BEIJING — The race to weaponize artificial intelligence for cybersecurity has entered a new phase. A Chinese firm now claims its bug-finding tool matches the capability of a leading American system. The implications are stark: software vulnerabilities discovered by AI can be exploited before anyone fixes them. 360 Security Technology announced its tool, Tulongfeng, at a cybersecurity conference in Beijing.
CEO Zhou Hongyi stated it is now comparable to Anthropic’s Mythos, the American system designed for the same purpose. Zhou described the tool as “this kind of powerful weapon that can alter the landscape of cyberwarfare,” according to reporting by the Wall Street Journal.
He added that such technology “can’t remain solely in American hands.” This is not about general AI prowess. China still lags behind on other fronts, including OpenAI products such as ChatGPT. The gap is closing in a specific, high-stakes corner: finding bugs in code.
Two Beijing-based firms, 360 Security Technology and Z.ai (also known as Zhipu AI), are leading this charge. Industry analysts have warned that widespread access to such tools could trigger a “bugmageddon” scenario — devastating cyberattacks on a global scale.
The competitive landscape is uneven across the broader AI sector. But in this narrow arena, the distance between Chinese and American capabilities has narrowed. The Trump administration has been imposing limits on public access to new Anthropic systems amid increased security concerns.
That move reflects a recognition of what these tools can do. President Donald Trump and Chinese President Xi Jinping have maintained what the US President terms a “good relationship.” The rivalry between the two superpowers remains tense. The Tulongfeng announcement arrives in that context.
Zhou’s language was blunt. He called the tool a powerful weapon.
He framed its development as a matter of national parity — not letting one country monopolize a technology that can alter the landscape of cyberwarfare. The Wall Street Journal’s Stephen M. Lepore reported the CEO’s comments from the conference. 360 Security Technology is a Beijing-based firm.
It is not a state-owned enterprise, but its announcement carries geopolitical weight. The company is effectively telling the world that China has caught up in one critical aspect of the AI arms race.
The question now is what comes next. If both sides possess comparable bug-finding tools, the advantage shifts from discovery to speed of patching. The window between finding a vulnerability and fixing it becomes the new battlefield.
Analysts have warned about bugmageddon for years. A scenario where automated systems find flaws faster than humans can patch them. Where cybercriminals or state actors exploit those flaws before defenses are updated.
The Tulongfeng announcement suggests that scenario is no longer theoretical. The broader AI sector remains lopsided.
American firms still lead in generative AI, large language models, and commercial applications. But cybersecurity is different. It is defensive and offensive at the same time.
Finding bugs is the first step in both protecting systems and attacking them. A tool that matches Mythos in capability gives its operator a significant advantage.
Zhou did not provide technical details about Tulongfeng’s performance. He did not release benchmark results. He made a claim at a conference.
The Wall Street Journal reported that claim. Independent verification has not been established. But the announcement itself signals intent.
It tells US policymakers that Chinese firms are investing heavily in this area. It tells the cybersecurity industry that the competitive landscape has shifted.
The Trump administration’s restrictions on Anthropic systems suggest Washington understands the stakes. Limiting public access to Mythos is a defensive move. It acknowledges that such tools are dangerous in the wrong hands.
But if Chinese firms have already matched that capability, the restrictions may be too late. The genie is out of the bottle — on both sides of the Pacific.
Sources
- Chinese AI GLM 5.2 Matches Mythos at Finding Bugs
- China matches Mythos on cyber bug-finding as the US restricts …
