SafeSpace is a place people write things they have not said out loud. That puts a higher duty on us than most apps carry, so this policy is written plainly and describes exactly what the app does.
Last updated: 4 September 2026
SafeSpace is a small, independently built app. Privacy questions, requests for a copy of your data and deletion requests go to [email protected], and a person reads them.
Your account: an email address or an Apple/Google identifier, used only to sign you in.
Your display name and avatar: assigned automatically when you join (for example QuietFern) and changeable by you. This is not your real name, and we never publish the name on your Google or Apple account.
What you write: vents, mood check-ins, gratitude entries, art therapy pieces, comments, and anything you choose to share to the feed or the gratitude wall.
How you use the app: which sessions you complete, your daily plan and whether you finished it, your streak, and how a session landed when you tell us.
Health data, only if you grant it: sleep, steps, exercise minutes, resting heart rate, heart rate variability, respiratory rate, blood pressure and mindfulness minutes. It is read from Apple Health or Health Connect, used on your device to shape your daily plan, and is never uploaded to our servers or shared with anyone.
Microphone audio, only if you use voice vents: speech is converted to text by your device’s own speech recognition. We do not record or store audio.
Everything you write is private by default.
A vent is only public if you tap “Share to the feed” after writing it. A gratitude entry is only public if you tick the box to share it on the wall. Your mood check-ins, your plan, your streak, your health data and your art are never public.
What is published is your display name, not your real name, and not your email.
When you finish a vent, its text is sent to Google’s Gemini API to produce a supportive reply, a suggested next step, and, if you choose to share it, tags. Text is also sent to Gemini for the safety check that runs before anything is posted publicly, and for mood insights on your history page.
For Pro members, the supportive reply is sent to ElevenLabs to be read aloud. If that is unavailable, your device’s own voice reads it instead.
These providers process the text to return a result. We do not use what you write to train any model, and we do not sell it. If you would rather nothing left the device, do not use voice playback, and note that the supportive reply cannot be produced without it.