Tuesday 3 November, 2026
Driven by the rapid advancement of AI agents, next-generation intelligent networks are increasingly evolving from connecting people and things towards also connecting distributed intelligence.
As network architectures and service paradigms continue to evolve, network environments may increasingly incorporate distributed, intelligent, and heterogeneous elements. While these developments unlock new capabilities, they also introduce new challenges related to security, trustworthiness, and governance. Zero Trust principles have been adopted in various conventional network environments, providing useful references for evolving network security architectures; however, as autonomous entities increasingly perform complex tasks, delegate actions, invoke network capabilities, and interact across different contexts and administrative domains, additional trust and authorisation considerations may arise beyond those addressed by conventional identityand access-oriented implementations.
This white paper focuses on two categories of scenarios: first, scenarios in which external AI agents (Third-party AI Agents) invoke network capabilities exposed by MNOs; and second, AI for Networks scenarios within the MNO, in which Network AI Agents and AI-enabled network functions participate in network operation, management and orchestration.
While autonomous-network frameworks already support intent-driven operation and closed-loop automation, agentic interactions introduce additional requirements to ensure that delegated and evolving actions remain trusted and within the authorised objective. To address these challenges, this white paper proposes a new trust framework for Agentic Networks: the MAIC-Trust Framework (hereinafter referred to as M-Trust).
