Tropic integrates AI into several features while maintaining strict standards for data security and privacy. Below are common questions we receive about how we handle your data when AI tools are involved.
Does Tropic use my data to train AI models?
No. Tropic does not allow customer interactions to be used for training public large language models. Your data remains private and is never shared with vendors for model improvement.
Is my data shared with other customers?
No. Tropic ensures data isolation across all AI tools. Prompts, responses, and documents are separated by client to maintain confidentiality and prevent cross-access.
Does the AI model retain my data after use?
No. Tropic AI interactions are stateless, meaning no data is stored in the model after processing. Each interaction is independent and does not persist in the system.
Are AI prompts or responses monitored by humans?
No. Tropic disables vendor-side monitoring of prompt and response data wherever possible. Human oversight is limited to protect your privacy.
How does Tropic ensure AI remains secure and compliant?
Tropic evaluates every AI-powered tool before release to minimize risk and protect customer data. This includes:
- Requesting vendor configurations that prevent data retention or model training
- Requiring data encryption for prompts, responses, and file uploads
- Enforcing client-level data separation
- Reviewing agreements to confirm privacy and security commitments
- Disabling vendor-side prompt logging, where possible
Example:
For tools like Ask Tropic, our team works with vendors to configure environments so that customer prompts are not logged or stored. We also ensure encryption is used during transmission and review contracts to confirm vendor obligations.
These practices are reviewed periodically and updated as needed.