What is SafeCircle?
An introduction to SafeCircle — what it does, how it works, and what data it collects.
SafeCircle is a family safety platform that helps parents stay informed about possible risks in their child's online chats, without reading their messages.
It has three parts you use today: a parent dashboard (a web app at app.safecircle.tech), a parent app for Android, and a child app for Android. The child app reads the text visible on screen in a few messaging apps and analyses it on the child's phone. When it sees a possible risk, you get an alert with the risk level, the categories, a confidence score and a short note. The message text itself is never sent to SafeCircle's servers.
An assistant, not a verdict
The detection can be wrong in both directions: it can flag harmless chats and miss real ones. Treat an alert as a prompt to talk with your child, not as proof. See AI limitations.
How it works
Set up the dashboard and pair the phone
You create a child profile in the dashboard (or the parent app) and generate a pairing code. On the child's Android phone you install the child app from the Google Play internal testing track (see Join the beta), scan the QR code or type the 6-digit code, and grant the Accessibility permission.
The child app reads on-screen text
Using Android's Accessibility Service, the app reads the text that is visible in the supported apps (WhatsApp, Instagram, Telegram, Snapchat and TikTok), and nothing else. It starts in Basic mode, which scans for keywords and needs no download. You can switch it to an on-device AI model for more accurate detection. See How the child app works.
Only a short summary is sent
When the result is medium or high risk, the app sends an alert with the category, confidence, timestamp, source app and a short note. Weaker results are discarded on the phone. The app analyses each monitored app at most once every 5 minutes.
You review the alert
The alert appears in the dashboard and as a push notification in the parent app, with a risk level (medium or high), the categories, a confidence score and the AI Reasoning note. For now that note is a plain summary of the detected categories, not a quote from the conversation. See Understanding alerts.
Supported apps and devices
- Messaging apps the child app reads: WhatsApp, Instagram, Telegram, Snapchat and TikTok. Other apps are not monitored. See Supported apps.
- Child's phone: Android 11 or later. There is no iOS app. See iOS.
- Parents: the dashboard works in any browser, and the parent app runs on Android 11 or later.
Privacy overview
| What SafeCircle stores | What it does NOT receive |
|---|---|
| Risk level, categories and confidence score | The text of messages, in any form |
| Timestamp, source app and a short note (a summary of the categories) | Screenshots or screen recordings |
| Your account, children, devices and pairing status | Contacts, location, photos or the microphone |
| Alert read state | Browsing history |
Traffic between the apps and SafeCircle's servers uses HTTPS. The recent chat text the app analyses stays on the phone, in memory only. Read AI privacy and Data and privacy for the full detail.
Beta
SafeCircle is in beta, and anyone can create an account.
Get started
Join the beta
From a new account to a paired device: sign-up, Google Play testing and the beta agreement.
Set up the parent dashboard
Add a child profile and generate a pairing code.
Set up the child app
Install, pair, and grant permissions on the Android app.
Get the parent app
Alerts, insights and push notifications on your Android phone.
Explore the docs
Platform
The dashboard, the Android apps, search, iOS status and Canopy OS.
How the AI works
On-device detection, risk categories, privacy and limitations.
SafeCircle Search
A private search engine with safe search. A separate site, not connected to the apps yet.
Developers
Architecture, API reference, authentication and data handling.
Support
FAQ, troubleshooting and known limitations.