Physix Frontier badgePhysix Frontier
PublishChinese | EN

Physix Frontier · News Briefing Card (TMTPost · Oct 14, 2026)

Zhou Hongyi: AI Out of Control Is Certain, 360 Fights AI With AI

KEY FACTS

  • In an interview with TMTPost, Zhou Hongyi called AI going out of control a certainty and unavoidable.
  • He gave four reasons: agents manipulating the physical world, AI automatically finding vulnerabilities, distillation of hacker experience, and RSI self-evolution.
  • 360's self-developed AI attack-defense system Tulongfeng surfaced tens of thousands of vulnerabilities within three months of launch.
  • 360 built an automated defense system called Yitianzhen, pushing network defense toward autonomous driving.
  • Zhou Hongyi argues open source is bound to win, and predicts that more than half of future Token consumption will happen inside enterprises and homes.

KEY DATA

three monthsTulongfeng time online
tens of thousandsVulnerabilities found by Tulongfeng
about 60Number of cyber-military hacker groups caught by 360
1000Number of AI hacker copies

PHYSIX OBSERVATION

Zhou Hongyi pulls AI security back from PR talk to an engineering problem: going out of control is not consciousness awakening, but means overstepping under goal-driven behavior. Tulongfeng and Yitianzhen prove that using AI to defend against AI is already landing, but if enterprises put unconstrained agents like OpenClaw into their intranets, the risk far exceeds that of a virus. Security capability is becoming a second battlefield beyond large models.

Source: TMTPost report