Qihoo 360 Security Technology has launched two new AI systems designed to find security flaws and defend networks, positioning them as a direct counter to Anthropic‘s Mythos.
Zhou Hongyi, the founder of Qihoo 360, presented these tools at a conference in Beijing. The first, Tu Long Feng, hunts for vulnerabilities. The second, Yi Tian Zhen, automates cyber defence. Zhou stated that Tu Long Feng has already identified 3,432 vulnerabilities.
Zhou claims the gap between leading Chinese AI models and their Western counterparts sits at 20 to 30 percent. To close this distance, Qihoo 360 employs an agent-based approach. This method pairs large models with specific security knowledge and automated utilities. “China cannot wait until model capabilities have fully caught up before beginning vulnerability discovery. We cannot afford to wait,” Zhou said.
Jie Tang, a professor at Tsinghua University who founded Z.ai and released the GLM-5.2 model, estimates a Chinese equivalent to Mythos will arrive before the first quarter of 2027.
Zhou frames AI vulnerability hunting as a nuclear deterrence problem
Without independent tests or benchmarks from companies like OpenAI and Anthropic, verifying Zhou’s claims is impossible. The rhetoric itself, however, demands attention.
Zhou compared Mythos’s capacity to find flaws and construct attack chains independently to “cyber-nuclear weapons of the AI era.” He argued that the reason there has never been a real nuclear war is that both sides possessed the weapons and deterred each other. He believes the same logic applies to cybersecurity.
“China needs an ‘equivalent strategic deterrent capability,'” Zhou said. “A weapon that can shift the entire balance of attack and defense must not be left solely in the hands of others.”
He warned against “one-sided transparency.” Zhou argued the United States could use Mythos to scan Chinese systems for weaknesses while China remains unaware. “With Mythos, it becomes a situation where the enemy is fast and we are slow, the enemy is many and we are few,” he said. “You are still relying on a few security experts for analysis, while the other side has already replicated a group of hacker agents to work simultaneously.”
Zhou pointed to export bans on Anthropic’s Fable 5 as proof of this dynamic. He described Fable 5 as the “civilian, neutered version of Mythos.” If someone jailbroke the software, the entire world could access Mythos-level capabilities. “This is what the U.S. government finds most intolerable. It must ensure that it alone possesses this capability, forming an absolute monopoly over this strategic asset,” Zhou said.
The United States government uses similar national security arguments to justify export controls on chips and Mythos. Both sides have spent years trading accusations over cyberattacks on critical infrastructure. It is the Cold War playbook, just applied to artificial intelligence. One side builds a capability; the other claims it cannot afford to fall behind.
The names of Qihoo 360’s tools carry a hint of war romanticism. They likely reference a classic Chinese martial arts novel where legendary weapons grant their owners supremacy.
Note: The Chinese transcript was translated using AI, and the excerpts were verified for accuracy using three additional AI systems.
What it means
For security teams in China, the message is clear: wait for the technology to mature before acting is not an option. The strategy shifts from waiting for a perfect model to building systems that combine imperfect AI with human expertise and automation immediately. This approach aims to create a defensive shield capable of matching the offensive speed of Western AI tools, framing the competition as a matter of strategic survival rather than just technical advancement.
