Last updated: September 25, 2026
This policy describes the iOS app, including live avatar rooms, optional AI voice chat, and safety reports. No account or login is required.
With your consent and camera permission, 3D Chat Room uses ARKit face tracking and the front camera, including the TrueDepth camera on supported devices. It processes camera images of your face, a three-dimensional face mesh, facial-expression blend-shape coefficients, head position and rotation, and eye positions and rotations. It derives a face texture, face-shape corrections and an average skin color for the avatar. The app does not access Face ID authentication templates or use your face to identify or authenticate you.
Face data is used only to draw and animate your live 3D avatar, including its appearance, expressions, eyes and head movement, and to adjust the viewpoint of the 3D scene. You can see your own avatar in the mirror view. There are no other planned uses. We do not use face data for advertising, analytics, profiling, identification, or training AI or machine-learning models.
In a room with other people, face textures, face-shape corrections, expression coefficients, and head and eye poses are sent directly to the participants you allow to display your avatar. Other participants are recipients of your face data. If you explicitly connect your iPhone to the Mac companion, your phone sends this data to that Mac, which renders it and can forward it to its room participants. In a solo AI room, face data is rendered locally and is not sent to xAI.
The iOS app does not send face data to the developer's signaling or moderation server, advertising or analytics services, or AI providers. Face data is not included in reports or block notifications. Apart from the participants and connected Mac described above, we do not disclose or share face data with third parties. Direct peer media is not end-to-end encrypted. Other participants may capture what their devices display; the app cannot control their separate screenshots or recordings.
The iOS app does not save captured face data to files, databases, backups, or cloud storage. It processes face data in volatile CPU and GPU memory on your device and receiving participants' devices. Leaving a room stops camera capture and transmission; working buffers, including your most recent face texture, can remain in the running app's memory until replaced or the app process ends. Closing the app process releases that memory. There is no developer-held face-data record to delete. You can stop further capture by leaving the room or revoking camera access in iOS Settings.
In rooms with other people, microphone audio is compressed and sent directly to participants you allow for live conversation. The app does not save recordings of these calls. Short working audio buffers exist in memory. Muting stops live microphone sharing; leaving stops microphone capture.
If you choose an AI room and accept the separate AI voice disclosure, your microphone audio and conversation text are sent to xAI to generate the AI companion's replies. Face data is not sent to xAI. The app holds audio and conversation text in memory and does not save a conversation history to disk. xAI's published API policy describes default retention of inputs and outputs for 30 days for abuse auditing, after which they are deleted; it says API data is not used for training without explicit permission. We do not authorize training on your conversations. See xAI API data practices and xAI's privacy policy. Declining AI sharing leaves rooms with people available.
The signaling server receives network addresses, port numbers, room selection and peer connection information so participants can connect directly. It does not relay room media. The app creates a random installation identifier so blocks continue to apply across app launches. This is not derived from your face and is not an advertising identifier. It is shared with room participants and the signaling server, and accompanies safety reports. Signaling room state is held in memory; server connection logs may contain network addresses and room events.
Reporting content or blocking a participant sends the action, reported participant's and reporting installation's identifiers, room, time, app version, selected reason, and any description you enter to the developer. No face imagery, face-tracking measurements or audio is attached. Please do not include private information in the description.
Reports are sent over HTTPS to a Google Apps Script operated by the developer and emailed to brkuprel@gmail.com for review. Google processes report metadata to host the receiver and deliver email. Unsent notification payloads are stored in the script's private storage for up to 30 days and removed after successful notification. A separate private review record contains the report ID, reported installation ID, room, reason, receipt time, 24-hour deadline, decision and review note. Open review records remain until resolved; resolved records are deleted within 30 days, or sooner when storage space is needed. Limited report identifiers, content hashes and delivery timestamps are retained for up to 30 days to prevent duplicate notifications. Delivery may be delayed by service outages or email limits. Notification emails are retained until the safety review and any related follow-up are complete, then deleted, unless retention is required by law. Pending reports are stored on your device and retried until delivery is confirmed; the local pending copy is then removed. Deleting the app removes unsent reports. Contact the developer to request deletion of a submitted report. See Google's privacy policy for its service data practices.
The app fetches current moderation policy over HTTPS on entering a human room and every minute while there. The policy exposes only banned installation identifiers and whether rooms are enabled; it does not expose reports, reporter identifiers, allegations or review notes. Google receives the network request, but the app does not include your installation ID in this policy request. Banned identifiers are kept for as long as the ban is in force so clients can remove offending live content and reject further contact.
Participant approvals and temporary hides last for the current room visit. New participants' media is withheld until you explicitly allow it. No audio transcription, image classifier, or third-party media-analysis service is used for this filter. A report or block does not upload any live media.
Your device stores terms and privacy consent, AI voice consent, the random installation identifier, blocked identifiers, appearance settings, and unsent safety reports. These are separate from face data. You can undo local blocks from the lobby. Deleting the app removes its local data; restoring a device backup may restore local settings. To request deletion of developer-held report data, contact the address below. No advertising or cross-app tracking is used.
3D Chat Room is intended for people aged 18 and older.
Privacy questions, safety concerns, or deletion requests: brkuprel@gmail.com.