Chinese cybersecurity firm Qihoo 360 Security Technology said it has developed AI-driven security systems capable of matching advanced frontier cyber capabilities, as the company escalates efforts to automate vulnerability discovery amid a fast-evolving security race shaped by emerging AI systems such as Anthropic’s “Mythos.”
Two core platforms under 360’s “Yitian Tulong” framework
At the ISC.AI 2026 Internet Security Conference in Beijing, founder Zhou Hongyi introduced two core systems under 360’s “Yitian Tulong” framework.
The first system, “Tulongfeng,” is designed to automatically discover software vulnerabilities at scale. The second, “Yitianzhen,” is an automated cyber defense system aimed at detecting and responding to threats with minimal human involvement.
Zhou said both systems represent a shift away from traditional, human-heavy security operations toward AI agents capable of continuously scanning, analyzing and reacting to cyber risks across complex digital environments.
Zhou cites “Mythos” as indicator of AI progress
Zhou repeatedly referenced Anthropic’s “Mythos,” framing it as a key example of how AI systems are reshaping cybersecurity dynamics.
He warned that such technologies could dramatically compress the time needed to discover and exploit software vulnerabilities, from months or years down to just hours.
In his view, this shift is turning cybersecurity into a “machine-speed” contest, where both attackers and defenders increasingly rely on AI systems rather than human teams.
China moves to link security and tech supply chain
Alongside the product rollout, Qihoo 360 announced the “Panshi Zhidun” (Rock Shield) initiative, bringing together semiconductor, operating system, cloud computing, database and cybersecurity firms, including Phytium and Kylin, to deepen coordination across China’s technology ecosystem.
Zhou said the goal is to build a more unified security structure across critical infrastructure, arguing that AI-driven threats require coordinated defenses that span the entire technology stack rather than isolated solutions.
Rise of automated cyber defense models
Zhou warned that AI-powered vulnerability discovery will likely increase both the number of exploitable flaws in global software and the speed at which they can be found, while also lowering the cost of discovery, forcing cybersecurity into a continuous, always-on operating model.
To respond, 360 built a multilayer, multi-agent AI system instead of relying on standalone large models, Zhou said, combining cybersecurity expertise, tooling and automated workflows into coordinated units capable of operating in real time.
Tulongfeng flags thousands of flaws in use
360’s “Tulongfeng” system has already identified thousands of vulnerabilities across open-source software, binary code and emerging AI platforms, with some findings confirmed by regulators and major technology vendors, Zhou said.
He also previewed “Yitianzhen,” an automated defense platform designed to simulate attacks, prioritize critical assets and respond to threats autonomously with minimal human involvement.
The system is positioned as a step toward fully automated cybersecurity operations, where human intervention becomes increasingly limited.
Defensive strategies evolve under AI pressure
The ISC.AI 2026 presentation highlighted how capable AI systems like Mythos are shaping cybersecurity thinking in China, as firms race to build defenses capable of operating at machine speed in an increasingly automated and contested digital landscape.
