A high-profile launch in AI health will always attract attention in schools. Senior leaders may wonder whether it could support pupil wellbeing, PSHE discussions or early signposting. But when a new service is explicitly unavailable in the UK, EEA and Switzerland, the first task is not enthusiasm. It is verification. Schools need a measured response that protects pupils, avoids accidental procurement drift and keeps safeguarding decisions firmly grounded in human judgement. If your team is already reviewing AI governance this term, it helps to align this conversation with broader policy work, such as a January AI policy sprint.
On 7 January 2026, ChatGPT Health launched as a health-focused AI offering, but not for users in the UK, EEA or Switzerland. That exclusion matters immediately for schools because many initial reports, social posts and vendor round-ups may mention new capabilities without making regional restrictions prominent. In practice, that means a school may hear about a feature before checking whether it can lawfully, contractually or technically access it.
For school leaders, the key point is simple. A launch announcement is not the same as service availability. Nor is a polished demo proof that your school can use the tool. Regional exclusions can apply at sign-up, contract, processing location, feature level or support level. A product may appear visible online while remaining unavailable for institutional use in your jurisdiction.
For schools in the UK, EEA and Switzerland, this is not just a commercial inconvenience. It affects planning, safeguarding and compliance. If a wellbeing lead assumes a tool is available and starts discussing pilots, staff time can be wasted and expectations can rise unnecessarily. Worse, a colleague might test the service using personal details, pupil scenarios or health-adjacent prompts before the school has checked the legal and safeguarding position.
This is especially sensitive where pupil support is involved. Health-related AI tools sit close to special category data, vulnerable pupils and safeguarding thresholds. Even if a service were available, schools would still need to ask whether it should be used at all in that context. Teams already thinking about AI and wellbeing may find it useful to revisit safeguarding boundaries in articles such as AI wellbeing copilots in schools, because the central issue is not novelty but appropriateness.
When a launch is widely covered, procurement assumptions can spread quickly. A governor reads a headline. A pastoral lead sees a conference clip. A supplier mentions “coming soon” in a broader platform conversation. None of that is enough.
A sensible school process starts with written verification. Ask the vendor to confirm, in writing, whether the exact product and feature set are available to schools in your geography, under what terms, and with what age restrictions. Ask whether availability differs between consumer and education accounts. Ask whether the service can be enabled by default through a larger platform contract, or whether it is blocked regionally even where the parent platform is already in use.
This matters because demonstrations often show global roadmaps rather than your actual tenant settings. The same caution applies across AI procurement more broadly, especially where features appear before governance catches up. If your team needs a wider structure for checking claims against reality, a procurement and governance playbook can help frame those questions.
It is easy to see why schools might be interested. PSHE teams may imagine scenario-based conversations about sleep, stress or healthy habits. Wellbeing leads may think about signposting to support, helping pupils phrase concerns, or offering out-of-hours guidance. Safeguarding teams may wonder whether AI could help triage low-level worries before they escalate.
These are understandable ideas, but they are also exactly where boundaries matter most. A pupil asking about anxiety, self-harm, eating, medication, family conflict or abuse is not engaging in a neutral edtech interaction. They may be disclosing distress, risk or harm. Even apparently simple prompts can become highly sensitive within seconds. That is why schools should treat health-adjacent AI tools differently from lesson-planning or report-writing tools.
Before any pilot or purchase, school leaders should ask a short set of hard questions. Is the product actually available in our region? Is it offered to schools or only to consumers? What minimum age applies? What data are processed, stored and used for model improvement? Is there a separate education agreement? Can the school disable the feature? What human escalation path exists if a pupil expresses risk?
The quality of answers matters as much as the answers themselves. Vague wording such as “enterprise-grade privacy” or “designed with safety in mind” is not enough. Ask for documentation. Ask for the data processing position. Ask who the contracting entity is. Ask where logs are retained. Ask whether conversations can be exported, deleted or centrally governed. A good cross-check is to use the same discipline you would bring to an AI privacy audit checklist, even if the interest began as a quick exploratory conversation.
Health-adjacent AI tools raise immediate data protection concerns because pupils may share information about physical health, mental health, family circumstances, identity, diagnosis or treatment. In many settings, that can quickly drift into special category data or other highly sensitive information. If staff are not careful, even “testing” a tool can involve real personal data.
