Skip to main content
HUMANS

Crisis protocol

Last updated: July 2026 · DRAFT pending legal review

If you or someone you know is struggling, help is available. This page explains how HUMANS detects signs of crisis and connects you with support. See also our safety page.

You are talking with an AI

The HUMANS guide is an AI, not a person, and it may not be suitable for some people. HUMANS is for adults 18 and over. The AI is not a crisis service, a therapist, or a doctor.

How we detect signs of crisis

We use two layers. First, a fast automated screen checks every message for language associated with self-harm or suicidal thoughts. When it flags something, a second AI classifier rates the level of risk. Its categories are based on a published, evidence-based framework, the Columbia Suicide Severity Rating Scale (C-SSRS). We store only that a referral happened, never the content of your message.

What the AI does

When a message suggests crisis, the AI stops trying to build your profile for that moment, responds with care, never produces content that could facilitate self-harm, and shows you crisis resources for your region. The resources are also displayed by the app itself, independent of the AI, so they reach you even if the AI phrases something imperfectly.

What we do afterward

A safety conversation is flagged for human review by our team, and that content is not used for matching. Our goal is your safety, not data.

If you are in danger right now

  • United States and Canada: call or text 988 (Suicide and Crisis Lifeline).
  • Costa Rica: call 9-1-1.
  • Brazil: call 188 (CVV).
  • Anywhere: find a helpline at findahelpline.com, or call your local emergency number.