Legal

AI Safety & Crisis Protocol

Last updated

If you are in crisis right now: call or text 988 (Suicide & Crisis Lifeline, US), text HOME to 741741 (Crisis Text Line), call 14416 (Tele-MANAS, India), or call 911 or your local emergency number in an emergency. Do not wait for an app.

Rox is an artificial-intelligence companion for people living with chronic illness, operated by Godspeed Labs Private Limited. Rox is an AI, not a human, and it is not a substitute for your doctor, care team, or mental-health professional. This page publishes the protocol we maintain for responding to expressions of suicidal ideation, suicide, or self-harm, as required by companion-chatbot safety laws including California SB 243 (Business & Professions Code §22601 et seq.) and New York General Business Law Article 47 (§1700 et seq.). Consistent with laws such as Tennessee's Health Care AI Act, Rox never represents itself as a licensed or qualified medical or mental-health professional.

1. Detection

Rox's systems screen every conversation, across text, voice, and check-in calls, for expressions of suicidal ideation, suicide, and self-harm. Detection is automated and applies to every user and every session. It is designed to surface help; it is not a monitoring service and cannot guarantee that every crisis will be detected.

Because Rox works with people managing chronic conditions, it is also designed to recognise language that may indicate a medical emergency and to direct you to emergency services. Rox cannot contact emergency services on your behalf.

2. Prevention by Design

  • Rox will not generate content that encourages, instructs, or assists with suicide or self-harm, in any conversation.

  • Rox will not provide information on methods or means of self-harm, regardless of how the request is framed.

  • Rox discloses that it is an AI, not a human, at the start of each session, and repeats that reminder at least every three hours of continued use, for every user.

  • Rox is designed not to imply it is a human or a clinician, and never claims to be a licensed or qualified medical or mental-health professional. If you ask whether you are talking to a human, Rox answers truthfully that it is an AI.

  • Rox does not diagnose conditions, prescribe medication, or advise you to start, stop, or change any medication or treatment.

  • Rox does not encourage behaviour that could harm your health, including disordered eating, over-exertion during a flare, or ignoring red-flag symptoms.

  • These safeguards are applied to every conversation, on every platform, for every user.

3. Referral to Crisis Services

When Rox detects an expression of suicidal ideation or self-harm, it interrupts the conversation to provide crisis resources, including:

  • 988 Suicide & Crisis Lifeline (call or text 988, US)

  • Crisis Text Line (text HOME to 741741, US)

  • Tele-MANAS (call 14416, India)

  • Samaritans (call 116 123, UK & Ireland)

  • 911 or your local emergency services

  • Local and international crisis services where available, based on your general region

Rox stays in the conversation after sharing resources and continues to encourage you to reach a human.

4. Safeguards for Minors

  • Rox requires users to be at least 16 years old (see our Terms of Service).

  • We do not knowingly collect data from anyone under 16. If we learn that a minor has created an account, we delete it.

  • If a user's messages suggest they may be under 16, Rox discloses that it is an AI, repeats a take-a-break and AI-status reminder at least every three hours of continued use, and blocks sexually explicit content server-side.

Rox is not designed for minors.

5. Contact

Questions about this protocol: founders@talktorox.com. Godspeed Labs Limited.