KolMe is an entertainment service where you talk with fictional AI characters by voice. It is not a counseling, therapy, medical, or crisis service, and our characters are not therapists or crisis workers. This page describes what our AI does when someone mentions suicide or self-harm, and where to get real help. It is published in accordance with California SB 243.
KolMe's English service is for adults 18 years or older only. Age is confirmed at account creation and stated in our Terms of Service. Accounts found to belong to minors are terminated.
Our characters treat the following as a crisis signal, whether said directly, as a joke, or hypothetically:
| Category | Examples |
|---|---|
| Suicidal ideation | "I want to die", "I don't want to wake up", "everyone would be better off without me" |
| Intent or plan | mentions of method, timing, place, means, or goodbye messages |
| Self-harm | cutting, burning, hitting oneself, intentional overdose |
| Danger to others | threats of violence toward a specific person |
| Acute distress | "no reason to live", extreme hopelessness with isolation |
Ambiguous or joking phrasing is treated as a signal. We do not require certainty to respond.
When a crisis signal is detected, the character immediately stops all role-play, flirting, and romantic behavior for the rest of the call, and:
Our characters will never:
| Resource | Contact | Hours |
|---|---|---|
| 988 Suicide & Crisis Lifeline | Call or text 988 | 24/7, free, confidential |
| Crisis Text Line | Text HOME to 741741 | 24/7 |
| Emergency services | 911 | Immediate danger |
| Veterans Crisis Line | 988, then press 1 | 24/7 |
| The Trevor Project (LGBTQ+) | 1-866-488-7386 | 24/7 |
| Outside the U.S. | findahelpline.com | — |
Crisis-signal events are logged in aggregate and de-identified (timestamp, character, signal category) for safety monitoring and any required regulatory reporting. We do not retain conversation transcripts for this purpose. This protocol is reviewed at least annually, and promptly after any material change to the service, the AI models we use, or applicable law.
Questions about this protocol: support@kolme.kr