Blab Tech
360 Security Unveils AI Suite to Counter Anthropics Mythos, Claims 3,432 Vulnerabilities Found
In a press‑release presented at the ISC.AI 2026 conference in Beijing, Chinese cybersecurity firm 360 Security Technology announced a new artificial‑intelligence (AI) platform called Yitian Tulong. The suite, named after the classic Chinese martial‑arts tale Heavenly Sword and Dragon Saber, contains two components: Tulongfeng, an automated vulnerability‑search tool, and Yitianzhen, a system designed to automate cyber‑defence and incident response.
The launch follows the United States’ recent decision to block Anthropic’s Claude Mythos model from foreign use. Mythos, a large‑language model that can identify software vulnerabilities, was highlighted by security researchers for its ability to discover thousands of zero‑day flaws across open‑source projects. Washington’s export‑control directive, issued in mid‑June, halted access to Mythos and its sibling Fable 5 for all non‑U.S. nationals.
360’s founder and CEO Zhou Hongyi said the move was a “direct, high‑profile answer” to the American system. According to the company’s statement, the Yitian Tulong suite is intended to prevent a “one‑way transparency” in which U.S. entities could scan global software while Chinese firms lacked comparable tools.
Tulongfeng is described as a “China’s version of Mythos.” 360 claims the tool has already flagged 3,432 potential vulnerabilities, of which 105 have been officially confirmed by Chinese authorities. Yitianzhen, on the other hand, focuses on automating defensive actions and incident response.
A key feature of 360’s approach is its strategy to sidestep U.S. chip export restrictions. The company said that domestic Chinese AI models typically lag by 20‑30 % in raw computing power because of limited access to advanced processors. Rather than waiting for hardware parity, 360 layers its existing security databases, automated tools, and human expertise over current AI models, a method the firm calls an “AI agent” strategy.
Zhou explained that the firm’s goal is “to organise a professional attack‑and‑defence team” rather than relying on the strongest chips. The company’s statement notes that this approach allows it to achieve capabilities it describes as “Mythos‑equivalent.”
The announcement comes amid a broader geopolitical contest over AI capabilities. Anthropic’s Mythos model, released earlier in 2026, was praised by security experts for its vulnerability‑finding performance but also raised concerns about offensive potential. The U.S. government’s export ban reflects those concerns, citing national‑security risks.
360’s Yitian Tulong is positioned as a domestic alternative that can be deployed without the restrictions that apply to U.S. models. The company said it will continue to develop the platform, with plans to expand its vulnerability‑discovery and defensive capabilities.
Industry observers note that the move signals China’s intent to maintain parity in AI‑driven cybersecurity tools. While the U.S. has tightened controls on advanced AI models, Chinese firms are accelerating domestic development to avoid dependence on foreign technology.
The U.S. export‑control directive remains in effect, and Anthropic has suspended worldwide access to Mythos and Fable 5 for all foreign customers. No official statement from Anthropic regarding the impact on its global customer base has been released.
360’s launch of Yitian Tulong is the first public demonstration of a Chinese AI system that claims to match the vulnerability‑discovery capabilities of Mythos. The company has not yet disclosed performance metrics beyond the number of vulnerabilities flagged.
As the U.S. and China continue to tighten controls on AI technology, the development of domestic AI security tools such as Yitian Tulong may influence the balance of cyber‑defence capabilities in the coming years. The international community will be watching to see how the new suite performs against the benchmark set by Mythos and whether it can be integrated into broader defensive frameworks.
The current status of Yitian Tulong is that it has been announced and demonstrated at ISC.AI 2026. 360 has not yet released a commercial product or detailed technical specifications. The U.S. export‑control directive remains in place, and Anthropic’s Mythos model is inaccessible to foreign users.
The broader implications of this development include potential shifts in the global AI‑cybersecurity landscape, increased competition for advanced AI tools, and a continued focus on national security concerns surrounding AI‑driven vulnerability discovery.
The launch follows the United States’ recent decision to block Anthropic’s Claude Mythos model from foreign use. Mythos, a large‑language model that can identify software vulnerabilities, was highlighted by security researchers for its ability to discover thousands of zero‑day flaws across open‑source projects. Washington’s export‑control directive, issued in mid‑June, halted access to Mythos and its sibling Fable 5 for all non‑U.S. nationals.
360’s founder and CEO Zhou Hongyi said the move was a “direct, high‑profile answer” to the American system. According to the company’s statement, the Yitian Tulong suite is intended to prevent a “one‑way transparency” in which U.S. entities could scan global software while Chinese firms lacked comparable tools.
Tulongfeng is described as a “China’s version of Mythos.” 360 claims the tool has already flagged 3,432 potential vulnerabilities, of which 105 have been officially confirmed by Chinese authorities. Yitianzhen, on the other hand, focuses on automating defensive actions and incident response.
A key feature of 360’s approach is its strategy to sidestep U.S. chip export restrictions. The company said that domestic Chinese AI models typically lag by 20‑30 % in raw computing power because of limited access to advanced processors. Rather than waiting for hardware parity, 360 layers its existing security databases, automated tools, and human expertise over current AI models, a method the firm calls an “AI agent” strategy.
Zhou explained that the firm’s goal is “to organise a professional attack‑and‑defence team” rather than relying on the strongest chips. The company’s statement notes that this approach allows it to achieve capabilities it describes as “Mythos‑equivalent.”
The announcement comes amid a broader geopolitical contest over AI capabilities. Anthropic’s Mythos model, released earlier in 2026, was praised by security experts for its vulnerability‑finding performance but also raised concerns about offensive potential. The U.S. government’s export ban reflects those concerns, citing national‑security risks.
360’s Yitian Tulong is positioned as a domestic alternative that can be deployed without the restrictions that apply to U.S. models. The company said it will continue to develop the platform, with plans to expand its vulnerability‑discovery and defensive capabilities.
Industry observers note that the move signals China’s intent to maintain parity in AI‑driven cybersecurity tools. While the U.S. has tightened controls on advanced AI models, Chinese firms are accelerating domestic development to avoid dependence on foreign technology.
The U.S. export‑control directive remains in effect, and Anthropic has suspended worldwide access to Mythos and Fable 5 for all foreign customers. No official statement from Anthropic regarding the impact on its global customer base has been released.
360’s launch of Yitian Tulong is the first public demonstration of a Chinese AI system that claims to match the vulnerability‑discovery capabilities of Mythos. The company has not yet disclosed performance metrics beyond the number of vulnerabilities flagged.
As the U.S. and China continue to tighten controls on AI technology, the development of domestic AI security tools such as Yitian Tulong may influence the balance of cyber‑defence capabilities in the coming years. The international community will be watching to see how the new suite performs against the benchmark set by Mythos and whether it can be integrated into broader defensive frameworks.
The current status of Yitian Tulong is that it has been announced and demonstrated at ISC.AI 2026. 360 has not yet released a commercial product or detailed technical specifications. The U.S. export‑control directive remains in place, and Anthropic’s Mythos model is inaccessible to foreign users.
The broader implications of this development include potential shifts in the global AI‑cybersecurity landscape, increased competition for advanced AI tools, and a continued focus on national security concerns surrounding AI‑driven vulnerability discovery.
