Published before you pay
If your teenager says something that suggests they are in danger, here is exactly what Atlas Guild does — and, just as importantly, what it does not do. No summaries, no reassurance we can't back up.
If you need someone right now
Call or text 988
The 988 Suicide & Crisis Lifeline is free, confidential, and staffed 24 hours a day across the United States. You can call, text, or chat at 988lifeline.org. You do not have to be in immediate danger to use it, and you can call about someone else.
Prefer to write rather than talk? Text HOME to 741741 to reach Crisis Text Line, also free and 24/7.
If someone is in immediate physical danger, call 911. Atlas Guild is not an emergency service and cannot summon help.
Worth knowing, since parents ask us: if you are worried about your teenager, asking them directly does not put the idea in their head — that is a persistent myth, and the research says the opposite. Nothing described below is a substitute for that conversation.
The crisis protocol runs when the conversation indicates a young person may be at risk of harm — to themselves or from someone else. It is not triggered by a bad mood, a hard day, an argument at home, or a sad story your teen is writing. Those are ordinary teenage conversations and the app treats them as such.
We publish the categories that raise an alert in full. We do not publish the thresholds inside them. A safety system that shows exactly where its own edges are stops being a safety system, because the edges become instructions.
Fixed, not improvised
In every other part of Atlas Guild, the character generates its own words. Here it does not. When the crisis protocol fires, the response is a fixed piece of text written by people, reviewed before it ever ships, and delivered exactly as written. Nothing is invented in the moment.
This is deliberate. A language model asked to be comforting will sometimes be agreeable instead, and agreeable is the wrong thing to be when a young person is describing harm. Removing the model from this one moment removes that risk entirely.
Cannot be turned off
Everywhere else on Atlas Guild, your family sets the rules. Not here. The crisis protocol is not a feature in your dashboard, it is not part of any plan, and it is identical on the free plan and the paid one. We do not sell safety as an upgrade.
Your teenager cannot switch it off either, and they are told so before their first message — not buried in terms, but on screen, in plain language.
Toward a person, not deeper into the app
The reply names a real, staffed service your teen can reach immediately — 988 by call or text, or Crisis Text Line — and encourages them to talk to an adult they trust. It does not offer to keep talking as a substitute. It does not suggest coping techniques, breathing exercises, or anything that could read as treatment.
An AI character is not what a young person in crisis needs, and the honest thing for it to do in that moment is say so and get out of the way.
Day to day, your teen decides what you see. That is not us being soft on them — it is the only arrangement that works. A teenager who believes every word is being forwarded to a parent stops saying anything real, and an app nobody says anything real to cannot help anyone.
The crisis protocol is the exception, and it is the exception your teen is told about up front. When it fires, you are notified. You are shown what triggered it and the part of the conversation that caused it — not the rest of their history, not last week, not the things that have nothing to do with it.
The full detail of who is told what, and at which level, is in the alert categories.
Atlas Guild does not diagnose anything, treat anything, or replace a doctor, a counsellor, or a crisis service. It notices, it says something honest and fixed, and it points at people who are qualified to help.
Nobody is watching a screen in real time. Atlas Guild cannot dispatch help, cannot call anyone on your behalf, and cannot intervene physically. If someone is in immediate danger, call 911 — do not wait for an alert.
We see what is said to Atlas Guild. That is all. We do not see their texts, their social apps, their browser, their photos, or anything else on the device. A monitoring app does that job and this does not replace one.
Every safety document you read will tell you what it catches. This is the part that tells you what it misses, because you are making a decision about your child and you should make it with the real picture.
Not built yet
The fixed crisis reply, the categories, and parent notification are in the September build. Two things are not finished, and we would rather say so here than let you discover it later:
How this document was written
This page and the crisis reply itself were written against the published safe-messaging guidance from the American Foundation for Suicide Prevention, the Suicide Prevention Resource Center, and the National Action Alliance Framework for Successful Messaging — no methods or means, no location detail, nothing that frames harm as inevitable or reasonable, help resources on every page that touches the subject.
Following published guidance is not the same as having been reviewed by a clinician, and we are not going to blur those two things. Independent clinical review is scheduled. When it happens we will publish what it says here, including anything it tells us to change.
See where we stand against every Common Sense Media condition →
