Usually yes, with one condition: the school can tell you who answers when the tool flags something serious, and how fast. The chatbots schools buy are built differently from the apps teenagers find on their own, and the better ones get a trained adult on the phone within minutes. None of them is a clinician. If your child is struggling right now, start with the people below, not an app.
If you need help today. In the US, call or text 988 for the Suicide and Crisis Lifeline. In the UK, Samaritans is 116 123 and Childline is 0800 1111. Your family doctor or pediatrician can refer, and your child's school has a designated safeguarding lead or head of year whose job is exactly this. None of these routes is anonymous in every circumstance, and you should not promise your child that it is. They are staffed by people who can act.
What just happened
On September 10, 2026, EdSource published a report by Vani Sanganeria on schools buying AI chatbots to sit alongside their counseling teams. The company profiled, Alongside, says its chatbot is now in more than 200 schools across 19 states and was used by around 31,000 students in the past year. Those are the company's figures, reported by a news organization rather than independently audited, and worth reading as such.
The reason schools are buying is not mysterious. The American School Counselor Association recommends one counselor for every 250 students. The US national average was 372 to 1 in the 2024 to 2025 school year, the most recent published. In England and Wales, the Youth Endowment Fund surveyed around 11,000 teenagers aged 13 to 17 for its Children, Violence and Vulnerability work, published in December 2025, and found a quarter had already turned to AI chatbots for mental health support, more than had used a helpline or a mental health website. Read those two facts next to each other and you have the whole story: children were going to talk to a machine anyway, as part of a much wider range of things they already use AI for, and schools decided they would rather it was a machine they chose.
That is the reading I would offer, and it is mine, not the researchers'.
Is the school's chatbot the same as the ones in the headlines?
No, and this is the part almost nobody tells parents.
In May 2026, Common Sense Media's Youth AI Safety Institute, working with psychiatrists at Stanford Medicine's Brainstorm Lab, ran more than 3,100 test exchanges across 13 clinical conditions through AI mental health apps. The consumer app Wysa was rated unacceptable for teenagers. The assessment found it failed to recognize signs of genuine psychiatric emergencies and had no human oversight when something went wrong. Two products aimed at schools, Alongside and Sonar, came out differently: in simulated crises they got a trained human on the phone with the test account's guardian within fifteen minutes.
So the headline you half-remember about AI mental health apps being dangerous for teenagers is true, and it belongs with the companion apps children find for themselves rather than the thing your child's school is offering. The difference the assessors kept finding was not smarter AI. It was whether licensed humans were wired into the product at all.
Notice what that means for your decision. You are not really being asked to judge a chatbot. You are being asked to judge a staffing arrangement with a chatbot on the front of it.
What I See in Schools
A wellbeing chatbot is one more place AI has arrived in a school, and there are already more of them than most parents realize. I have sat in enough safeguarding meetings to know where this particular one breaks, and it is almost never at the moment of detection.
A school buys a tool. The tool works. It flags a student on a Thursday afternoon. The alert lands in an inbox belonging to a counselor who is, that week, covering two other schools. Nothing sinister happens. Nobody is negligent. The alert just sits there over a long weekend, because the part of the system that was supposed to be a person was actually a duty roster with a hole in it.
The pattern I see most often when schools adopt anything that touches pastoral care is that the purchase is planned in detail and the response is planned in a sentence. The demonstration shows the alert arriving. Nobody in the room asks who is holding the phone at seven on a Friday evening in the middle of a school break.
That is why I would not spend your one conversation with the school asking whether the AI is any good. Ask this instead.
The Five Minute Question is what I would ask about any wellbeing chatbot a school gives your child: when it flags something serious, who gets the alert, how fast, and what happens if that person is not in the building? The software is not the safeguarding. The person at the end of the alert is.
A wellbeing chatbot is a doorbell, not a doctor. It can tell somebody your child is at the door. It cannot open it.
What to ask the school this week
Three questions, in this order. They take a two-line email.
- When the tool flags something serious, who gets the alert, how quickly, and who covers evenings, weekends and school holidays?
