You decide what KKMatch may capture and use. Consent is requested explicitly before every session and can be reviewed at any time.
Camera (optional), microphone (optional), behavioral analysis (required to build a model), storage (required to save it), research (optional), and sharing (optional).
We do not infer medical, psychological, demographic, or sensitive attributes. The model is about communication and interaction only.
Voiceprint (the biometric characteristics of your voice) is treated as sensitive personal information under the applicable standard GB/T 45574-2025. It is therefore requested as a separate, standalone consent — never bundled with the other consents — and it is off by default.
If you do not enable it, no voiceprint template is derived and no voiceprint-based analysis is performed. If you do enable it, it is used only to compute interaction features (pause length, turn-taking, vocal energy) inside that session, and is never used to identify you or to verify your identity. You can withdraw voiceprint consent at any time from the consent panel; withdrawing stops all further voiceprint processing.
We keep data only for as long as the purpose you approved requires:
Deleting your Human Model triggers a cascade that permanently removes your model, session, conversation turns, behavioural signals, feedback, analytics events, decision-log records, challenge links, and the training samples for that session. Only the statutory billing records described above survive.
Our application servers are located in mainland China. Some processing is carried out by service providers outside mainland China:
Transfers carry only what the purpose requires: text and derived numerical features — never your raw audio or video. Outbound model and voice traffic exits through a dedicated egress so that the transfer path is fixed and auditable. Cross-border processing happens only for sessions you have consented to start, and stops when you delete your data.
Q: Can I delete my data?
A: Yes. Delete removes your model, session, and associated events.
Q: Is my video stored?
A: Only with your storage consent, and only as needed to provide the service you approved.
Q: Who can see my model?
A: Only you, unless you explicitly choose to share or challenge.
Q: Do you use my voiceprint to identify me?
A: No. Voiceprint consent is separate, off by default, and used only to derive interaction features — never for identification or verification.
Q: How long do you keep my data?
A: Raw audio and video are never stored. Your session and model stay until you delete them; training feature samples expire within 24 months; analytics events within 13 months. Billing records are kept for the statutory tax period only.