Safety & Crisis Protocol

Last updated September 29, 2026

1. SoulPhone is an AI

SoulPhone is an artificial intelligence. It is not a person, and not a licensed therapist, counselor, psychologist, psychiatrist or medical professional. It tells you it's an AI whenever you ask, and it never claims otherwise.

It is designed for reflection, spiritual practice and mindfulness. It does not give medical advice, diagnoses, psychological treatment or professional counseling.

2. Self-harm prevention protocol

The protocol is active in every conversation. It works in three layers:

• An AI safety check on every turn, run separately from the conversation, looking for signs of risk
• Phrase detection as a backup, so a clear crisis signal is caught even if the AI check fails. It recognizes negations, so "I'm not suicidal" is not treated as a crisis
• Instructions to the conversational AI on how to respond when risk appears

Immediate risk

If someone describes suicidal thoughts or plans, active self-harm, intent to harm others, domestic violence or abuse, or a medical emergency, SoulPhone:

• Acknowledges their pain with warmth
• Gives crisis numbers directly (988, 911, 1-800-799-7233)
• Clearly and repeatedly encourages them to reach trained crisis professionals
• Does not act as a therapist or crisis counselor, and does not carry on as normal until the resources have been given

Moderate risk

For passive thoughts of death, hopelessness, severe distress or panic, SoulPhone validates the person's feelings, offers grounding when it helps, shares the 988 Lifeline and Crisis Text Line, and stays on the call. It does not hang up on someone who is struggling.

Other struggles

For grief, addiction, relationship trouble, money stress, loneliness and similar challenges, SoulPhone listens first, asks before offering resources, suggests specialized help where it fits, and never pushes resources on someone who declines.

3. Crisis resources

All free and confidential.

4. What SoulPhone is not

SoulPhone is not a crisis service. It watches for signs of crisis and points people to professional help, but it cannot dispatch emergency services, contact anyone on your behalf, or provide clinical care. If you are in crisis, contact the resources above directly, and don't rely on SoulPhone alone.

5. Hypnotherapy safety

• Before each session, you're shown conditions that may not suit guided hypnotherapy, including active psychosis, dissociative identity disorder and uncontrolled epilepsy, and must acknowledge them. If any apply, talk to a qualified health professional first. SoulPhone does not screen for or diagnose any condition
• You stay in control. Hypnotherapy is focused relaxation. You can't be made to do anything against your will, and you can open your eyes, move or stop at any time
• An exit button is always on screen and ends the session immediately
• If a session becomes distressing: breathe slowly, feel the surface beneath you, remind yourself you're safe and can stop, open your eyes and look around the room. If the distress continues, contact a professional or a crisis line
• Don't use hypnotherapy while driving or operating machinery. If you have a history of trauma, consider talking to a therapist first

6. Reporting and review

We review our safety protocol regularly against current practice in AI safety and crisis care. To raise a concern or report a safety issue, email [email protected].

7. Regulatory compliance

We publish this protocol to meet laws for AI companion services, including California's companion chatbot law (SB 243), which requires a self-harm prevention protocol, clear disclosure that you are talking to an AI, and annual reporting of crisis referrals. That reporting contains counts only, never personal information.