- Do you need my permission, can I opt my child out without them being singled out, and how do I do that?
- What happens to what my child types: who can read it, how long is it kept, and does any of it end up in a record that follows my child?
A good answer to the first one has a named role and a number of minutes in it. A weak answer describes the software. Schools vary widely on the second: one North Carolina district, Cumberland County Schools, tells parents plainly that permission is required for grades 6 to 12 and that they may opt out at any time by emailing the company's privacy address. Yours may do something entirely different, which is why you ask rather than assume.
Does the answer change with your child's age?
Yes, and mostly because of what a child that age thinks the chatbot is.
Under 8. Almost no school is offering this to children that young, and the products in the news start around fourth grade. If your child's school is offering one to a five year old, that is the question, before anything about safety settings.
8 to 12. This is the youngest band these tools reach, and the risk is not that the chatbot says something harmful. It is that a nine year old does not have a category for a friendly listener that is not a friend. Before consenting, ask what the school tells children about what the thing actually is.
13 to 15. The peak years for this, and the years when your child is most likely to use it instead of telling you. That is not a betrayal. Talking to something that cannot look disappointed is easier, which is the whole argument I made in my post on teenagers confiding in AI.
16 to 18. Old enough to want the privacy, old enough to be told the truth about its limits. Say plainly that the tool passes serious things on, because finding that out by surprise is how a teenager decides never to type anything honest again.
Words you can actually use
Short ones work better than a prepared speech.
You could say: "Your school has this chat thing for when you're stressed. Have you used it? What's it like?"
You could say: "I don't mind you using it. I'd just rather be the second person who knows, not the last."
You could say, to a younger child: "It's a computer that's good at listening. It isn't a person and it can't remember you like a friend does. If it's something big, it tells a grown-up at school, and that's on purpose."
And if your child asks whether it is private, do not soften it: "Mostly, yes. But if it thinks you're in danger it tells someone who can help. I'd want it to."
What no school and no chatbot can promise you
Be clear-eyed about the limits, because the marketing will not be.
It cannot catch everything. The EdSource reporting notes these tools miss context, misread how young people talk, and struggle in particular with signs of psychosis. Detection is a filter, not a net.
It does not appear to make children less anxious or less depressed on its own. The reported evaluations show effects on identifying students who need help, not on the underlying condition across a whole school population. Surfacing a child is not the same as treating one.
The strongest numbers come from the company. Figures on students identified and crises averted are supplied by the vendor and reported by journalists. That is not an accusation, it is a stage of evidence, and independent published evaluation is what comes next.
And the school's rules come first on some of this. If a safeguarding lead judges that something must be escalated, your preferences do not override that, nor should they. Equally, if your child has a diagnosed condition or is under any professional care, that clinician's advice outranks anything a school app does and anything in this article. And if you are worried about your child today, the numbers at the top of this page, your doctor and the school's safeguarding lead are all faster than any of this.
What I would watch next
Two things over the coming months. First, whether any of these products publishes an independent, peer-reviewed evaluation rather than a company pilot summary. Second, whether the state laws now landing on AI chatbots and children treat school-commissioned wellbeing tools differently from consumer companion apps, because at the moment much of the drafting does not distinguish them, and a rule written for one could quietly remove the other. If your district follows the pattern I wrote about when New York restricted AI in schools, the wellbeing tool is the one most likely to be swept up by accident.
The honest summary is that a school offering your child an AI listener is not evidence that the school has stopped caring. In most cases it is evidence that the school has more children in distress than it has adults to sit with them, and has decided a triage tool beats a waiting list. That is a reasonable decision. It is also a decision you are allowed to inspect.
Ask who answers. Everything else follows from that.
If your child's school has not yet talked to parents about how it chooses and checks AI tools, that gap is the work I do with schools: helping a school get genuinely ready before it buys, rather than after. It is worth forwarding to your head of year or principal.
I write a Sunday newsletter that keeps teachers and parents current on this without the panic. Parents are very welcome.
Dan Fitzpatrick is a former teacher, assistant headteacher and school trustee who now trains schools on AI. More about Dan.
