Used carefully, AI can help tutors, mentors and pastoral teams ask better questions, spot emerging patterns and prepare for difficult conversations. It can give a tired tutor at the end of the day a structured set of prompts, or help a pastoral lead summarise weeks of notes into a clear picture of how a student is doing.
However, AI must never be positioned as a counsellor, therapist or crisis responder. It cannot hold duty of care, build genuine relational trust, or judge risk in a nuanced way. It can generate plausible but unsafe advice, misunderstand cultural context, and miss subtle warning signs.
The safest framing is simple: AI is a thinking aid for adults and, where appropriate, for students – a way to structure reflection and planning. Humans remain responsible for interpretation, safeguarding decisions and all direct support.
If you have not already explored where AI can help and harm more broadly, you might find it useful to read about when AI helps vs harms learning alongside this wellbeing-focused guide.
A safeguarding-first approach rests on a few non-negotiable principles.
AI acts as a prompt, not a person. Its job is to suggest questions, structures and possible interpretations, not to give definitive answers about a student’s mental health. Think of it as a planning partner helping you prepare, not a participant in the conversation.
AI acts as a mirror, not a judge. When you feed in anonymised patterns (for example, “Student A has missed three homework deadlines and seems withdrawn in class”), the AI can reflect back possible themes and questions to explore. It should not label, diagnose or predict outcomes.
Humans hold the relationship and the risk. Every AI-assisted prompt or summary must feed into a human-led conversation where you listen, respond and decide what happens next. Escalation decisions should always be made by trained staff, following existing safeguarding policies.
Finally, AI use around wellbeing should be explainable to a student and their family in plain language. If you would feel uncomfortable describing a workflow out loud, it probably needs redesigning.
Wellbeing check-ins look very different for a seven-year-old and a seventeen-year-old. AI can help you tailor language and structure, but you must set the boundaries.
For younger pupils (roughly ages 7–11), AI is best used to help adults plan. A class teacher might ask an AI tool: “Suggest five gentle, age-appropriate questions to ask a pupil who seems quiet and is avoiding group work.” The teacher then chooses, edits and uses these in person. Students at this age should not be chatting directly with AI about their feelings.
For early adolescents (around 11–14), you might introduce simple, structured self-reflection prompts, but still mediated by an adult. For example, a weekly form where students choose from AI-generated sentence starters such as “This week I felt proud when…” or “One thing that made school harder was…”. The AI can help staff summarise patterns across a tutor group, but individual responses are read by humans.
Older students (14–18+) may be ready for more direct, but still carefully constrained, AI-supported reflection. For instance, a sixth-form student might use a school-approved AI tool to help them organise their thoughts before a meeting with a mentor: “Help me list what’s been stressing me about school lately, and turn it into three points I can discuss.” Clear guidance, boundaries and monitoring remain essential.
Across all ages, AI should never be used for crisis triage, suicide risk assessment or replacing access to real counselling. Any sign of immediate risk must trigger your existing safeguarding procedures, not further AI prompts.
To make this concrete, here are examples of how staff might use AI behind the scenes.
A tutor preparing for a one-to-one might write:
“Act as a pastoral support planning assistant for a secondary school tutor. I will describe a student’s recent behaviour and context. Suggest 6–8 open, non-leading questions I can ask in a short check-in, focused on listening and understanding, not fixing. Avoid any medical or diagnostic language. Here is the context: [insert anonymised summary].”
A mentor following up on repeated lateness could ask:
“Based on this anonymised pattern – frequent lateness, incomplete homework, appearing tired – suggest three possible underlying factors I should consider, and three practical, non-judgemental ways to start a conversation about them. Keep the tone supportive and curious.”
A pastoral lead summarising notes might use:
“I will paste anonymised notes from several brief student conversations over the last month. Organise the information into: key themes, changes over time, strengths to build on, and questions to explore next time. Do not make any diagnoses or predictions. Focus on what an adult should pay attention to.”
In each case, the AI is scaffolding your thinking, not speaking to the student. You retain control over what to ask, how to phrase it and what to do with the answers.
Ready to Revolutionise Your Teaching Experience?
Discover the power of Automated Education by joining out community of educators who are reclaiming their time whilst enriching their classrooms. With our intuitive platform, you can automate administrative tasks, personalise student learning, and engage with your class like never before.