Red flags include unclear retention periods, uncertain model training use, no straightforward deletion route, weak admin controls, and no clear separation between consumer and institutional accounts. Another warning sign is a product that appears to support emotionally significant conversations without a clear lawful basis analysis or a robust explanation of how schools should avoid entering identifiable pupil information.
For many schools, the safest starting point is a strict rule: no real pupil data in any unapproved health or wellbeing AI tool. Use synthetic examples only during evaluation. If your school is refreshing its wider AI rules this term, build that line into your acceptable use policy refresh.
There are also things AI should not do in pupil wellbeing workflows. It should not replace a trusted adult. It should not assess suicide risk in isolation. It should not decide whether a concern is “serious enough” to escalate. It should not reassure a pupil that a harmful situation is normal. It should not become a hidden pastoral layer that staff cannot see, review or supervise.
In school settings, AI may sometimes support information access, signposting or staff preparation. But the moment a pupil’s safety, health or vulnerability is in question, the workflow must return to human-led systems. DSLs and pastoral teams need clear boundaries on what staff may ask an AI tool to do, what pupils may be directed towards, and when immediate referral procedures apply. This is particularly important in digital harm contexts, where wellbeing and safeguarding overlap, as explored in work on incident response and cyberbullying scenarios.
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.
If the service is unavailable this term, schools still need support pathways that work now. A no-regrets contingency plan starts with human capacity. Reconfirm pastoral referral routes. Remind staff how pupils can seek help confidentially. Check that PSHE resources on stress, sleep, healthy routines and help-seeking are current. Make sure external support services used by the school are up to date and easy to access.
Then review approved alternatives already within your estate. In some schools, this may mean using existing platform controls, approved chat tools for staff-only drafting, or curated wellbeing resources rather than conversational AI for pupils. The point is not to replicate the unavailable product. It is to meet the underlying need safely. If your wider AI environment is still developing, a minimum viable AI toolkit can help schools focus on governed, low-risk use first.
A one-page briefing for SLT, DSLs, DPOs and governors should be short and factual. State what launched, the date, and the regional exclusion. Note that schools in the UK, EEA and Switzerland should not assume access, pilot rights or procurement readiness. Summarise the main risks under four headings: availability, data protection, safeguarding and reputation. Then set out the school’s temporary position for the term, including whether staff may explore the tool at all, whether synthetic-only testing is permitted, and which existing support routes should be prioritised instead.
This style of briefing works best when it avoids hype. Governors do not need a product tour. They need assurance that the school is not sleepwalking into a health-adjacent AI decision without proper scrutiny.
For January 2026, keep the checklist practical. Verify regional availability in writing. Pause any pilot planning until legal and safeguarding checks are complete. Ban real pupil data from exploratory testing. Confirm whether existing contracts or platform settings expose staff or pupils to adjacent features. Re-state safeguarding escalation routes. Brief governors and senior leaders in one page. Review policy wording so staff know what is and is not approved.
Most importantly, keep the educational aim in view. Schools are not falling behind because they did not adopt a headline launch on day one. In sensitive areas like pupil wellbeing, restraint is often a sign of strong leadership. The right question is not “How quickly can we use this?” but “How safely can we meet the need in front of us?”
May your decisions this term be careful, clear and firmly centred on pupils.
The Automated Education Team
Microsoft Build 2026 brought a fresh wave of Copilot announcements, but …
Summer term is often the last realistic window for school leaders and AI …
Schools do not need a whole-school AI rollout to learn what works. In many …
Once the papers are marked, many departments want feedback that is sharper …
Primary assessment week can feel intense, especially when SATs and spring …
Exam week often creates pressure to do more, faster, with less time to …
Many school AI rules still rely on vague disclosures such as “I only used …
AI voice tools for modern foreign languages have improved sharply by 2026, …
Spring assessment season puts pressure on teachers, pupils and families to …
Suur AI tervisevaldkonna lansseerimine võib kiiresti äratada huvi koolijuhtides, PSHE eestvedajates …
En större lansering inom AI-hälsa kan snabbt väcka intresse hos skolledare, PSHE-ansvariga och …
Merkittävä AI-terveyspalvelun julkaisu voi nopeasti herättää kiinnostusta koulun johdossa, …