Zhou Hongyi, founder of 360 Group and chairman of the ISC.AI conference, has highlighted that the advancement of AI agent technology is fundamentally disrupting and challenging the traditional cybersecurity industry.
He noted that the most formidable threat often comes not from direct competitors but from outsiders entering the field with disruptive new technologies.
As an example, he cited the large language model Mythos from the American company Anthropic, which possesses the capability to autonomously discover, analyze, and construct attack software based on vulnerabilities. This model has been designated by the U.S. government as a "nuclear weapon of the AI era" and is subject to export controls. The emergence of such capabilities completely upends the traditional cybersecurity defense paradigm, which was built on the premise of "vulnerability scarcity."
Zhou outlined four major paradigm shifts brought by Mythos: first, a hundredfold increase in speed, moving network attack and defense from "human speed" to "machine speed"; second, a massive increase in the number of vulnerabilities discovered, as AI can conduct exhaustive, parallel searches 24/7; third, a drastic reduction in the cost of finding vulnerabilities, as AI primarily consumes computing power, which is relatively inexpensive; and fourth, the democratization of attack capabilities, allowing even those without programming knowledge to generate attack code using large models. These changes render the traditional defense logic of security companies—relying on "stacking hardware and buying software"—increasingly ineffective.
He pointed out that China's cybersecurity industry is facing a second crisis of "one-way transparency," shifting from a state of "the enemy is hidden, we are exposed" to one of "the enemy is faster, we are slower; the enemy has numbers, we are few." In response, he explicitly stated that China must develop its own equivalent of Mythos to establish a deterrent based on "discover first, verify first, patch first." He introduced 360's Chinese version, named "Tulong Feng," which already has the capability to mine vulnerabilities in open-source code, binary systems, and cutting-edge AI agents, transforming vulnerability discovery from a lottery-like chance into a systematic, assembly-line process.
Addressing the trend of increasingly frequent future cyberattacks, Zhou emphasized that merely finding and patching vulnerabilities is insufficient; automated defense is essential. He unveiled 360's automated cybersecurity defense system, "Yitian Zhen."
Finally, Zhou called for the establishment of China's own security collaboration system and officially launched the "Unbreakable Shield" initiative. This plan brings together several domestic tech giants, including Tongxin, Kylin, Hygon, and China Mobile Cloud, to prioritize offering the capabilities of "Tulong Feng" and "Yitian Zhen" to key domestic innovation units and critical infrastructure organizations for trial use. The aim is to collectively strengthen the nation's cybersecurity defenses before vulnerabilities can be exploited on a large scale.
Comments