360 founder urges China to build its own Mythos-class cyber AI
Synopsis
360 Security Technology founder Zhou Hongyi declared Anthropic's Mythos — which accelerates vulnerability discovery 100x — a 'cyber nuclear weapon' and warned that China's exclusion from the Project Glasswing alliance leaves it strategically exposed, demanding a domestic equivalent for deterrence.
Key Takeaways
Zhou Hongyi , founder of 360 Security Technology , called for China to build a domestic equivalent to Anthropic 's Mythos AI model at a cybersecurity conference in Beijing on 25 June 2026 .
Mythos , launched in April 2026 , reportedly accelerates software vulnerability discovery a hundredfold while slashing associated costs.
Anthropic 's Project Glasswing alliance gave more than 40 organisations access to Mythos Preview for cyber defence — China was excluded.
Zhou described Mythos as a 'cyber nuclear weapon' and argued a Chinese equivalent would provide 'reciprocal strategic deterrence capability.' The 56-year-old founder warned that a tool capable of reshaping cyber warfare 'cannot be held solely in the hands of others.'
Zhou Hongyi, founder of 360 Security Technology, warned on Wednesday, 25 June 2026 at a cybersecurity conference in Beijing that China must develop a domestic equivalent to Anthropic's Mythos model — or risk falling into a permanent strategic blind spot in AI-driven cyber warfare.
Why Mythos alarms Beijing's security establishment
Mythos, released in April 2026, can autonomously identify software vulnerabilities and has reportedly accelerated vulnerability discovery a hundredfold while dramatically cutting costs. Zhou described this as a 'terrifying change' that had effectively 'democratised' cyberattacks, likening the model to a 'cyber nuclear weapon' in terms of its strategic weight.The Project Glasswing exclusion
Anthropic launched the Project Glasswing alliance in April 2026, granting more than 40 organisations access to Mythos Preview to bolster their cyber defences. China was excluded from the initiative entirely. Zhou called the move 'astute' on the part of the US, noting the asymmetry it creates: 'This means US organisations can use Mythos to scan your vulnerabilities, but you don't even have the qualification to catch a glimpse of Mythos,' the 56-year-old founder said.The case for strategic deterrence
A domestic Mythos-class model, Zhou argued, would give China a 'reciprocal strategic deterrence capability' — drawing an explicit analogy to nuclear-armed states that deter one another through mutually assured capability. He characterised such a tool as a 'game-changing weapon in cyber warfare' that 'cannot be held solely in the hands of others.'Competitive backdrop
Zhou's remarks arrive as China's AI sector races to close gaps with frontier Western models across multiple domains. The cybersecurity vertical is emerging as a particularly sensitive front, given that AI-powered vulnerability scanners blur the line between defensive tools and offensive weapons. The exclusion of Chinese entities from Project Glasswing mirrors broader patterns of technology decoupling, including export controls on advanced semiconductors and restrictions on cloud AI access.What's next
Whether China's cybersecurity industry can marshal the talent, compute, and specialised training data to produce a credible Mythos rival remains an open question. The call from a high-profile founder like Zhou Hongyi is likely to intensify state-level attention and funding toward AI-native security research — and the global cybersecurity community will be watching closely to see how quickly that gap narrows.Point of View
A rhetorical playbook that has historically unlocked significant government funding in China. What mainstream coverage underplays is that Project Glasswing's exclusion of Chinese entities is as much a data and intelligence advantage as a tooling one: the 40+ allied organisations feeding defensive telemetry back to Anthropic are collectively training a security feedback loop that China cannot access. This mirrors the semiconductor export-control dynamic, where the real damage is not just the missing chip but the missing ecosystem of tooling, talent, and iterative improvement. The deeper risk for China is that by the time a domestic Mythos-class model arrives, the US-aligned alliance will have iterated through multiple generations of the same capability.
NationPress
25 Jun 2026
Frequently Asked Questions
What is Anthropic's Mythos model and why is it significant?
Mythos is an AI model released by Anthropic in April 2026 that can autonomously identify software vulnerabilities, reportedly accelerating their discovery a hundredfold while cutting costs. Its significance lies in effectively democratising cyberattacks, lowering the barrier for sophisticated vulnerability exploitation.
Why is China excluded from Project Glasswing?
Project Glasswing is a US -led alliance through which Anthropic granted more than 40 organisations access to Mythos Preview for cyber defence purposes. China was not included, a move Zhou Hongyi described as strategically deliberate, leaving Chinese entities unable to use or even evaluate the model.
What did Zhou Hongyi say about Mythos at the Beijing conference?
Zhou Hongyi called Mythos a 'cyber nuclear weapon' and a 'game-changing weapon in cyber warfare' at a cybersecurity conference in Beijing on 25 June 2026 . He argued that such a tool 'cannot be held solely in the hands of others' and that China needs its own equivalent for strategic deterrence.
What would a Chinese Mythos equivalent achieve?
According to Zhou Hongyi , a domestic Mythos -class AI would give China a 'reciprocal strategic deterrence capability,' analogous to how nuclear-armed states deter one another. It would allow Chinese organisations to scan their own vulnerabilities using comparable AI-powered tools rather than remaining exposed to one-sided scanning by adversaries.
How does this fit into the broader US-China tech rivalry?
The exclusion of China from Project Glasswing is the latest front in a widening technology decoupling that already encompasses semiconductor export controls and cloud AI restrictions. Zhou 's call mirrors earlier pushes for domestic chip and large language model development, signalling that AI-powered cybersecurity is now a priority battleground in the US- China tech competition.
