Last updated: 14 November 2026
StreamSim is a simulation for fun and practice. There is no real audience: every viewer, message and gift in the app is generated.
The camera picture is shown only on your phone. Speech recognition runs through Apple's speech service to turn what you say into text so the simulated chat can react.
When AI viewers are on, the app sends a small still frame (about 512 pixels) from your camera and the last sentence you said to our server roughly every 7 seconds. The server passes it to Anthropic's Claude model to write chat messages and returns them to the app. Frames and text are used only for that request. We do not store them, and we do not use them to train models. Anthropic processes the request under its commercial terms and does not train on it.
For cost and abuse control the server logs, per request: the time, the number of AI tokens used and a shortened one-way hash of a random install ID. No pictures, no words, no name, no email.
Recordings you make with iOS screen recording stay on your device unless you share them yourself.
StreamSim is not directed at children under 13.
support@yamosta.com