Chinese cybersecurity firm 360 Security Technology has released two AI models designed to find software flaws and automate cyber defense. The system, called Yitian Tulong, was presented at the ISC.AI 2026 conference in Beijing.
The setup has two parts: Tulongfeng hunts for vulnerabilities, while Yitianzhen handles automated defense and incident response. According to the company, Tulongfeng already found 3,432 software flaws, with 105 confirmed by the Chinese government.
Founder Zhou Hongyi acknowledged a 20-30% capability gap compared to US models but argued that building a professional attack-and-defense team matters more than relying on a single advanced AI. The company’s strategy focuses on operational capability rather than chasing the most advanced individual model.
This contrasts with the US approach, which Zhou characterized as relying on a “genius hacker” model. 360’s pitch is that a stable, continuously operating system delivers more practical value than one breakthrough model.
The announcement adds another data point to the growing intersection of AI and cybersecurity — a space where Chinese companies are investing heavily to close the gap with Western counterparts.
