Skip to main content

Data Security Overview

In the AI-driven digital era, data security has become a core safeguard for enterprise operations and growth. Guandata always puts data security first and has built a protection framework that covers the full data lifecycle, including collection, transmission, storage, and application. This document explains the platform's data security measures and protection capabilities across five dimensions.

This document is intended for all users of Guandata Dashboard Insights, especially data security administrators, IT owners, and compliance teams.

Data Minimization: What You See Is What You Get, with Zero Sensitive Data Exposure

Core mechanism

The system strictly follows the principle of data minimization. By relying on the BI platform's aggregation capabilities, it processes raw data before sending only dashboard structure definitions (metadata) and aggregated dashboard result data to the LLM. Raw detailed records are never transmitted to the model.

Security safeguards

  • Source-level filtering: Raw detailed data is never transmitted, eliminating sensitive-data exposure risks at the source.
  • Permission control: Combined with field-level permissions, users can access only the aggregated information within their authorized scope, fundamentally preventing leakage caused by excessive data transmission.

Financial-Grade Encrypted Transmission: Prevent Interception and Tampering

Encryption framework

The platform uses HTTPS as the baseline transmission protocol and integrates AES-128/AES-256 standards for end-to-end protection. TLS 1.3 secures the handshake process and effectively resists man-in-the-middle attacks, while AES encrypts transmitted data byte by byte.

Integrity verification

Each data packet is protected with dynamic encryption salts and message authentication codes (MAC), providing dual protection against interception and tampering and ensuring data integrity throughout transmission.

Zero Data Retention: Compliant with GDPR and MLPS 2.0

Core commitment

Guandata strictly follows a zero-retention policy. In Guandata Dashboard Insights, no conversation data exchanged with the LLM is captured or retained in any form. This policy aligns with the GDPR principle of minimum retention duration and also meets the data-storage security requirements of MLPS 2.0.

Provider coordination

LLM service providers such as OpenAI, SiliconFlow, and DeepSeek explicitly prohibit storage of customer conversation data in their service agreements. As a result, any data sent to the LLM is not retained and is deleted after the response is returned, creating a second layer of protection.

Relevant service agreement clauses: Alibaba Bailian (6.2.5), Volcengine Ark (3.1), SiliconFlow (1.4.1), DeepSeek

Secure Proxy Control: Eliminate Secondary Leakage Risks

Access principle

When using public services provided by LLM vendors, adopt a zero-trust architecture and connect directly to the vendor's official API endpoint.

Risk avoidance

Do not use any unauthorized third-party proxy services. This avoids potential data leakage or misuse introduced by untrusted intermediaries.

Private Deployment: Build a Local Security Fortress

Applicable scenarios

For industries with extremely high data security requirements, such as finance, state-owned enterprises, and government, the platform supports integration with privately deployed LLMs based on models such as DeepSeek-V3 and Qwen3.

Security assurance

Both the data processing engine and the LLM inference service can be fully deployed on the enterprise's local servers or private cloud environment, so data can complete the entire workflow from ingestion to analysis to insight generation without leaving the internal network.

With this multi-layer security architecture, Guandata Dashboard Insights delivers end-to-end protection from data processing to practical business use. Whether for everyday enterprise analytics or for highly regulated scenarios such as finance and government, this security framework provides reliable protection and makes Smart Insights a safe driver of data value discovery.

If you have further questions about data security, contact our product team or your customer success manager.