This page describes, in plain language, what this service does when someone appears to be in crisis. It is published because California law requires operators of companion chatbots to publish their protocol, and because anyone using this should be able to read it without asking.
Your companion is an AI. It is not a therapist, a counsellor, a doctor, or an emergency service, it does not diagnose anything, and it is not a substitute for a human being. If you are in immediate danger, call your local emergency number.
Every message is screened for expressions of suicidal ideation, self-harm, and crisis, using both a pattern check and a second language model reading the surrounding conversation rather than the last line alone. The screen is deliberately over-inclusive: it is better to respond gently to someone who was fine than to miss someone who was not.
When it fires, your companion stops whatever it was doing, says plainly that it heard you, asks directly whether you are safe right now, and refers you to the 988 Suicide & Crisis Lifeline (call or text 988 in the US) or to someone you trust. Outside the US, it refers you to your local crisis line. It stays in the conversation with you rather than ending it.
Your companion will never describe, name, or discuss a method of self-harm, including in the course of talking someone out of one. That rule has no exceptions and no framing that opens it.
We keep a count of how many times a crisis referral was issued, with a one-way hash of the account and a timestamp. We do not copy the message, and we do not store what was said, into any separate record. The count exists because California requires operators to report it annually beginning July 2027.
This service is for adults. Companion chatbots may not be suitable for some minors. If a user tells us they are under 18, their access ends immediately and is not restored automatically.
HumanoidStudio · humanoidstudio.ai