An agent security risk assessment method and system based on multi-threat index dynamic weighting
The intelligent agent security risk assessment method, which uses dynamic weighting of multiple threat indicators, solves the problem of inaccurate risk assessment in intelligent agent systems, achieves unified quantitative and adaptive risk assessment, and provides interpretable protection strategies.
Patent Information
- Application Number
- CN202610749613.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-05-28
- Publication Date
- 2026-07-17
AI Technical Summary
Existing security assessment methods for intelligent agent systems cannot effectively unify multiple risk indicators, resulting in inaccurate risk assessments and difficulty in guiding targeted protection strategies.
A multi-threat indicator dynamic weighting method is adopted. By calculating the injection rejection rate, privacy leakage rate and dynamic tool-mediated security latency, it is converted into risk components with consistent direction. The weights are dynamically adjusted based on the scenario state vector to calculate the comprehensive risk score and level, and output the protection strategy.
It achieves unified quantification of multiple heterogeneous threats, adapts risk assessment to different scenarios, provides interpretable risk source guidance, and supports pre-deployment assessment, version regression testing, and runtime monitoring.
Smart Images
Figure CN122419951A_ABST