Last updated November 2026
Every session in Socially Awkward is a live, spoken role-play: you talk, and an AI plays the other person — a barista, a coworker, a stranger — in real time. That requires sending your voice to an AI provider capable of holding a live spoken conversation, and generating a written review afterward requires sending the session transcript to an AI model. This notice explains exactly what that involves.
Before your first AI voice session, the app shows a consent screen explaining that your voice will be sent to a third-party AI provider to run the conversation. You won't be put into a live session without seeing that screen first.
You can delete all of this earlier, at any time, using Delete My Data in the app's Settings — see Support.
OpenAI provides the real-time voice conversation. Google (Gemini) generates the text of your post-session review from the session transcript. We send each provider only what's needed for that part of the session — we don't send either provider your purchase history, your device identifier, or any other app data beyond what a given session requires.
The app only accesses your microphone while a live session is active, and only after you've granted microphone permission in iOS. You can revoke this at any time in Settings → Socially Awkward → Microphone, which will prevent new sessions from starting until you grant it again.
We use session metadata (which scene, which tier, how long, whether it was completed) in aggregate to understand which scenes are useful and which need work. This is product-improvement analysis, not profiling of individuals, and it's never used to tailor advertising.
Questions about how AI is used in the app: hi@sociallyawkwardapp.com. See also our Privacy Policy.