Huawei Leads AI Firewall Standard Development, Debuts Security Framework

At HUAWEI CONNECT 2026, Wu Jiaxing, President of Huawei's Data Communication Product Line Security Domain, said in an interview that the company is taking an 'AI against AI' approach to security, shifting from rule-driven protection to AI-driven, intent-aware and autonomous defense.

Wu said the rapid adoption of large models and AI agents has expanded the attack surface, while AI tools have lowered the barrier for attackers and accelerated malware mutation. He noted that more than 100 million new viruses are generated each year in the AI era, and risks such as prompt injection, model poisoning and compute theft have become key bottlenecks for enterprise AI transformation.

To address these challenges, Huawei positions its AI Firewall (AIFW) as the next-generation firewall, and has co-led the development of AIFW technical standards with more than 20 organizations including CCIA. Wu stressed that AIFW is not a firewall with AI features but a fully new product generation. The company also introduced an AI computing center security solution and the Xinghe AI three-layer security fence.

The three-layer fence is designed to counter three major risks at AI computing centers – compute theft, prompt injection and model poisoning. The outer network layer embeds a mining detection model with a 95% detection rate to block malicious traffic in milliseconds; the middle layer is an Agent fence; and the inner host layer monitors process behavior in real time to prevent data theft and model tampering.

Wu said AI security will become increasingly intelligent and autonomous in the next three to five years, with firewalls evolving into intelligent edge hubs and the AI security market entering a phase of scaled growth.