Privacy Boundaries and Data Compliance

When using the TiRTC AI Chat service, audio data, captions, and related interaction context are transmitted to the Tange platform for processing. This document explains the data flow, storage, and security privacy boundaries to assist your enterprise in compliance assessments.

Media Data Processing

1. Signaling and Transport Encryption

   • All data streams (including media and signaling) between the TiRTC device and the cloud are encrypted through platform-level transport layer security mechanisms. This ensures data security during network transmission and prevents eavesdropping or tampering.
   • Control flows and authentication requests (e.g., requesting peer_id and token) between your application server and the platform enforce HTTPS, and requests are validated using the TGV1-HMAC-SHA256 signature algorithm.

2. Audio Data Retention

   • Device uplink audio streams (user voice) are primarily used for ASR (Automatic Speech Recognition) and streaming understanding. Once recognition is complete, the audio stream is not persistently stored.
   • TTS downlink audio streams synthesized by the platform are cleared from the processing buffers immediately after being sent to the device.
   • The Tange platform does not permanently store the original user recordings or use them for foundation model training, unless diagnostic or tuning services are explicitly enabled under a specific agreement and with customer authorization.

Text and Caption Data

1. Captions and Interaction Context
   • Captions generated during the AI conversation (user intents and AI responses) are passed to the LLM as temporary context for that round of interaction.
   • To meet customer requirements for troubleshooting and historical tracking, this text log is kept in the cloud for a short period by default (typically 7 to 30 days, depending on the service plan or specific agreement), after which it is automatically destroyed.

Regions and Globalization

1. Service Regions and Data Isolation

   • The Tange Cloud AI Chat service cluster supports multi-region deployment. When establishing a connection, the associated media and context data are strictly restricted to the data center in the geographic region specified when creating the application.
   • If your applications and devices are distributed globally, it is recommended to assign devices in different regions to their corresponding regional nodes to avoid non-compliant cross-border data transfer.

2. Compliance Responsibilities

   • As the Data Processor, the Tange platform provides secure, encrypted transmission channels and processing facilities, and promises not to disclose customer business data to third parties.
   • As the Data Controller, your application server is responsible for signing privacy agreements with end users (e.g., device owners), service authorization, and permission assignment (including binding devices to role_id), and must ensure the legal identity and ownership of the device when issuing tokens.