人工智能安全治理框架2.0
AI智能总结
AISafetyGovernanceFramework2.0 全国⽹络安全标准化技术委员会NationalTechnicalCommittee260onCybersecurityofSAC 国家计算机⽹络应急技术处理协调中⼼NationalComputerNetworkEmergencyResponseTechnicalTeam/CoordinationCenterofChina 2025年9⽉ 目 录 1.⼈⼯智能安全治理原则......................................................................2. 2.⼈⼯智能安全治理框架构成................................................................3 3.⼈⼯智能安全⻛险分类......................................................................4. 3.1⼈⼯智能技术内⽣安全⻛险.............................................................3.2⼈⼯智能技术应⽤安全⻛险.............................................................3.3⼈⼯智能应⽤衍⽣安全⻛险.............................................................468 4.技术应对措施.....................................................................................9 4.1技术内⽣安全⻛险的应对措施.........................................................4.2技术应⽤安全⻛险的应对措施.........................................................4.3应⽤衍⽣安全⻛险的应对措施.........................................................91112 5.综合治理措施....................................................................................13 5.1建⽴健全⼈⼯智能安全法律法规..........................................................13 5.2构建⼈⼯智能科技伦理准则..................................................................13 5.3提升研发应⽤全⽣命周期安全能⼒......................................................14 5.4强化开源⽣态安全和供应链安全.....................................................5.5实施应⽤分类及安全⻛险分级管理...............................................5.6推⼴⼈⼯智能⽣成合成内容可追溯管理........................................5.7安全有效释放重要⾏业应⽤需求...................................................5.8建设⼈⼯智能安全测评体系...........................................................5.9共享⼈⼯智能安全⻛险威胁信息....................................................5.10完善数据安全和个⼈信息保护规范..............................................5.11增进协同应对⼈⼯智能失控⻛险的共识......................................5.12加⼤⼈⼯智能安全⼈才培养⼒度.................................................5.13提升全社会的⼈⼯智能安全意识.................................................5.14促进⼈⼯智能安全治理国际交流合作.........................................1414141515151516161616 6.⼈⼯智能研发与应⽤的安全指引.......................................................17 6.1⼈⼯智能模型算法研发的安全开发指引........................................6.2⼈⼯智能应⽤建设部署的安全指引................................................6.3⼈⼯智能应⽤运⾏管理的安全指引................................................171920 附件1⼈⼯智能安全⻛险的分级原则....................................................6.4⼈⼯智能应⽤访问使⽤的安全指引................................................附件2可信⼈⼯智能基本准则..............................................................附件3术语........................................................................................致谢...................................................................................................2124272931 Content 1. Principles for AI safety governance.................................................35 2. Framework for AI safety governance..............................................38 3. Classification of AI safety risks.........................................................39 3.1 Inherent safety risks of AI technology...........................................3.2 Application safety risks associated with AI technology..............3.3 Derivative safety risks from AI application..................................404346 4. Technological countermeasures to address risks.........................49 4.1 Safeguards against inherent safety risks......................................4.2 Safeguards against application safety risks.................................4.3 Safeguards against application-related secondary safetyrisks........................................................................................................495153 5.1 Formulating and improving laws and regulations for AIsafety......................................................................................................55 5.2 Establishing ethical principles for AI technology........................5.3 Enhancing safety throughout the full life-cycle, includingR&D and application...........................................................................5.4 Strengthening open-source ecosystem safety and supplychain safety...........................................................................................5.5 Implementing AI application classification andriskgradingmanagement........................................................................................5.6 Promoting traceable management of AIGC...............................5.7 Unlocking key industry application demands in a safe andeffective manner...................................................................................5.8 Establishing an AI safety assessment system..............................5.9 Sharing information on AI risks and threats................................5.10 Improving data security and personal informationprotection regulations...............................................
