|
Work item:
|
X.fr-SGAI
|
|
Subject/title:
|
Functional requirements for security gateway for AI application services
|
|
Status:
|
Under study
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
2029-12 (Medium priority)
|
|
Liaison:
|
ISO/IEC JTC1 / SC 27 and SC42
|
|
Supporting members:
|
China Telecom Communications Corporation, China Unicom, China Information Communication Technologies Group
|
|
Summary:
|
A security gateway for AI application services serves as an intermediary control layer between AI applications and AI-related services, providing a unified point of governance for AI interactions. It enables centralized enforcement of security policies, monitoring of interaction and usage behaviors, and consistent application of security controls across heterogeneous AI components, including models, agents, tools, and knowledge-based services. By providing unified governance and interaction control, it helps ensure that AI interactions remain secure, reliable, and controllable across different application environments, particularly in scenarios involving complex, distributed, and multi-component AI systems.
This Recommendation defines a functional framework and associated requirements for such security gateways to address security challenges arising from the increasing integration and orchestration of large language models, intelligent agents, tools, and knowledge-based services. It specifies functional requirements organized into five parts covering asset security management, protection, detection, response, and auditing. These functional requirements collectively support secure governance of AI-related interactions, including access control, data and context protection, interaction monitoring, agent behavior governance, resource usage management, risk response, and security traceability. In addition, this recommendation provides reference deployment models for cloud, on-premise, and hybrid environments to support flexible implementation across different operational scenarios.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2026-06-10 15:13:41
|
|
Last update:
2026-06-10 15:18:22
|
|
