Safety

How Driftstone handles crisis situations

Last updated: 17 November 2026. Published in line with California Business and Professions Code § 22602 (SB 243) and New York General Business Law Article 47.

Driftstone is an AI tool for self-reflection for adults. It is not a crisis service. This page describes what the app does when what someone writes suggests they may be in danger.

What the app looks for

How it checks

  1. On the device, before anything is sent. The app checks the text against a list of unmistakable statements.
  2. On our server. The same list is checked again.
  3. By the AI model. The model is instructed to flag any of the situations above, including ones that use no explicit words, and to prefer flagging when unsure.

What happens when something is flagged

These resources are always available under Settings > Crisis resources.

What the AI is not allowed to do

Reflections must not encourage self-harm, give medical or therapeutic advice, diagnose, or claim to be a person or to have feelings. Driftstone does not role-play and does not hold conversations.

AI disclosure

Every reflection is labelled "Written by AI". The app states during setup and in Settings that it is not a person, not therapy, and not a substitute for professional care.

What we record

The app keeps a count, on the device, of how many times the crisis page was shown. It does not record what was written, and the count stays on the device.

Limits

Automated checks miss things and sometimes flag things that are not a crisis. If you are struggling, please reach out to a person, whatever the app says.

Contact: support@getdriftstone.com