How to Tell if an AI Tutor Is Safe and Accurate

You can tell an AI tutor is safe and accurate by checking five things before your child uses it: where its answers come from, what it does when it doesn't know something, how it handles a child's data, whether a real, verified human stands behind it, and what happens when a question needs more than a chatbot can give. A tool that passes all five is worth using. One that fails even one — especially the last two — is a risk dressed up as convenience.

That checklist matters because "AI tutor" now covers everything from a bare chatbot wrapped in a study app to a curriculum-grounded assistant sitting inside a marketplace of verified human tutors. They look similar on a screen. They are not similar in what happens when something goes wrong.

"Safe" and "accurate" are two different questions

Parents often ask about safety and accuracy as if they're the same check, but they fail in different ways. An AI tutor can be perfectly safe — no inappropriate content, no data mishandling — and still be wrong about the syllabus, the mark scheme, or the method a particular exam board expects. Equally, a tool can be broadly accurate on the facts and still be unsafe in how it handles a child's data, a distressing question, or a moment where a human should be involved instead.

That's why a single "is it safe?" question isn't enough. You need to check both: is the information right, and is the way it's delivered appropriate for a child. The five checks below cover each in turn.

1. Where do its answers actually come from?

A general-purpose chatbot answers from whatever its underlying language model happens to hold — a vast, unfiltered mix of the internet, with no guarantee any of it maps to the exam board your child is actually sitting. It will answer confidently either way. Confidence is not the same as correctness.

A safe AI tutor should be able to show its working: does it draw from a curated, curriculum-aligned knowledge base, or is it free-associating from general training data? Ask it where an answer came from. A tool built for education should be able to tell you; a bare chatbot usually can't, because it doesn't know either.

2. What does it do when it doesn't know?

This is the single biggest tell. A poorly built AI tutor will guess and present the guess as fact — the behaviour researchers call hallucination. A well-built one is designed to flag uncertainty, say "I'm not sure" rather than invent an answer, and track what a learner has actually mastered rather than assuming.

For a parent, this is easy to test directly: ask it something genuinely obscure or slightly wrong, and see whether it corrects you or plays along. A tool that never says "I don't know" is telling you something important about how it was built.

3. What happens to your child's data?

Under UK data protection law's Children's Code, a service likely to be used by under-18s has to meet minimum standards — acting in the child's best interests, limiting profiling, and being clear about what data it collects and why. That's the legal floor. A safe AI tutor should be able to state plainly what it stores, why, and who can see it, without you having to dig through a generic privacy policy written for adults.

The Department for Education's guidance on generative AI in education sets out a similar expectation for schools and edtech tools: protect pupil data, keep a human accountable for decisions that affect a child's learning, and treat AI as a support tool, not an unsupervised decision-maker. A tool that can't answer a straightforward "who sees what my child types" question hasn't cleared that bar.

4. Does it know when to hand over to a human?

An AI tutor that never escalates is a tool with no ceiling. Real learning eventually hits something a curriculum-grounded chatbot can't safely handle on its own — a child in genuine distress, a safeguarding concern, a question that needs a qualified adult's judgement, not a generated paragraph. The design question is whether the tool is built to recognise that boundary and step back, or whether it will keep generating text regardless of what's actually being asked.

This is also where "AI tutor" and "AI tutoring platform" split. A standalone chatbot has nowhere to hand a child off to. A platform that pairs AI support with a marketplace of real, checkable tutors does — which is the actual point of building one alongside the other rather than instead of it.

5. Is there a real, verified human standing behind it?

This is the part most AI-safety checklists miss, because they treat the AI in isolation. But the honest answer to "is this AI tutor safe" is rarely just about the AI — it's about what backs it up. A chatbot with no accountable human behind it is asking you to trust software with no consequences for getting it wrong. A tutoring platform where the AI sits alongside verified human tutors is asking you to trust a person whose credibility you can actually check.

On Tutorwise, that check is called Credibility as a Service, or CaaS. Every human tutor on the platform is scored across six categories built from real, verifiable signals — not a self-written bio and not a star rating anyone can buy with enough reviews. The categories cover delivery (how sessions actually go), credentials (real, checked qualifications), network (professional standing), trust (identity verification and an enhanced DBS check), digital presence, and demonstrated impact. No tutor gets a score at all until they've verified their identity or completed onboarding — there's a hard floor before any number appears. You can read the full mechanism in how CaaS works and what a credibility score actually measures.

