2026年10月16日，中国商务部与科技部提出扩大《禁止出口限制出口技术目录》，新增约40项技术，涵盖人工智能、生物医药与光伏。文件将AI技术分为禁止出口的敏感AI技术（涉及跨境数据传输的深度伪造检测、网络攻击技术、基于"核心数据"训练的基础模型、面向关键基础设施与国防的智能体等）和列入限制清单的AI技术（覆盖基础模型、AI智能体、AI芯片与算力集群），并以累计算力、参数规模或全球领先水平等条件界定门槛，对已公开发布的开源模型予以豁免。野村认为，短期内对中国基础模型行业影响可控，已发布的开放权重模型可获豁免；但独立AI实验室若拟将未来模型部署于海外云基础设施服务海外用户，则面临较高门槛，而已在海外运营数据中心的阿里、字节跳动和腾讯更具优势。报告还指出，核心数据与重要数据的界定、评级机构标准等仍存模糊空间，监管执行有调整余地。
On 16 October, China's Ministry of Commerce and Ministry of Science and Technology proposed expanding the Catalogue of Technologies Prohibited and Restricted from Exports to cover 40 additional technologies spanning AI, biomedicine and photovoltaics, with consultations running through 15 November. The proposal places AI technologies into two categories: sensitive AI technologies prohibited from export, and AI technologies subject to export controls. Prohibitions focus on sensitive data and security-related technologies, including advanced deep fake detection for cross-border data transfers, certain AI-enabled cyberattack technologies, foundation models trained on "core data", and high-performance AI agents for critical infrastructure, defence or core government functions. The restricted list covers foundation models, AI agents, AI chips and AI computing clusters, with thresholds based on accumulated computation, parameter count or globally leading capabilities. Open-source models alrea