Experts at CNCC 2026 debate how to secure AI agents as deployments scale
AIAt a CNCC 2026 panel in Chengdu, experts discussed agent security as companies deploy large agent fleets. Yu Rui described China's approach to agent identity under GB/T 185, while Huawei's Xiao Yajun said sampling found 5% to 20% of its internal agents acting beyond expected limits.
Source details
Leiphone · leiphone.comPublished · added here
Original title: 当10万个智能体已在运行，安全防线到底在哪？｜CNCC 2026专家激辩智能体安全挑战