Cybersecurity buffs have a new nightmare scenario, and it has a name: Tulongfeng. The bug-finding tool from Beijing’s 360 Security Technology is, according to its CEO Zhou Hongyi, now on par with Anthropic’s Mythos system. Zhou dropped that comparison at a cybersecurity conference in Beijing.
The Wall Street Journal reported his blunt assessment: this is “this kind of powerful weapon that can alter the landscape of cyberwarfare.” His follow-up was just as direct. The tool, he said, “can’t remain solely in American hands.” The statement lands in a specific, tense corner of the U.S.-China tech rivalry.
Not the flashy consumer AI race dominated by OpenAI’s ChatGPT, where China still trails. The contest here is quieter, more dangerous. It is about finding software bugs before anyone else does.
Use an AI to spot a vulnerability in widely used code, and you own a window of time — before a patch exists — to break into systems, steal data, or shut down infrastructure. That window is the whole game.
Two Beijing firms are driving the push. 360 Security Technology and Z.ai, also called Zhipu AI, have closed the gap with American competitors specifically on bug-hunting, according to reporting from Stephen M. Lepore. That gap-closing matters.
Widespread access to these tools, industry analysts warn, could trigger a “bugmageddon” — a wave of devastating global cyberattacks launched by criminals who move faster than defenders can fix their code. President Donald Trump and Chinese President Xi Jinping maintain what Trump calls a “good relationship.” The rivalry beneath that diplomatic language, though, remains tense. The Trump administration has already been clamping down on public access to new Anthropic systems.
The logic is clear: if bug-finding AI is a weapon, you do not hand it out freely. But the weapon is no longer exclusive.
Chinese firms are building their own. This is not about matching every American AI achievement. It is about winning a specific, high-stakes race.
A bug found first is a bug exploited. A bug exploited is a network breached.
A network breached can be a power grid shut down, a hospital system frozen, a military communication channel tapped. The consequences scale fast. Zhou’s language — “powerful weapon” — is not hyperbole in this context.
It is engineering reality. The broader AI sector remains uneven. China still lags on consumer-facing products like ChatGPT.
That gap gets the headlines. But the gap that keeps security experts awake at night is the one narrowing in Beijing.
Tulongfeng and its counterparts represent a shift in what counts as winning the AI race. The prize is not a better chatbot. The prize is the ability to see the next exploit before your adversary does.
What comes next depends on how fast patches move versus how fast attackers move. The “bugmageddon” scenario depends on scale.
One hacker with one zero-day is a problem. Hundreds of hackers with AI tools churning out zero-days daily is a different order of threat. The global software ecosystem is vast and fragile.
A single widely used library, cracked open, can cascade through millions of devices. Zhou’s announcement at the Beijing conference is a signal. The technology is real.
The competition is real. And the old assumption that American firms hold a permanent lead in cyberweaponry is no longer safe.
The landscape just got more dangerous. No one is calling it a good relationship in the server rooms where these tools are deployed.
{“@context”: “https://schema.org”, “@type”: “NewsArticle”, “headline”: “Chinese Cybersecurity Firm 360 Says Its AI Tool Rivals Anthropic’s Mythos System”, “description”: “Cybersecurity buffs have a new nightmare scenario, and it has a name: Tulongfeng. The bug-finding tool from Beijing’s 360 Security Technology is, according to its…”, “datePublished”: “2026-06-28T11:19:00+00:00T09:30:00+00:00”, “author”: {“@type”: “Person”, “name”: “Li Wei”}, “publisher”: {“@type”: “Organization”, “name”: “InfoPulse Today”, “url”: “https://newsfeed.infopulsetoday.com/”}}
