At the opening of the ISC.AI 2026 conference, the founder of 360 Group, Zhou Hongyi, officially launched the company's two core AI security capabilities, named "Yitian Tulong." These consist of an automated vulnerability discovery intelligent agent called "Tulong Feng" and an automated network security defense system named "Yitian Zhen."
Concurrently, 360 initiated the "Rock Shield" security collaboration plan, partnering with industry allies in the domestic IT ecosystem, security, cloud computing, large language models, and computing power.
Zhou Hongyi argued that the global shockwaves caused by the AI tool 'Mythos' stem from its ability to autonomously discover and analyze vulnerabilities, even constructing cyber weapons. He likened it to a "cyber nuclear weapon" for the AI age, creating a new form of strategic deterrence.
The AI Threat and a New Era of Vulnerability
In his address, Zhou issued a stark warning: Mythos represents a paradigm-shifting blow to the traditional security industry, rendering old defenses obsolete. He cautioned that without new countermeasures, China's cybersecurity faces a "second era of one-way transparency."
Zhou reflected that 360 spent two decades confronting the first such challenge, where overseas APT groups operated with long-term stealth, creating a dynamic of "the enemy in the dark, us in the light." The advent of tools like Mythos, however, has shifted the balance to "the enemy fast, us slow; the enemy many, us few." Attackers can now deploy swarms of hacker agents simultaneously, while defense still relies on manual analysis by a handful of experts.
"When AI can rapidly and batch-discover vulnerabilities, our systems appear like sieves in the eyes of attackers, full of potential entry points," Zhou stated.
The Imperative for a Domestic Solution
Zhou predicted a high-incidence period for cyberattacks on China's critical infrastructure and vital industries in the coming years. He emphasized that China's cybersecurity sector must possess its own version of Mythos. The goal is to see and patch one's own vulnerabilities first, rather than reacting passively after others discover them. "Only by holding the cards in our own hand can we have confidence," he said.
He further noted that China should not simply replicate the foreign approach of relying solely on brute computing power and model scale. Instead, it should leverage its engineering strengths to pursue an intelligent agent pathway. This involves organizing the capabilities of large models, the experience of security experts, and vulnerability knowledge bases into a collaborative system of intelligent agents.
Introducing the 'Yitian Tulong' Capabilities
This philosophy underpins the newly launched "Yitian Tulong" AI security capabilities. The "Tulong Feng" agent is positioned as a Chinese counterpart to Mythos, focused on proactive vulnerability discovery. It has reportedly uncovered 3,432 vulnerabilities to date, with 105 confirmed by regulators and several classified as high-risk by the national vulnerability database. The "Yitian Zhen" system is designed for security operations and automated defense, capable of autonomously planning tasks, analyzing alerts, and coordinating responses within real network environments.
Comments