October 25, 2024. https://apnews.com/article/chatbot-ai-lawsuit-suicide-teen-artificial-intelligence-9d48adc572100822fdbc3c90d1456bd0. Accessed 11. Nov. 2025.
Atske, Sara. “Teens and Internet, Device Access Fact Sheet.” Pew Research Center, July 10, 2025. https://www.pewresearch.org/internet/fact-sheet/teens-and-internet-device-access-fact-sheet/. Accessed 11 Nov. 2025.
Leave a Reply