Qihoo 360 unveils Tu Long Feng and Yi Tian Zhen to rival Mythos

Qihoo 360 Security Technology unveiled two AI tools at a conference in Beijing on Jun 28, 2026, saying one has already flagged 3,432 vulnerabilities. Founder Zhou Hongyi presented Tu Long Feng for automated vulnerability hunting and Yi Tian Zhen for automated cyber defense and compared the capabilities to a strategic deterrent.

What did Qihoo 360 unveil?

Qihoo 360 introduced two named systems: Tu Long Feng, which the company says automatically hunts for vulnerabilities, and Yi Tian Zhen, which automates cyber defense; Zhou Hongyi said Tu Long Feng has already flagged 3,432 vulnerabilities. The company positioned both tools as China’s answer to Anthropic’s Mythos and emphasized operational use alongside human security expertise.

How does 360 say the tools work?

Zhou said 360 uses an agent-based approach that pairs AI models with security expertise and automated tools, a setup intended to make up for what he called a 20 to 30 percent capability gap between top Chinese models and the most capable Western systems. The company’s pitch is that agents plus automation can multiply analyst capacity and perform simultaneous scanning and analysis rather than relying on a few experts.

Zhou framed that capability gap as urgent. He said, "China cannot wait until model capabilities have fully caught up before beginning vulnerability discovery. We cannot afford to wait." The company also argued this approach reduces reliance on single experts by deploying groups of automated agents.

How does Zhou compare these tools to U.S. systems and exports?

Zhou singled out Anthropic’s Mythos-class capabilities and likened advanced vulnerability models to "cyber-nuclear weapons of the AI era," arguing China needs an equivalent strategic deterrent so the capability is not monopolized by others. He cited the U.S. export ban on Anthropic’s Fable 5 as evidence that Washington seeks exclusive control over such strategic assets and warned against "one-sided transparency" in which the U.S. could scan Chinese systems while China remains blind.

The article also notes a separate estimate from Tsinghua professor Jie Tang, who said a Chinese "Mythos" class model would arrive before Q1 2027. The piece adds that without independent tests and benchmarks like those OpenAI and Anthropic provide, Zhou’s capability claims cannot be externally verified.

Why it matters

Qihoo 360’s presentation reframes automated vulnerability discovery as a strategic, not merely technical, issue: the company ties tool development to national deterrence and to parity with Western capabilities. If large-scale automated vulnerability hunting becomes operational, it changes how states and companies think about exposure, disclosure, and control of offensive and defensive cyber tools.

What to watch

Look for independent tests or benchmarks of Tu Long Feng and Yi Tian Zhen, and for any public disclosures that verify the claim that Tu Long Feng flagged 3,432 vulnerabilities. Also monitor developments toward a Chinese "Mythos" class model, which Jie Tang estimated could appear before Q1 2027.

Note: the article’s Chinese transcript was translated using AI and the excerpts were verified for accuracy using three additional AI systems.