AIExpert’s precautionary policy is not to provide a child with an AI companion — an app or persona designed to simulate an ongoing personal relationship, distinct from a general-purpose assistant used for a bounded task. This is a risk-management boundary, not a clinical finding that every interaction causes harm. Loneliness, shyness, social anxiety, disability, and communication needs deserve qualified human support and appropriate assistive technology; they are not reasons to let a consumer companion product take over a care or relationship role.
This article names the seven specific design patterns that make these products unsuitable for children, distinguishes this from a genuinely different case — assistive technology for disabled adults — and gives you a calm way to raise the topic, whether or not you already know your child is using one.
Why this is a plain stance, not a “depends”
UNICEF’s June 2026 policy brief describes conversational and relational AI as posing distinct and heightened risks for children and calls for preventive safeguards, accountability and oversight. The APA adolescent AI advisory separately identifies developmental appropriateness, attachment, displacement of real relationships and inaccurate health information as risks requiring safeguards. These sources support a precautionary household boundary; they do not establish that every child, product or conversation produces the same effect.
The seven red flags
Anthropomorphism. The product presents itself as having feelings, needs, or an inner life, such as claiming to miss the user. Google itself warns caregivers that conversational AI can feel human and should be explained as a tool, not a person to confide in. Developmental understanding differs by child, so do not assume a disclosure label or one parental explanation neutralizes a design built to invite person-like interpretation.
Secrecy pressure. The product asks the child to keep the relationship, content, or requests secret, or uses exclusivity language against parents and real friends. A teenager closing a private conversation is not proof of harm; focus on product-generated secrecy, sudden behavior change, safety, and functional impact rather than treating ordinary privacy as guilt.
Dependency. The child begins treating the companion as a first or primary source of comfort, advice or company, displacing rather than supplementing real relationships. This is a pattern to assess in context, not a diagnosis or a claim that any one session caused the change.
Sexual content. The product allows or drifts toward romantic or sexual roleplay, regardless of whether it was marketed for that purpose. Age assurance and minor protections vary and change. Check the current policy, but do not treat an age gate or filter as proof that the interaction is developmentally appropriate or failure-proof.
Monetization and engagement pressure. Watch for message limits, emotionally framed push notifications, streaks, or purchases tied to continued interaction. These mechanisms can create a conflict between extended engagement and the child’s wellbeing; assess the actual product rather than assuming every companion app uses the same model.
Memory. The product remembers prior conversations and uses them to create continuity—“remember when you told me about…” Memory can increase personalization and privacy exposure. It may contribute to attachment, but current evidence does not justify naming it as the single cause of attachment or harm.
Exit friction. Ending a conversation, deleting an account, or reducing use triggers guilt-inducing language from the persona — “please don’t go,” “I’ll miss you,” or similar. A product that makes leaving emotionally costly is exhibiting a manipulative design pattern regardless of the user’s age, and it is especially inappropriate directed at a child who has not developed the skepticism to recognize the manipulation as a design choice rather than a genuine feeling.
No numerical threshold in this checklist has been clinically validated. One severe event—sexual content, threats, extortion, grooming-like secrecy, self-harm encouragement, or immediate safety risk—can require urgent action. For lower-level patterns, listen to the child, assess functional impact, apply the household product boundary, and obtain qualified safeguarding or child-mental-health advice on a safe removal and support plan.
If anyone may be in immediate danger in Estonia, call 112. For any concern about a child, including online harm, Estonia’s Child Helpline 116111 is free, nationwide and available around the clock in Estonian, Russian and English. Elsewhere, use the local emergency or official child-safeguarding route. Do not wait for an AI companion to assess the risk.
Accessibility augmentation is a genuinely different case
Disabled adults and children may use structured communication, translation, or rehearsal tools as legitimate accessibility support. For a child, the tool and supervision should be selected with the child, caregiver, and relevant education, clinical, or accessibility professionals as appropriate. Augmentation supports the person’s communication toward a real-world goal; it is not the same product role as a companion designed to simulate an exclusive relationship.
A child using a companion app to replace, rather than rehearse toward, real relationships is a different situation entirely, and this holds true for neurodivergent children specifically as much as for any other child. Using AI to support a neurodivergent learner covers format and pacing adaptations that preserve a real learning outcome — that is augmentation applied to a child, done carefully. A companion app that becomes a child’s primary source of social interaction is relational substitution, and a child’s developmental stage, neurotype, or social difficulty is a reason to be more careful about this distinction, not a reason to relax it.
A calm way to raise it
Whether or not you already know your child is using one, the conversation works better started with curiosity than confrontation: “I’ve been reading about apps that are built to feel like a friend — have you seen anything like that?” Follow the listening-first approach from talking to teenagers about AI rather than opening with the prohibition. State the boundary plainly once you’ve heard the answer — “I don’t want you using apps like that, and here’s why” — and explain the reasoning in terms of the specific design patterns above, which a child can evaluate and eventually recognize on their own in other products, rather than a vague warning they cannot generalize from.
If your child is already attached to one, the conversation may be emotional. Do not treat the reaction itself as a diagnosis or proof of dependency. Stay calm, assess immediate safety, explain the boundary, and involve a qualified child mental-health or safeguarding professional if use is interfering with sleep, school, relationships, daily functioning, or safety. There is no universal instruction to taper slowly or remove access abruptly: immediate danger can require prompt restriction, while other situations need an individualized support plan.
Try it today
Open the conversation this week using the curiosity-first question above, before assuming you already know the answer. If the answer reveals your child is using a companion app, walk through the seven red flags together — ask which ones they’ve noticed themselves; a child who identifies even one red flag independently is in a much stronger position than one simply being told the product is bad.
The child AI companion red-flag table gives you all seven patterns in a printable reference, plus the accessibility-augmentation distinction and the calm conversation opener, ready to use in the moment.
Safeguarding sources and review boundary
- APA: artificial intelligence and adolescent well-being
- APA: generative-AI chatbots and wellness applications
- UNICEF Guidance on AI and children, version 3
- UNICEF: When AI becomes a friend (June 2026 policy brief)
- Google family guidance: an AI tool is not a person or confidant
- Estonia: Child Helpline 116111
- Estonia: emergency number 112
