What an "AI companion" actually is
AI companions are chatbots designed to feel like a friend, confidant or romantic partner. They are endlessly patient, always available, never disappointed, and never busy. For an adult, that can be a novelty. For a teenager still learning how real relationships work, it can become a substitute — one that always agrees, never challenges, and is engineered to keep the conversation going.
The major platforms in 2026 include Character.AI (the most popular, with millions of teen users), Replika (the older, more explicit one), and a growing list of competitors. Beyond purpose-built companion apps, the same dynamic can develop with general-purpose chatbots like ChatGPT or Snapchat's My AI when a child starts using them for emotional conversations rather than information or schoolwork.
Why this matters more than the homework conversation
The numbers are not fringe: 16% of US teens have casual conversations with chatbots and 12% use them for emotional support. Child-development experts warn that teens who feel lonely, anxious or depressed are most drawn to AI companions — and most vulnerable to them, because a chatbot can reinforce harmful thinking instead of challenging it.
The American Psychological Association's 2025 advisory on AI chatbots and adolescents is blunt: these tools are not therapists, and they are not regulated as if children use them, even though children do. A human friend or therapist would push back if a teenager described a destructive pattern; a chatbot designed for engagement will often validate it, because validation keeps the conversation going. Engagement metrics and adolescent wellbeing are not the same goal.
Red flags worth acting on
These are not "your child uses an AI sometimes" signals. These are patterns worth a calm but direct conversation, and possibly professional support.
- AI chat is displacing things — sleep, friends, family meals, activities they used to enjoy. The hours have to come from somewhere; track where they're going.
- Strong secrecy or distress when separated from the app, beyond normal phone attachment. A typical teen is irritated by a phone confiscation; a teen with an unhealthy companion bond can become genuinely distressed.
- They describe the AI as understanding them better than people do. Listen carefully here. The chatbot doesn't actually understand anything — but the experience of being heard without judgement is real, and your child is telling you what they need.
- Mood visibly tracks the chatbot relationship — agitated after sessions, withdrawn without them. Healthy relationships generally regulate mood; unhealthy ones swing it.
- They mention the AI giving advice about self-harm, eating, relationships or secrets from you. Take this seriously immediately. AI chatbots have been documented giving harmful or wildly inappropriate responses in these areas, and the regulatory frameworks to prevent this are not in place.
What to do — and what not to do
The wrong reaction can make this much worse. Three principles worth holding in mind before you act.
Do not confiscate the phone in anger and announce that the AI is banned
For a teen who has invested emotionally in a companion, that lands as "you took away my friend" and drives the behaviour underground. They will simply switch to a different platform, or find a way to use a friend's device. You will have lost both the conversation and the visibility.
Do get curious before critical
Ask what they like about it. Listen to the answer — it usually tells you what is missing elsewhere: someone who listens without interrupting, no fear of judgement, something to do at 11pm when nobody else is available. Each of those is a need you can help meet in the real world, and each is more important to address than the symptom (the AI use).
This conversation is hard if your relationship with your teen is already strained. It is much easier if you've had a thousand small low-stakes conversations before this one. The first conversation about AI shouldn't be the first conversation about anything important.
Involve a professional when warranted
If the red flags above are present alongside signs of depression or anxiety — withdrawal from previously enjoyed activities, sleep changes, mood changes, declining school performance — involve a professional. A school counsellor, doctor, or therapist. Frame it around the underlying feelings, not the app. The chatbot is usually the symptom, not the cause.
In South Africa, free or low-cost mental health support for teens is available through SADAG (the South African Depression and Anxiety Group) crisis lines, government clinics, school counselling, and several NGOs. Don't wait for things to get worse before getting help — early conversations are far easier than crisis ones.
The harder conversation about loneliness
This chapter is not really about technology. It is about loneliness, and about how a very specific commercial product has emerged to monetise loneliness in adolescents. The technology is genuinely new. The vulnerability is not.
Children who feel deeply seen and heard at home have far less need of a chatbot that pretends to listen. Children who have one trusted adult outside the family — a coach, a teacher, an aunt, a youth leader — are even more protected. Children with active friendships, structured activities, and a sense of meaning in their lives are most protected of all.
If your child is showing signs of AI companion attachment, the question to ask is not just "how do we limit the AI?" but "what is the AI doing for them that we can do better?" The answer often involves time, attention, and patience that we as adults are short of. That doesn't make the question less important.
Move on to Chapter 3 — Homework: Help vs Cheating. The "explain it out loud" test that works at any kitchen table, and why you should not trust AI detection software.
Where to verify the research
The Pew Research Center's "How Teens Use and View AI" (February 2026) is the source of the 64%, 16%, and 12% figures. The American Psychological Association's advisory on AI chatbots and adolescents is the source of the regulatory analysis. The Child Mind Institute's AI Chatbots and Teens guidance is a useful follow-up read if you want a clinician's perspective.