BEIJING, June 24 (Parliament Politics Magazine) – Chinese cybersecurity firm 360 Security Technology announced on Wednesday the development of a domestic alternative to Anthropic’s Mythos.
The company presented the tools at the ISC.AI 2026 conference in Beijing, framing them as a necessary strategic asset for national cyber security.
Introducing Yitian Tulong
360 founder Zhou Hongyi unveiled two AI-powered security tools under the collective banner Yitian Tulong. The name is derived from a classic Chinese martial arts novel, translating to Heavenly Sword and Dragon Saber.
According to Zhou, the suite includes a tool named Tulongfeng, which is designed to automatically discover software vulnerabilities. He explicitly identified this system as China’s version of Mythos. The second tool, Yitianzhen, was built to automate cyber defence processes and manage incident responses.
The company stated that Tulongfeng has already identified 3,432 software vulnerabilities, with 105 of those findings confirmed by Chinese authorities. These claims could not be independently verified.
The development of Mythos by Anthropic, which was previewed in April, has caused significant concern across the global cybersecurity industry. The system is designed to detect software vulnerabilities, but experts have warned that such capabilities could potentially be used to accelerate cyberattacks.
The United States government recently ordered Anthropic to suspend exports of a less powerful version of the programme, citing national security risks. Zhou argued that China faced the danger of one-way transparency if U.S. entities could use models like Mythos to scan systems while Chinese firms were restricted from accessing comparable technology.
“This kind of powerful weapon that can change the landscape of cyber offence and defence cannot be held only by others,” Zhou said.
Shifting Engineering Approaches
Zhou acknowledged that U.S. export controls on advanced AI chips, in place since 2022, have hindered the progress of Chinese models compared to American rivals. He noted that domestic models still face a 20% to 30% gap in base capabilities.
However, 360 has opted for a different engineering path to circumvent hardware limitations. Instead of relying solely on massive, singular large language models, the company is using an agent-based approach. This strategy combines models with existing security expertise, large-scale vulnerability databases, and automated testing tools.
Zhou argued that while the U.S. approach focuses on training a high-end genius hacker model, 360 is focused on organizing a professional team of automated defense and attack systems. He claimed this method allows the company to deploy Mythos-equivalent capabilities while working within the constraints of current hardware availability.
The tension between China and the U.S. regarding cyber operations remains a significant factor in the development of these technologies. Both nations have frequently accused the other of conducting offensive cyber operations against critical infrastructure.
For 360, the launch of Yitian Tulong represents a move to secure a level playing field. As a member of China’s top political advisory body, Zhou framed the development of vulnerability-finding AI as a strategic asset. He emphasized that China could not afford to wait for domestic base models to catch up to their American counterparts before developing systems capable of identifying security gaps.
The competitive nature of this technology is highlighted by the scale of vulnerability detection reported by major players. Anthropic previously stated that the Mythos Preview found thousands of major vulnerabilities in operating systems and web browsers. As these tools continue to evolve, the global focus remains on the balance between automated defense and the potential for offensive exploitation in critical infrastructure.
