Zhou Hongyi, founder and chairman of Chinese cybersecurity and internet company 360 (Qihoo 360), spent the summer of 2026 making two arguments that, taken together, sum up how he has positioned the company for the current phase of the global AI race: that China is dangerously exposed on AI-powered cybersecurity, and that 360 itself already has a working answer for AI-powered productivity. Both arguments came with real, dated announcements attached rather than abstract commentary.
“A Cyber Nuclear Weapon” China Can’t See
Speaking at a cybersecurity conference in Beijing on June 25, 2026, Zhou singled out Anthropic’s most advanced model, Mythos, and its role in a disclosed AI-powered vulnerability-discovery effort, describing the technology in blunt terms as a “cyber nuclear weapon” that had “democratised” cyberattacks by putting sophisticated offensive capability within reach of far less skilled actors. His central complaint was one of access, not just capability: because China had been excluded from the disclosed program, he said, US-aligned organizations could use the model to scan for vulnerabilities in their own systems while Chinese entities had “no qualification to catch a glimpse” of the same tool. Zhou argued a domestic, Mythos-class model was the only way to give China what he called “reciprocal strategic deterrence capability” — explicitly comparing the dynamic to the logic of mutual deterrence between nuclear-armed states, and framing the gap as something that “cannot be held solely in the hands of others.”
Then He Shipped the Product Side of the Same Argument
Roughly six weeks later, on August 3, 2026, Zhou unveiled Nano Work, 360’s own enterprise AI agent platform, at a launch event in Beijing — a product aimed squarely at the productivity side of the AI race he had just spent June warning China was losing on the security side. “When enterprises embrace AI, the boss must be the first to use it,” Zhou said, framing Nano Work as a tool built for business owners, startup founders, sole proprietors, and rank-and-file staff rather than only large enterprise IT departments. He was explicit that the product was meant to be more than a thin interface layered on top of an existing model: “Nano Work is not a simple wrapper for any single agent,” he said, describing it instead as software that “turns advanced capabilities into ready-to-use productivity” by combining multi-agent processing, integration across multiple underlying AI models, cloud-based operation, several distinct operating modes, and a set of specialized expert agents built for specific business functions.
Five Months of Testing Before the Public Launch
360 backed the launch with specific development numbers rather than only a demo: the company said it ran a five-month internal testing phase ahead of release, deploying the agent platform across more than 1,000 real business scenarios, collecting 56,000 pieces of user feedback, and shipping 166 iterative updates before the public unveiling. To drive early adoption, 360 offered trial users 100 million complimentary AI tokens along with training and support services, targeting an initial rollout across 1,000 small businesses — a deliberately hands-on go-to-market approach for a company better known internationally for free consumer antivirus software than for enterprise AI tooling.
A Founder Who Has Long Built a Public Persona Around Security Threats
Zhou co-founded Qihoo 360 in June 2005 and built its early reputation, and much of its Chinese user base, around free antivirus and internet-security software distributed at a scale few Western security vendors ever matched domestically; the company later listed on the New York Stock Exchange in 2011. That security-first identity has carried directly into how Zhou has positioned 360’s AI strategy: rather than competing head-on with China’s largest general-purpose chatbot makers on consumer chat quality, he has repeatedly framed AI through the lens of national and enterprise security exposure, a positioning that both the June cybersecurity remarks and the August Nano Work launch reinforce from opposite ends — one warning about a capability gap, the other claiming to have already closed a related one commercially.
Part of a Broader Domestic Push on AI-Native Security
Zhou’s June remarks, made as US semiconductor export controls and cloud AI restrictions continued to shape the broader US-China technology relationship, are widely expected to add pressure for increased Chinese government funding into AI-native security research specifically — a narrower, more specialized category than the general large-language-model race that has dominated most public attention on China’s AI sector. His argument that offensive AI capability requires a defensive AI counterweight, rather than purely regulatory or diplomatic responses, reflects a security-industry framing distinct from how most consumer-facing Chinese AI labs, focused on chat assistants and agent products, have typically discussed the stakes of the broader AI competition.
What This Means for Philippine Founders
Zhou’s Nano Work launch is a useful data point for Philippine founders sizing up enterprise AI adoption: 360’s own account of the product — five months of testing across 1,000 real business scenarios and 56,000 feedback points before shipping — is a concrete reminder that a credible AI agent platform requires substantially more validation work than wrapping a single model behind a chat interface, a bar worth measuring any locally-built “AI agent” product against before pitching it to enterprise buyers. His cybersecurity warning is also directly relevant for Philippine companies and government agencies dependent on both US and Chinese-origin software and cloud infrastructure: as AI-powered vulnerability discovery tools become more capable and more geopolitically restricted in who can access them, Filipino IT and security teams should expect the gap between well-resourced organizations with access to frontier AI security tooling and those without it to widen, regardless of which country’s technology stack a given Philippine organization ultimately depends on.
Share this article