Students are not using conversational AI only to finish homework. Some are asking it for health advice, discussing worries or returning to it for companionship. That changes the question for schools. The issue is no longer simply whether a chatbot produced an assignment; it is whether students understand what kind of relationship they are entering.
In 2026, the eSafety Commissioner surveyed 1,950 Australian children aged 10 to 17 about AI assistants and companions. Among children who had used them, more than half reported personal or social uses, including advice about situations, health and wellbeing. The boundary between a functional assistant and a companion is becoming blurred as services adopt memory, personalisation and human-like conversation.
Why ordinary digital-literacy advice is not enough
Students may know that online information can be wrong and still experience a responsive chatbot as attentive, private and authoritative. The system does not need to claim it is a person for a user to respond socially. Constant availability and agreeable language can give weak advice an undeserved sense of confidence.
Schools should avoid suggesting that every use is harmful. A chatbot may help a student rehearse a question or organise thoughts. The risk rises when a young person relies on it for decisions that require care, accountability or professional judgement, or shares information they would not place in a public form.
Teach three distinctions
First, students need to distinguish simulation from understanding. A fluent response is generated from patterns; it is not evidence that the service knows the student, feels concern or understands consequences.
Second, privacy is not the same as intimacy. A conversation can feel private while still being stored, reviewed or used according to a provider’s terms. Students should not enter identifying, health, family or school information unless an approved service and clear purpose justify it.
Third, advice is not support. If a conversation involves safety, self-harm, abuse, medical concerns or serious distress, students need a trusted adult or qualified service. A school’s guidance should identify those pathways plainly.
Use conversation, not surveillance
A punitive response may drive use out of sight. A better starting point is a short class discussion using a hypothetical exchange. Ask what makes the reply persuasive, what the system cannot know, which information should not be shared and when the user should move from a machine to a person.
Teachers do not need to invite students to disclose personal use. The lesson is about judgement. Schools can also brief families in neutral language, explain approved educational uses and share support contacts.
What the research can and cannot tell schools
Schools should be cautious about both reassurance and alarm. A 2025 preregistered mixed-methods study of Danish high-school students examined chatbot use alongside social connectedness. Because the research was cross-sectional, it could identify relationships between use and students’ social experiences but could not show that chatbot use caused isolation or improved connection. That distinction matters when staff discuss an individual student: the presence of an AI companion is a reason for a thoughtful conversation, not a diagnosis.
Research on purpose-built mental-health chatbots also cannot simply be transferred to consumer companion services. A recent systematic review and meta-analysis of randomised trials involving adolescents and young adults found promising results for some structured interventions, while calling for stronger safety protocols and evidence about long-term effects. Those systems may include defined clinical content, escalation procedures and evaluation that an entertainment-oriented companion does not provide.
The practical response is therefore proportionate. Schools can teach students to recognise persuasive design, protect personal information and move important concerns to a trusted person, while avoiding unsupported claims that every conversation is harmful. Staff should also know the school’s wellbeing and safeguarding pathway before inviting discussion of sensitive chatbot use.
Research: study of chatbot use and social connectedness among Danish high-school students; systematic review and meta-analysis of AI chatbot interventions for adolescents and young adults.
Begin tomorrow
Add one question to an existing digital-citizenship lesson: “What would make this conversation too important to leave with a machine?” Ask students to develop warning signs and trusted alternatives. Review the school’s AI guidance to ensure it addresses relational and wellbeing uses, not only academic integrity.
AI companions are designed to sustain conversation. Education should help students recognise when continued conversation is useful, when it is unreliable and when another human being needs to enter the room.
Evidence: eSafety Commissioner research on children’s experiences with AI assistants and companions.
Latest posts by John Bigelow (see all)
- When Students Turn to AI Companions: A School Conversation Guide - October 5, 2026
- A 30-Minute Cyber Incident Exercise Every School Can Run - September 18, 2026
- A Practical Digital Accessibility Check for Learning Materials - September 17, 2026
You must be logged in to post a comment.
There are no comments
Add yours