How Deux responds to suicide and self-harm
Deux is an AI helper for conversations. It is not a crisis service, a therapist, or an emergency service. This page describes how Deux is built to respond when someone may be thinking about suicide or self-harm, as California's law on companion chatbots (SB 243) asks us to publish. If you or someone near you may be in danger, contact your local emergency number or a crisis line now.
How Deux notices
- Word check. Before Deux asks the AI service for an answer, it checks what you wrote for clear signs of suicide, self-harm, or danger. This check runs on Deux's own servers and reads English.
- AI safety check. A separate AI safety check reads what you wrote in any language, including every language the app is shown in, while the answer is being written. It also reads Deux's answer before you see it.
- Tips Deux offers on its own. When you turn on active mode in a chat, the same checks read the words a tip would be based on first. A sign of danger means Deux offers no tip.
What Deux does
- When a check finds such signs, Deux does not use an answer written by the AI service. It shows help lines for your country instead (in the US, call or text 988, or call 911 in an emergency), in your language where the app has it, and suggests reaching someone you trust.
- If an answer written by the AI service touches suicide, self-harm, or danger, Deux does not show it. It shows the same help instead.
- Each help line in the app was read from its service's own page, and the app keeps the date it was checked.
Counting
Deux counts how many times a day it shows such a help notice, with no names, no messages, and no way to tell who saw one. California's law asks us to report this number to its Office of Suicide Prevention every year, starting in July 2027.
Questions
Use our contact form. You can also read the Deux Privacy Policy.