Crisis Support Protocol

Every character on BeCraved AI is an artificial intelligence, not a person, and cannot help someone in a crisis. This protocol explains what happens when a message suggests that a user may be thinking about suicide or self-harm. We publish it as California's companion chatbot law (SB 243) requires.

1. How we recognise it

Each message a user sends in chat and on live cam is checked for language that suggests suicidal thoughts or self-harm. The check is designed to err on the side of showing help, while leaving out everyday expressions such as "dying laughing". Messages are never blocked for this reason: reaching out should not be punished.

2. What the user sees

Crisis resources appear on screen straight away, next to the conversation, with a reminder that the character is an AI. The notice can be dismissed and appears again whenever a later message calls for it. Characters are instructed not to engage with self-harm or suicide content.

3. Where to get help

• United States: Call or text 988 (Suicide & Crisis Lifeline), or text HOME to 741741 (Crisis Text Line)
• United Kingdom and Ireland: Call Samaritans on 116 123
• Netherlands: Call 113 or 0800-0113 (113 Zelfmoordpreventie)
• Anywhere else: Find a free, confidential helpline at findahelpline.com

If you are in immediate danger, call your local emergency number.

4. What we record

We record that a crisis notice was shown: the date, the account and where it appeared. We do not keep the message for this purpose. The record is used only to count how often crisis resources are shown, which we report as the law requires, and is never shared in a way that identifies anyone.

5. Contact

Questions about this protocol can be sent to [email protected].