Don’t let administrative tasks overshadow your passion for teaching. Sign up today and transform your educational environment with Automated Education.
AI becomes most useful when embedded into predictable routines rather than used ad hoc. Many schools already run tutor-time check-ins, learning journals or mentoring programmes; AI can quietly support these without changing their human core.
For example, a weekly tutor-time reflection form might include three rotating questions generated with AI, tailored to the year group’s current pressures. The AI can then help the tutor quickly scan responses, summarising common themes and flagging where a human should read more closely. The tutor still reads any concerning answers and follows safeguarding procedures.
Similarly, a mentoring programme might use AI to help mentors plan sequences of conversations over a term. After each meeting, the mentor writes a short, anonymised reflection. An AI tool can suggest possible follow-up questions, highlight small positives to reinforce, and remind the mentor of patterns they might otherwise miss.
For students, AI can support metacognitive reflection rather than emotional disclosure. Linking to ideas from future-proofing students’ skills AI can’t replace, you might encourage learners to use AI to think about how they cope with stress, how they plan their week, or how they ask for help, while keeping sensitive personal details out of the chat.
Any AI use touching wellbeing must sit inside your existing safeguarding and data protection frameworks, not alongside them.
First, decide what must never be entered into an AI tool: identifiable details, specific disclosures of harm, names of alleged perpetrators, medical diagnoses and anything that could put a student at risk if leaked. Work with your safeguarding lead and data protection officer to define clear red lines.
Second, set explicit escalation rules. For example, if a check-in response suggests a student feels unsafe, is self-harming, or is being harmed, staff should stop using AI and follow the safeguarding policy immediately. AI is not consulted on whether to escalate; that decision rests with trained humans.
Third, choose tools with appropriate privacy controls. Locally hosted or education-focused platforms with clear data-processing agreements are preferable to consumer chatbots. This aligns with wider digital safety work you may already be doing, such as your approach to digital citizenship and AI.
Finally, audit logs matter. Any AI-supported workflow should leave a clear, human-readable trail: who used it, for what purpose, and what decisions were made as a result.
Trust is built through transparency. Students and families should understand, in simple language, how AI is being used around wellbeing – and how it is not.
When introducing AI-supported check-ins or planning tools, explain that:
In a parent information evening, you might show anonymised examples of AI-generated question sets and how staff adapt them. Emphasise that AI is never left alone with a child in distress, and never replaces access to school counsellors, psychologists or external services.
With students, involve them in setting boundaries. Ask older pupils what feels comfortable, what feels intrusive, and how they would like their reflections to be used. This co-design can surface concerns early and prevent misunderstandings.
To move from ideas to practice, start small and review often.
Begin with one narrow use case, such as AI-assisted tutor question planning for a single year group. Train the staff involved, agree red lines, and run a short pilot. Collect feedback from tutors and students about whether conversations feel more focused, supportive and timely.
Monitor both intended and unintended effects. Are staff noticing patterns earlier? Are students more willing to talk? Is there any sign of over-reliance on AI, or of staff feeling less confident in their own judgement?
Be prepared to stop or redesign quickly if anything feels unsafe, confusing or misaligned with your values. A safeguarding-first approach means AI use is always provisional, always under review, and always secondary to human relationships.
Above all, hold onto the core purpose: better human conversations. AI can tidy notes, suggest prompts and highlight patterns, but it is the quiet, patient, face-to-face conversations that change lives.
Happy connecting!
The Automated Education Team
Classroom displays can be powerful teaching tools, but only when they are …
This ready-to-run INSET Day workshop helps staff move from general AI …
Timetabling and daily cover are high-stakes operational puzzles, but AI is …
An AI Acceptable Use Policy written once and filed away won’t keep pace …
2025–26 brings sharper, more practical expectations for how schools manage …
A back-to-school AI plan succeeds when it is small, predictable and safe …
Clearing moves fast, but students deserve decisions that are calm, …
Results season can feel like a rush of numbers, narratives and urgent …
Summer reading can slip quietly, especially for pupils who need extra …
Tehisintellekt saab aidata töötajatel märgata heaolumustreid varem ning pidada õpilastega …
AI kan hjälpa personal att upptäcka välmåendemönster tidigare och hålla mer fokuserade, empatiska …
AI voi auttaa henkilökuntaa huomaamaan hyvinvointiin liittyviä muutoksia aiemmin ja käymään …