Safety
Safety built into the rules, not bolted on later.
Homeroom is not a school-security company, and we will not pretend to be one. What we do is make the system itself safe by construction: a child can only be reached in an emergency, sensitive records are hidden by default, the content filter never shows a minor the words it blocks, and our AI keeps nothing. Here is exactly what that means — and what we leave to the specialists.
This page lists only what is shipped today, plus an honest list of what we do not yet offer.
Safe because of how it is built
Emergencies only, by construction
There is no marketing or fundraising call lane in the system, so a non-emergency call cannot be sent — it is refused. When the emergency calling channel is turned on, a call reaches only a student’s permitted guardian-of-record contacts. Emergency-only scope shipped
Content moderation, blocked by default
Submitted images and text are checked, and the content-filter word list stays on the server — a minor never sees the words being filtered. If the check ever hiccups, it blocks by default rather than letting something through. Shipped
AI that keeps nothing
All AI text hides personal information, keeps nothing afterward, filters its own output, and refuses to create images of people. Shipped
No alert fires twice
A re-mark or a re-send never notifies the same guardian twice — a safety property built into the system, not a habit staff have to remember. Shipped
Extra care where it matters most
Some records need more than ordinary privacy. Self-harm, harassment, and law-enforcement incidents are hidden from lists and shown only to an administrator, one record at a time. For any hidden student, the incident narrative, the student name, and referral notes are removed by default. This is the same permission and privacy spine the rest of the platform runs on — not a separate, bolt-on policy. Shipped
What we leave to the specialists — for now
We are honest about the school-safety tools we do not offer. Homeroom does not do visitor management or sex-offender screening at the front door, panic-button alerting or the reunification workflows tied to it, device and web content filtering on student laptops, or always-on monitoring of student accounts for self-harm. Those are real, important jobs — and today they belong to dedicated safety vendors built for them. We will not imply otherwise. The emergency phone-call channel that does exist is in early access: it turns on only once a per-school calling key is set up, so Homeroom does not place live emergency calls on its own today. Emergency calling channel in early access
We’re honest about what’s shipped
The emergency-only message scope, content moderation blocked by default, AI that keeps no personal data, no-double-notify alerts, and extra-care hiding of sensitive incidents are live today. The emergency calling channel is in early access, turning on with a per-school calling key. Visitor management, panic-button alerting, device filtering, and account monitoring are not offered — and we tell you plainly rather than pretend.