Why staff need a safe place to ask policy questions

Staff need a safe place to ask policy questions because the safeguarding questions that matter most are the ones people are least willing to say out loud. A teaching assistant in her second term is not going to walk into the Designated Safeguarding Lead's office to ask what counts as a concern. She sat through the induction. Asking now would reveal that she did not follow it, or that she has forgotten, and she is on a one-year contract in a country that is not hers.

So the question goes sideways. She asks the colleague at the next desk, who is also not sure and answers confidently anyway. Both of them now hold the same wrong understanding, and the school has no idea. I have watched this happen in staffrooms in more than one school, and the striking part is how reasonable everybody's behaviour is at each step.

A school can measure training completion easily. It cannot measure the questions nobody asked. That absence is invisible in every report a Head reads, and it is the part of the system where things actually go wrong.

Amy Edmondson's work on psychological safety describes the mechanism precisely. She defines team psychological safety as a shared belief that the team is safe for interpersonal risk taking, and her 1999 study in Administrative Science Quarterly links it to whether members of a team ask questions, seek feedback and discuss errors. Where that belief is absent, people stay quiet and the team learns less. A staffroom is a work team by that definition, and a safeguarding question is exactly the kind of interpersonal risk she describes.

Schools add their own multipliers to that cost. Hierarchy is one: in many of the systems international schools recruit from, a junior member of staff questioning a procedure reads as questioning a senior colleague. Language is another: a driver or a kitchen supervisor who works in the local language will not ask a question in English in front of an audience. Contract security is a third, and it weighs heavily on staff whose visa depends on the school.

Time is the least discussed and the most common. The DSL is teaching, in a meeting, or with a child. A member of staff with a small uncertainty and a five-minute window does not start something that might take twenty minutes. She defers it, and small uncertainties that are deferred usually stop being uncertainties. They become assumptions.

Schools usually respond to this with encouragement. The Head says at the start of term that there are no stupid questions in safeguarding, and means it. That helps a little and it does not change the arithmetic for the teaching assistant, because the cost she is weighing is social and immediate while the reassurance is general and three weeks old.

What changes the arithmetic is a route where asking costs nothing observable. No colleague sees the question. No meeting is booked. Nobody learns that a member of staff did not know something. The answer comes from the school's own policy rather than from a colleague's recollection, and it arrives in under a minute, in the language the person actually works in.

An AI policy assistant is one way to build that route, and the honest description of it is narrow. It answers questions from the school's approved documents. It does not hold a concern, assess a threshold, or advise about a specific child. Getting that boundary right matters more than the answer quality, because a tool that is vague about its limits is worse than no tool. The related problem of staff who hesitate to report because of hierarchy or culture needs the same care.

The refusals define the tool. An assistant that will answer anything is a liability in safeguarding, because the questions that sound like policy questions are sometimes disclosures wearing a policy question's clothes. Someone typing about a specific child needs a person, immediately, and the system's only correct response is to say so and name the Designated Safeguarding Lead and the route to reach them.

The second refusal is about thresholds. Whether a particular set of facts crosses the line into a referral is a professional judgement that belongs to the DSL, with the Head as the risk owner. An assistant should say what the policy states about thresholds and stop there.

The third refusal is about gaps. When the school's policy does not answer a question, the correct output is that the policy does not cover it and here is who to ask. A model that fills the gap with a plausible general procedure teaches a member of staff something the school never approved, which is precisely the failure that generic training already causes.

• Anything naming a child: route the member of staff to the DSL, do not record details in the chat.
• Threshold decisions: quote the policy, name the decision-maker, stop.
• Questions the policy does not cover: say so plainly and point to a person.
• Anything urgent: give the reporting route immediately, before any explanation.

A teacher who is unsure about a procedure has several routes to an answer. She has a head of department, a year leader, a mentor and a staffroom of colleagues who work in the same language and hold the same training. The safety net around her is thick, and she will usually find the right answer within a day.

The driver waiting at the gate has none of that. He works alone for most of the day, his shift ends before the staff briefing, he received a shorter induction than the teaching staff, and he may not use English at work. The same is true of the night security guard, the kitchen supervisor and the cleaner who works the corridors after five. These adults see children in the least supervised parts of the school day, and they are the least likely to ask anyone anything.

Any route a school builds for questions has to work for that group first, which usually means the local language, a phone rather than a staff laptop, and no requirement to find a colleague. If it works for them it will work for everyone else, and the same reasoning applies to safeguarding training for drivers, security, catering and nursing staff.

The second reason to build this route is that it produces information a school cannot get any other way. When forty questions in a term concern the same paragraph of the child protection policy, the school has learned something specific. That paragraph is unclear to the people who have to act on it.

That finding is far more useful than a training completion figure. It tells the DSL exactly which sentence to rewrite, and it tells the Head where the next short training module should be aimed. When I ran a policy audit against a well-established school with two accreditation cycles behind it, the pattern of confusion pointed at the same handful of clauses that the audit flagged independently, which was a good check on both.

Question data has to be handled carefully. Aggregate patterns are useful to the DSL. Individual questions attached to individual names would rebuild the exact cost the route was designed to remove, which is why the reporting should show the topic and the frequency rather than the person. A school that gets this wrong will see the volume of questions fall to zero within a month.

Every safeguarding decision in a school belongs to a trained human being. Keeping children safe in education is explicit that staff should raise concerns with the Designated Safeguarding Lead, and no system that answers policy questions changes that duty. The value of the assistant is that it removes the smaller uncertainties before they turn into silence, so that the questions reaching the DSL are the ones that need her.

A DSL I worked with put it well. She said the problem was never the staff who came to her. It was the ones who nearly came to her and then decided it was probably nothing. Lowering the cost of the first question is how a school hears from those people, and it is a large part of what separates a safeguarding culture from a compliance record.

Staff can ask the SafeguardIS policy assistant questions that are answered from the school's own policies, in English, Russian or Kazakh. The DSL keeps every safeguarding decision, and no child's details belong in a chat window: anything that looks like a concern about a real child routes the member of staff to a person. If your staff work in a language the platform does not ship yet, I build that language in for your school as part of the pilot.

The assistant sits next to the training rather than replacing it. Training is built from the school's own safeguarding and child protection policies, the Designated Safeguarding Lead reviews and approves every module before staff see it, and publishing requires a named reviewer's declaration with the content hashed and logged. There is also a policy audit that checks a school's documents against ITFCP-aligned expectations and produces an action plan, which is where most schools find the clauses their staff keep asking about. The role of the Designated Safeguarding Lead in an international school stays exactly where it was.

I have worked in international education since 2008, as an IB and MYP teacher, an IB DP economics teacher, an IB and CIS evaluator and workshop leader, and most recently as CIS accreditation coordinator. I am not a Designated Safeguarding Lead. To see what your staff would get back when they ask a question from your own policy, book a 20-minute walkthrough.