Online Safety
AI Companions and Kids: Questions to Ask Before Allowing Access
Understand how an AI companion differs from a homework tool. Review age rules, relationship features, privacy, spending, and a child's reasons for using it.
An AI companion is designed to simulate a personal relationship through conversation. That is a different starting purpose from a tool used for a bounded task such as explaining a calculation. Before allowing access, look at what the service actually encourages the child to do and how it responds to personal disclosures.
The boundary can be blurry. eSafety's 2026 survey of Australian children found that young people also used general AI assistants for personal and social reasons. A product category alone does not describe every interaction.
Start with the child's reason
Ask what they hope the tool will provide: imaginative role-play, company, advice, a place to vent, or something friends are discussing. Listen without ridiculing the experience.
You can acknowledge that a conversation feels meaningful while explaining that the responses come from a system. Warm language does not establish human understanding, loyalty, confidentiality, or professional competence.
Review a service before deciding
| Question | Why it changes the decision |
|---|---|
| Is the child eligible under the current age and regional rules? | A household preference does not override service restrictions |
| Does the service encourage romantic, exclusive, or secretive interaction? | The relationship design matters beyond ordinary content filters |
| What happens to conversations and uploaded information? | Personal disclosures may create privacy risks |
| Are there payments tied to attention, messages, or relationship features? | Emotional engagement and spending can become connected |
| How can a user stop, report a response, or delete an account? | The family needs a practical exit route |
If answers are unclear, that uncertainty is part of the decision. You do not have to create a child account to inspect marketing and public documentation.
Write the boundaries in ordinary language
An original family note might say: “We will not use a chatbot as the place to handle emergencies, serious health questions, or threats. We will talk to a trusted person. We will not upload other people's private messages or identifying records.”
If the family allows a suitable service, define its purpose and check-in arrangement. Do not present parental monitoring as proof that every interaction is safe.
eSafety's AI companion advisory discusses harmful content, privacy, and relationship risks. UNICEF also addresses these issues in When AI becomes a friend.
Notice the effect on ordinary life
Ask whether use is making it harder to sleep, maintain relationships, complete responsibilities, or ask real people for help. Avoid inventing a diagnostic threshold based on minutes or message counts.
If a child is distressed about leaving the chatbot, approach the feeling seriously. Shaming the attachment may make the next conversation harder. Work on a practical plan with appropriate professional support when the difficulty is persistent or substantial.
Keep help available outside the app
Identify real people the child can contact and make the route usable. If there is an immediate safety concern, seek local emergency help rather than relying on a chatbot response.
Our teen digital mental-health guide provides broader context. For an appointment about a recurring concern, the pediatrician conversation guide helps you bring specific observations instead of an app-use label.
Sources
We identify organizations and evidence types so readers can evaluate each claim. See our editorial policy.
- 2026 survey of Australian children — eSafety Commissioner. Source guidance.
- AI companion advisory — eSafety Commissioner. Source guidance.
- When AI becomes a friend — UNICEF. Source guidance.