一种服务化人工智能模型的多级安全防护系统及方法
By constructing a multi-level security protection system for service-oriented artificial intelligence models, and adopting SM2/SM3/SM4 algorithms and Docker container technology, the system solves the problems of fragmented development models and insufficient security protection, realizes standardized model supply and full lifecycle security management, and improves model security and reusability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- IND INFORMATION SECURITY (SICHUAN) INNOVATION CENT CO LTD
- Filing Date
- 2026-05-09
- Publication Date
- 2026-07-17
AI Technical Summary
The fragmented development model of existing artificial intelligence models, the lack of full life-cycle security protection, and the insufficient autonomy and controllability of cryptographic technology result in low model reuse rate, poor security, and low integration efficiency.
A multi-level security protection system for service-oriented artificial intelligence models is constructed, including a three-level progressive security protection system of registration level, application level and deregistration level. The SM2/SM3/SM4 algorithms are used for identity authentication, encryption and decryption, model encapsulation and key management. Combined with Docker container technology, the model is standardized service-oriented encapsulation and multi-level autonomous and controllable security protection are realized.
It has enabled the standardized service provision of models, improved the security and availability of models throughout their entire lifecycle, reduced the risk of models being stolen and tampered with, increased model reuse rate and integration efficiency, and met the requirements of independent and controllable security standards.
Smart Images
Figure CN122160196B_ABST