That distinction — a computed, checkable score versus an unverifiable rating — is also why reviews alone aren't enough to trust a tutor. A five-star average tells you sessions felt fine. It doesn't tell you whether the person is who they say they are, or whether they've been through an enhanced DBS check, which is a specific, legally defined process — not a box someone ticked on a form. Tutors on Tutorwise also carry safeguarding duties set out in law, not just platform policy.

Sage: an AI tutor built this way, not bolted onto one

Tutorwise's own AI tutor, Sage, is built on this principle rather than around it. Sage answers from a curriculum-focused knowledge base, is designed to flag uncertainty instead of guessing, and tracks what a learner has genuinely mastered over time. When a question needs more than Sage can safely give — real accountability, a qualified human judgement call, or simply a level of teaching a chatbot shouldn't attempt — it points towards a verified tutor on the platform, whose CaaS score is visible before you ever book a session.

Here's what that looks like in practice. A parent checking whether Sage is safe for their Year 10 child to use for GCSE maths revision can do exactly what this checklist recommends: ask Sage a curriculum-specific question and see whether the explanation matches the actual syllabus; ask it something it's likely to be unsure about and watch whether it says so; check what data it holds on the account; and, if the child needs more than a chatbot can responsibly give, see who Sage hands them to and whether that person's credibility is checkable rather than asserted. That last step is the one a standalone AI app can't offer, because there's no verified human on the other end of it.

This is also why AI genuinely helping a child learn and a platform overclaiming what its AI can do are two different things. Sage is useful precisely because it's honest about its limits and backed by people whose standing you can verify — not because it claims to replace a qualified human tutor. It doesn't, and it isn't designed to.

Red flags — when the checklist fails

A few signs an AI tutor hasn't been built with these questions in mind: it answers every question with total confidence regardless of subject or difficulty; it can't say where its information comes from; its privacy policy is generic and doesn't mention children specifically; there's no path from the AI to a real, checkable human when a question needs one; and any "verification" or "trust" claims are marketing copy rather than something you can independently confirm, like a DBS check or an identity check. If a tool fails two or more of these, treat it as a study aid to supervise closely, not something to leave a child alone with.

Overclaiming is itself a warning sign worth naming on its own. Any AI tool that markets itself as a full replacement for a qualified human — "never need a tutor again," implying it can handle safeguarding, or claiming a level of accuracy no language model can honestly promise — is telling you it's optimising for the sale, not for your child. The safest tools are usually the ones that are upfront about what they can't do, because that honesty is itself evidence the rest of the design was done carefully.

Why this matters more than it used to

The number of AI study tools and chatbots marketed at parents and students has grown quickly, and the underlying technology genuinely has improved — modern language models are far better at giving grounded, curriculum-relevant explanations than earlier versions. That's a real gain. But it also means confident-sounding answers are now the default output of almost any AI product, whether or not the product was built with a child's specific needs, data, or safeguarding in mind. The confidence of the output has stopped being a useful signal of quality. What's left is exactly the checklist above: provenance, honesty about uncertainty, data handling, escalation, and a verifiable human behind the claim.

FAQ

Is Sage free to use? Yes. Sage is Tutorwise's AI tutor, and you can start using it without paying, working alongside a verified human tutor when you want or need one.

How is Sage different from ChatGPT or a general AI chatbot? A general chatbot answers from broad training data with no accountability behind it. Sage answers from a curriculum-focused knowledge base, is designed to flag uncertainty rather than guess, and sits inside a marketplace of tutors whose credibility is a checkable, computed score rather than a claim. For revision using general-purpose tools, see how to use ChatGPT for GCSE revision safely.

Can an AI tutor replace a human tutor? No — and a well-designed one won't claim to. AI is strong at instant, patient, curriculum-aligned explanation; it can't replace a qualified human's judgement, accountability, or the safeguarding a real relationship requires. The full comparison is here.

What should I actually check before trusting an AI tutor with my child? Where its answers come from, whether it admits uncertainty, what it does with your child's data, whether it can hand off to a real human when needed, and whether that human's credibility is independently verifiable rather than self-declared.

Is AI tutoring safe for a child with additional needs? It depends heavily on the platform and the specific need — a curriculum-grounded AI tutor backed by verified human specialists is a different proposition to a generic chatbot. This covers AI tutoring for SEN students specifically.

---

Traderwise, Trainerwise and Adspots are separate Tutorwise verticals; this article covers tutoring only.

More in this series:

• Can an AI Tutor Replace a Human Tutor? An Honest Answer
• How to Use ChatGPT for GCSE Revision Safely and Well
• AI Tutoring for SEN Students: Useful Support or Risk?