When someone writes about suicide, self-harm or being in danger.
If you are thinking about suicide or hurting yourself, or you are in danger, call or text 988 to reach the 988 Suicide & Crisis Lifeline. It is free, confidential and open all day, every day. If you are in immediate danger, call 911.
Veterans and service members: dial 988, then press 1.
You can also text HOME to 741741 to reach the Crisis Text Line.
Where this applies
The AI conversations on this platform: Civic Advisor, Ask PVOS, Goal consultation and Future Self. Each one says, on screen: “You are talking to an AI, not a person.” None of them has a human name, a face or a persona. Each also says: “This AI may not be suitable for some minors.” All of them are open only to members confirmed as 18 or older, outside classroom mode.
What starts it
A message is checked before any AI sees it. The protocol starts when the words:
- mention suicide, or wanting to die, to end your life, or to not be alive;
- mention hurting, cutting or burning yourself, self-harm or an overdose;
- say you are in danger or not safe, or that someone is threatening or trying to hurt or kill you.
The check reads the words themselves. It does not guess at mood, and it does not look at anything else about you.
What the AI does
- It stops. No AI model is called for that message. On the Civic Advisor, the message is also left out of everything sent to the AI afterwards. Future Self ends the conversation.
- It shows where to get help: call or text 988 for the 988 Suicide & Crisis Lifeline (veterans and service members press 1), text HOME to 741741 for the Crisis Text Line, and call 911 in immediate danger.
- It generates no crisis content. The reply is fixed text written in advance. No AI writes about suicide or self-harm, gives advice on it, or keeps the conversation going about it.
What is kept
One record, each time the protocol runs: which of the four screens it was, and the time. Nothing else. No name, account, email, message text or internet address is stored with it, so the record cannot be traced to you. No one at the platform is alerted, and no one is shown what you wrote.
Those records are counted once a year for the report California asks for (California SB 243, first due July 1, 2027). The report holds the count and this protocol. It holds nothing about any person.
What this is not
This platform is not a crisis service, and no one here monitors conversations. If you or someone near you is in danger now, call 911.
Questions about this protocol: privacy@cross-hug.com. See also the Privacy Policy, section 10.