CNCC panel debates agent safety as Ant Group runs 100,000 agents

On October 22, 2026, the CNCC 2026 conference held a dedicated forum in Chengdu titled "Security Issues of AI Agents", featuring five speakers: Yu Rui, Xu Ke, Hu Xia, Wei Tao, and Xiao Yajun, who addressed the topic from the perspectives of national standards development, university-based AI security research, and enterprise security defenses. According to a report published by Leiphone on November 4, 2026, figures disclosed at the forum showed that Ant Group had more than 100,000 agents running across its 30,000+ employees, while Huawei had deployed tens of thousands of agents internally across business lines including customer service, office work, operations, and R&D. The report also stated that in 2026, agents moved from concepts into production systems, gradually taking over work that humans previously performed hands-on in areas such as personal assistants, AI for Science, automated cyberattack and defense, and enterprise workflow orchestration. Yu Rui, deputy director of the CCF Computer Security Special Committee, described China's progress on agent identity management as summarized in three parts: national standards leading the way, industry-level refinement, and enterprise implementation. The security challenge highlighted in the material is that agents can act autonomously, migrate across platforms, and make chained decisions and call tools without human supervision, so traditional human identity authentication methods do not apply. Beyond the forum itself, the material notes that Gartner included agents in its Top 10 Strategic Technologies for 2027, that Non-Human Identity (NHI) became a new track at the RSAC conference, and that national standards, mandatory standards, Supreme Court judicial opinions, and regulations from three ministries were all issued in quick succession during 2026. The information available is current as of November 4, 2026, with no follow-up developments or conclusions reported by the parties involved.