Report evidence and camera limits
The optional camera estimates visible facial movements on your device, not your actual emotions. Turning it on or off does not change the report scope or accuracy. Current camera handling in the Privacy Notice
Your drawings, short writing, and changes made while drawing.
The service uses what you leave across five steps to create self-reflection questions and reference statements. Camera video and face-related values are not report evidence.
It cannot determine your personality, attachment style, actual emotions, another person’s thoughts, or future behavior.
The result does not replace a medical or clinical diagnosis, psychological assessment, treatment, or professional counseling.
Only if you choose it. When enabled, it estimates visible facial movements on your device.
Video and face values are not stored or sent off-device, and camera-off users complete the same five-step reflection and free result.
MediaPipe Face Landmarker can return face landmarks and, when enabled, blendshape classifications for rendering a 3D face model. These are technical coordinates and animation coefficients, not judgments about emotion, personality, or health. MediaPipe documentation · FaceLandmarkerOptions definition
The camera starts only after you deliberately choose it and grant permission. Blendshapes estimate mouth-corner, eyebrow and mouth movements and changes from the previous observation, not emotional probabilities. Video, landmarks, coefficients and estimates stay transiently in screen memory and are neither stored nor sent to the server or an external AI. Camera data boundary in the Privacy Notice
The current and future handling described here follows the same boundary published in the Privacy Notice.
| Data | On-device use | Off-device handling |
|---|---|---|
| Raw video | Processed transiently in browser memory after camera choice and permission | Not stored or sent |
| Face landmarks | Created transiently on-device to estimate visible movements | Not stored or sent |
| Blendshapes | On-device movement estimates and changes from the preceding observation | Not stored or sent |
| Camera choice | Off or on-device use | Only the choice is stored as a session setting |
| Report evidence | Drawing, writing, and drawing process | No camera values |
Meta Forest maps facial movement to avatars to support interaction between therapists and clients. We reference the interaction pattern only. It is not evidence that automated psychological judgments are accurate or that our reports are more effective. Original SKT Newsroom article
A comprehensive review by Barrett and colleagues found that facial movements do not provide a sufficiently reliable one-to-one map of emotion across context, culture, and individuals. The service therefore creates no anxiety, tension, or calm tags and uses no facial movement as report evidence. Original paper by Barrett and colleagues
The shared NotebookLM material was used only as a starting point for questions. This page links directly to official documentation and original sources.