Global Public Security Cooperation Forum Held; Zhou Hongyi Proposes a New Path for AI Security Governance: "Using Models to Govern Models"

Recently, the 2026 Global Public Security Cooperation Forum opened in Lianyungang, Jiangsu. Zhou Hongyi, founder of 360 Group, emphasized at the conference that AI has changed the landscape of cyber offense and defense, and proposed a new automated offense-defense system centered on "using models to govern models." He mentioned that in the face of the changing offense-defense dynamics in the AI era, 360 was the first to propose "using models to govern models," building an AI security offense-defense system covering vulnerability discovery, automated defense, and agent governance: "China's version of Mythos," Tulongfeng, has discovered more than 10,000 vulnerabilities cumulatively since its release in June, uncovering long-dormant Windows and Office zero-day vulnerabilities, as well as high-value vulnerabilities in AI platforms such as OpenClaw and Codex.