safety

she tells you she's an ai

her first text says so. after that, the first message she sends you after three hours or more ends with “(reminder: i'm an ai, not a person.)” That includes her morning and evening notes.

if you text about suicide or self-harm

every message is checked before any ai writes a reply. when a message says you want to die, hurt yourself, or end your life, zemi does not improvise. she stops the normal conversation and sends this, word for word:

i'm really glad you told me, and i'm taking it seriously.

i'm not the right thing to carry this alone with you, and i want you safe. please reach out right now:

- call or text 988 (Suicide & Crisis Lifeline) - free, 24/7
- text HOME to 741741 (Crisis Text Line)
- if you might act on this or you're in danger, call 911

i'm still here and i'm not going anywhere. talk to me. but talk to them too - they're trained for exactly this.

these resources are in the united states. zemi keeps a record that this reply was sent and what kind of message it answered. that record never includes your message itself. if you are outside the us, contact your local emergency number.

minors

companion chatbots may not be suitable for some minors.

contact

questions go to support@zemi.chat.