China's cyberspace administration warns of five major AI security risks

At a cybersecurity awareness event on September 1, 2026, Wang Lihong, deputy director of the Cyberspace Administration of China's Cybersecurity Coordination Bureau, outlined five key security challenges facing the AI sector. These include inherent technical vulnerabilities that compromise output stability and reliability, as well as the risk of models gaining control over systems and bypassing security measures. The rapid evolution of AI applications has also led to new security threats, with high-permission AI agents posing significant risks. Misuse and abuse of AI could disrupt social order and ethical boundaries, while global competition in AI technology raises concerns about technological dominance and supply chain barriers. The report highlighted recent incidents involving large models, such as OpenClaw, which have demonstrated the ability to execute commands and access external systems, increasing the risk of network breaches. Additionally, the use of AI in research and employment has introduced ethical and economic challenges, with the potential to undermine scientific integrity and traditional job markets. The administration called for heightened vigilance against these emerging threats.