Wuzhen Summit Explores AI Governance for Agents and Robots: Trust, Permission Isolation, and Dynamic Defense

At the 2026 World Internet Conference Wuzhen, industry leaders discussed AI governance for agents and robots, emphasizing that trust, authorized access, and physical safety are prerequisites for real-world deployment. Experts proposed isolating agent actions from user credentials via trusted gateways and shifting security from static compliance to dynamic, real-time risk assessment. The discussion highlighted the need for clear responsibility boundaries among device makers, model providers, and network services to ensure human oversight as AI systems gain autonomous capabilities.