CAC: The field of artificial intelligence faces five major security risks

Wang Lihong, Deputy Director of the Cybersecurity Coordination Bureau of the Cyberspace Administration of China, said at the press conference for the 2026 National Cybersecurity Publicity Week on September 1 that as artificial intelligence technology accelerates its penetration into various industries, related security risks are also rising, and there are currently five major challenges. These five challenges are: inherent technical vulnerabilities affect the stability and reliability of models; improvements in frontier model capabilities may change traditional cybersecurity defense methods, bringing the risk of loss of control; the rapid development of new application forms such as agents is causing security risks to evolve accordingly; misuse, abuse, and malicious use may impact social order and ethical boundaries; and there may be issues of technological hegemony and supply chain barriers in the global artificial intelligence industry chain. The press conference noted that artificial intelligence has, on the one hand, improved efficiency and expanded the scope of applications, but on the other hand, it has also placed new demands on cybersecurity governance, content management, research ethics, and industrial collaboration. Relevant authorities reminded that risk identification, assessment, and protection capabilities need to be continuously strengthened.