China cybersecurity company 360 (三六零) said it developed its own AI system to counter Mythos, a security vulnerability detection model by U.S. artificial intelligence (AI) corporations Anthropic.
According to Reuters and China Securities Journal on the 25th, 360 founder Zhou Hongyi on the previous day unveiled the vulnerability auto-detection AI "Tulongfeng" (圖龍鋒) and the automated defense system "Yitianzhen" (儀天陣) at the 14th Internet Security Conference held in Beijing. Zhou introduced Tulongfeng as the "China version of Mythos," saying it can automatically find software vulnerabilities, and explained that Yitianzhen is a system that automates cyber defense and incident response.
360 said Tulongfeng has so far found 3,432 software vulnerabilities, of which 105 were confirmed by Chinese authorities. It also claimed that through an "AI agent approach" that combines security expertise, vulnerability databases, and automation tools in the AI model, Tulongfeng has acquired capabilities equivalent to Mythos.
However, Reuters said it could not independently verify these claims. Mythos is a model that Anthropic released in Apr., known to be capable of finding large-scale vulnerabilities in operating systems and web browsers. The U.S. administration recently announced export control measures restricting access by foreign nationals to Mythos, citing national security concerns.
Zhou said Mythos not only finds software vulnerabilities but can also analyze them to create attack methods, calling it a "cyber nuclear weapon of the AI era." He argued that if China does not acquire capabilities like Mythos, China could face a situation where, even if the United States analyzes China's core systems, China cannot respond.
Zhou said, "China's major infrastructure and key industries could be increasingly exposed to AI-based cyberattacks in the future, so we must find and fix vulnerabilities first," stressing the need to build independent AI security capabilities.