Chinese Firm Develops AI Cybersecurity Tool Rivaling US Mythos System

360Security Technology, a Chinese cybersecurity firm, announced the development of a local alternative to the Mythos system created by the American company Anthropic. This move highlights the intensifying competition between Beijing and Washington in AI tools designed for cybersecurity offense and defense.

The announcement was made at the ISC.AI 2026 cybersecurity conference in Beijing, where the company described Mythos as a "strategic capability that China cannot remain deprived of," especially following recent US export restrictions citing national security concerns, according to Reuters.

Anthropic introduced Mythos in April as a system capable of automatically detecting software vulnerabilities. However, cybersecurity experts warned that such capabilities might equip cyber attackers with more powerful tools for widespread cyberattacks.

Earlier in June, the US government ordered Anthropic to suspend exports of a less powerful version of Mythos to all foreign destinations, including international users, based on national security considerations.

At the ISC.AI 2026 conference, 360Security Technology's founder, Zhou Hongyi, revealed two new AI-powered tools named Yitian and Tulong. These names are drawn from a classic Chinese martial arts novel, meaning "Heavenly Sword" and "Dragon Sword."

Zhou explained that Tulongfeng, the first tool, is designed for automatic detection of software vulnerabilities and described it as "the Chinese version of Mythos." The second tool, Yitianzhen, aims to automate cybersecurity defense and incident response processes.

He stated, "Such powerful weapons capable of shifting the balance of cyber offense and defense should not be possessed by only one party," considering AI-based vulnerability detection technologies as strategic national assets that can protect critical infrastructure while enabling offensive advantages in cyberspace.

This development occurs amid a long history of mutual accusations between the US and China regarding cyber intrusions targeting each other's critical infrastructure.

The announcement by 360Security Technology represents the most prominent Chinese response so far to the Mythos system, which has raised widespread concerns in the US, other capitals, and within the cybersecurity sector due to its ability to identify vulnerabilities in sensitive systems.

Anthropic reported in April that its Mythos Preview version successfully detected thousands of significant vulnerabilities in operating systems, web browsers, and other software.

Zhou Hongyi expressed concern that China faces a risk of what he called "one-sided transparency" if American institutions use models similar to Mythos to examine software and critical systems while Chinese companies are denied similar capabilities.

The Chinese company stated that Tulongfeng has so far identified 3,432 software vulnerabilities, including 105 that Chinese authorities have officially verified. Reuters was unable to independently verify these figures.

Zhou added that his company does not aim to replicate the American approach, which he described as relying on "the strongest model, the largest computing power, and the best chips."

He noted that since 2022, US restrictions on exporting advanced chips to China have prevented Chinese models from fully catching up with their American competitors, including Anthropic, although the gap has narrowed over the past year.

The US justifies these restrictions by stating they prevent the Chinese military from enhancing its capabilities through AI.

Zhou said, "Objectively, local models still lag behind their foreign counterparts by about 20% to 30% in core capabilities. But China cannot wait for this development to be complete before starting to use AI for vulnerability detection because the cost of waiting would be high."

He explained that the company chose a different path based on the concept of AI Agents by integrating language models with cybersecurity expertise, vulnerability databases, and automation tools. Zhou asserted that 360Security Technology is the only company that has successfully implemented this approach, granting Tulongfeng capabilities comparable to Mythos.

Zhou compared the two approaches by saying, "If Mythos represents a high-performance electronic chip, what we are building is a fully integrated device that operates steadily 24/7 and reduces errors. If the American approach focuses on creating a genius hacker, our approach organizes a professional, integrated team for attack and defense."

In an effort to manage an escalating crisis with the White House, Anthropic sent several senior employees to Washington to hold direct meetings with US administration officials.

Last year, Anthropic revealed that hackers exploited vulnerabilities in its AI assistant Claude to launch attacks targeting approximately 30 global organizations.

A joint study by IBM and Palo Alto Networks involving 1,000 executives found that 67% of participants reported their organizations experienced AI-powered cyberattacks in the past 12 months.

Zhou Hongyi is recognized as one of China's leading internet pioneers and a prominent figure in technical discussions. He founded 360Security Technology, which became one of the country's most well-known cybersecurity companies through its antivirus software before expanding into cybersecurity solutions for enterprises and government entities.