On June 24, at the ISC.AI 2026 conference in Beijing, 360 Security Technology founder Zhou Hongyi introduced a new AI tool for automatic vulnerability discovery — Tulong Feng. The development is positioned as a direct Chinese counterpart to Anthropic's Mythos 5 model, highlighting the growing competition in the field of AI-based cybersecurity.
Technical Details and Results
Tulong Feng is part of the Yitian Tulong package, which also includes the automated protection system Yitian Zhen. According to Zhou, the tool has already discovered 3,432 software vulnerabilities. Of these, 105 have been confirmed by Chinese regulators, and several have been added to the national database as high-criticality. It is important to note that these data have not yet received independent verification, leaving room for skepticism.
China's Strategic Approach
Unlike the American focus on "the strongest model, the most powerful computing, and the most advanced chips," 360 emphasizes AI agents. These agents combine language models, expert knowledge from security professionals, and vulnerability databases. This approach may be more effective under conditions of limited access to advanced chips, but requires careful integration and testing.
Global Race Context
The launch of Tulong Feng occurs against the backdrop of strict US restrictions. The government required Anthropic to suspend access to the Fable 5 and Mythos 5 models for foreign nationals, fearing technology leaks to China. In response, OpenAI not only launched the specialized GPT-5.5-Cyber model for vulnerability discovery and patching, but also restricted access to GPT-5.6 Sol, Terra, and Luna to a small group of trusted partners.
Earlier, Chinese startup Zhipu AI released its flagship model GLM-5.2 with a context window of 1 million tokens, also indicating active development of the AI ecosystem in China. However, incidents involving the distillation of Claude, in which Anthropic accused operators linked to Alibaba, show that tensions in this area are only increasing.
Expert Opinion
While Tulong Feng demonstrates impressive numbers, the key question remains its real effectiveness compared to Western counterparts. Without independent audits and transparent benchmarks, claims of 3,432 vulnerabilities could be either a breakthrough or a marketing move. In the context of the global AI security race, China is clearly betting on its own solutions, but whether it can catch up with the leaders without access to advanced chips — only time will tell.