AI-Native Network Controller: A Modular Framework for Safe Agentic Control of Multi-Domain Network Infrastructure

Researchers introduce AI-Native Network Controller (AI-NNC), an open-source modular framework enabling coordinated AI control across heterogeneous network infrastructure spanning radio access, optical transport, and core networks. The system prioritizes safety by routing AI agent commands through validated domain-specific applications rather than direct equipment access, addressing a critical gap in 6G network management.

The emergence of AI-Native Network Controller represents a meaningful shift in how complex infrastructure systems approach autonomous management. The paper addresses a genuine technical gap: existing network controllers operate in silos—O-RAN RIC handles radio domains while optical and core networks lack integrated AI automation—creating coordination inefficiencies and safety risks. This fragmentation becomes increasingly problematic as 6G deployment approaches, requiring seamless orchestration across historically separate domains.

The framework's design philosophy prioritizes safety through architectural constraints rather than trust. By implementing protocol-agnostic adapters for individual devices while restricting AI agents to domain-specific application layers, the system creates a validated control boundary. This resembles security-by-design principles seen in critical infrastructure, where automation must maintain human-verifiable safeguards. The modular approach also enables practical benefits: unified dataset collection, controlled experimentation environments, and coordinated testbed operations reduce fragmentation costs for network operators and researchers.

For telecommunications infrastructure and network equipment vendors, this development signals increasing pressure to embed AI-native capabilities and standardized interfaces. Enterprise network operators face strategic questions about controller architecture investments and interoperability standards. The open-source positioning may accelerate adoption among research institutions and smaller operators, potentially creating competitive pressure on proprietary solutions. The framework's emphasis on safe agentic control addresses legitimate industry concerns about deploying autonomous systems in mission-critical infrastructure, suggesting a pragmatic regulatory alignment that could facilitate faster real-world deployment compared to less cautious approaches.

• →AI-NNC enables coordinated autonomous control across previously siloed network domains—radio, optical, and core—critical for 6G infrastructure
• →Safety-first architecture routes AI commands through validated applications rather than direct device access, creating governance checkpoints for critical infrastructure
• →Open-source modular framework reduces vendor lock-in and standardizes interfaces, potentially accelerating industry adoption and interoperability
• →Unified testbed and dataset collection capabilities address practical research and deployment challenges beyond pure control functionality
• →Design patterns may establish reference models for safe autonomous control in other critical infrastructure sectors beyond telecommunications