Self-Hosted AI Processing
The most important thing to know about Torki's privacy approach: your conversations are processed on our own dedicated GPU servers. Unlike most AI platforms that route your messages through third-party API providers, Torki runs its AI models locally. Your chat messages never leave our infrastructure for AI processing.
Encryption
All data transmitted between your device and Torki is encrypted with TLS encryption, ensuring your messages cannot be intercepted in transit. OAuth tokens for connected services (Google, Microsoft) are stored encrypted at rest, so even in the unlikely event of a data breach, your connector credentials remain protected.
No Training on Your Data
Your conversations are never used to train AI models. What you say to Torki stays between you and Torki. This is a fundamental commitment, not a setting you need to enable.
Minimal Data Collection
Torki collects only what is needed to provide the service: your account information, chat history (if enabled), and connector tokens. We do not sell data, serve ads, or share your information with third parties for marketing purposes.
Connector Security
When you connect external services like Google Calendar or Microsoft OneDrive, Torki uses industry-standard OAuth 2.0 authentication. Torki never sees or stores your Google or Microsoft passwords. Tokens are encrypted and can be revoked at any time by disconnecting the service.