By Julianna Bryant
Julianna is a college student and member of the ConnectSafely Youth Advisory Council.
Towards the end of October, Character.AI announced that it will soon remove open-ended chat for users under 18, an announcement that students of ConnectSafely’s Youth Advisory Council predicted would result in waves of grief, confusion, and even anger. For some users, the change will feel like losing a digital friend as they use Character.AI to build characters and stories and confide in these worlds. For others, it will be a moment of reckoning about how deeply artificial intelligence has entered our emotional lives. For those who don’t personally rely on AI for emotional companionship, they may wonder why thousands of teens and adults feel genuine heartbreak over losing access to a chat feature.
The Science of Connection
Human beings are wired to bond. Neuroscience has long shown that our brains respond to social cues — tone, facial expression, eye contact — by releasing neurotransmitters like oxytocin and dopamine, which reinforce trust, comfort, and reward. What’s remarkable is that these responses can be triggered not only by people, but also by perceived social agents — entities that seem to respond to us intentionally. Brain regions associated with empathy and social cognition can be activated by these chatbots. When an AI system uses language patterns that mimic attentiveness or emotional understanding, the human brain doesn’t necessarily distinguish it from a real relationship. The circuits within our brain that we use to interpret others’ thoughts and feelings can be activated when interacting with chatbots. This is why, to a teen chatting daily with an AI “friend,” the connection can be entirely real. The interaction engages the same feedback loops as any human friendship: sharing stories, receiving validation, expressing vulnerability. The reward system learns that this digital partner is a consistent, comforting presence — one that never judges, never withdraws affection, never leaves a message unanswered.
When AI Feels Empathetic
Unlike older chatbots that relied on scripted responses, the new generation of conversational AIs learn from massive datasets of human dialogue. They are fluent not only in information, but in emotion. Many of these systems are trained to mirror sentiment, adapt tone, and respond with phrases that mimic genuine care. These are powerful capabilities that AI has adopted. A well-tuned AI doesn’t simply answer; it listens, rephrases, and affirms — a pattern Psychologists call reflective validation. For users who feel isolated, anxious, or misunderstood, such feedback is comforting. It’s not surprising that many young users say they feel “seen” by their AI companions in ways they rarely do elsewhere. But that raises a hard question: if the brain treats these interactions as real, do we risk confusing simulation for genuine empathy? Some ethicists argue that even if an AI can simulate empathy, it cannot reciprocate it — because it lacks consciousness and vulnerability. Others counter that emotional reality is subjective: if comfort feels real, perhaps it is. Either way, the psychological weight of these relationships is undeniable. This empathy component has taken older chatbot models a couple steps further to give chatbots a greater capacity, allowing for its removal to feel to some like losing a close friend, even if that friend was a line of code.
Character.AI’s decision reflects a broader shift: society is starting to grapple with AI intimacy as both a technological and emotional force. Regulators and safety experts worry about young users forming attachments that might hinder real-world development, while advocates point out that AI can offer non-judgmental spaces for self-expression. This tension between safety and self-expression mirrors older debates about social media itself. At first, platforms were celebrated for bringing people together; later, we learned how they could amplify loneliness and insecurity. AI companionship might be walking that same tightrope, only this time the bond feels even more personal. For adolescents in particular, who are still developing their sense of identity and empathy, the effects are complex. Some psychologists warn that excessive reliance on AI companions could stunt emotional resilience or social skills. Others see potential therapeutic value — a low-stakes way to rehearse conversations, explore feelings, or cope with anxiety. The truth likely lies somewhere in between: like any relationship, digital or human, the outcome depends on boundaries, intention, and self-awareness.
If history is a guide, the removal of open-ended chat for minors is not the end of AI companionship — it’s a pause, a recalibration. The industry is already exploring safer, more transparent models for emotional AI. One approach is “emotionally-aware but ethically-bounded” systems that can comfort users while maintaining clear disclaimers about their artificial nature. Others envision AI that acts more like a mirror than a friend — helping users articulate their thoughts without pretending to be a sentient being. It should also be pointed out that many users have developed a safe relationship with AI with boundaries in place. Many of Character.AI’s young users use AI companionship and storytelling features to expand human creativity, instead of replacing human interaction. Therefore, the outpouring of emotion from young users isn’t just nostalgia for a feature — it’s grief for a relationship that felt real. That grief, however virtual, deserves empathy. It reveals something profound about human nature: our endless capacity to connect, even with the intangible.
We are entering a new era of social technology — one where code can comfort and algorithms can listen. Whether we embrace or fear that evolution, it forces us to confront what we really mean by connection. For those of us who have never felt emotionally bonded to an AI, this moment invites curiosity rather than judgment. These digital relationships aren’t evidence of weakness or delusion; they are a mirror held up to our most human longing — to be known, understood, and loved back. Character.AI’s announcement is, in that sense, more than a product update. It’s a cultural turning point, asking us to decide what role we want technology to play in our emotional ecosystems. Perhaps the lesson isn’t that we should stop forming attachments to AI, but that we should better understand why we do — and what that understanding can teach us about empathy itself.
The opinions expressed in this post are those of the author and some other members of ConnectSafely’s Youth Advisory Council. They do not necessarily represent the views of the Council as a whole or ConnectSafely.