1. Background
Formulated by the National Cybersecurity Standardization Technical Committee (TC260) and officially released at the main forum of the National Cybersecurity Publicity Week on September 9, 2024. The framework responds to the Global AI Governance Initiative and aims to provide systematic and actionable guidance for AI safety governance and promote the construction of a safe, credible, and controllable AI application ecosystem.
2. Governance principles
Adhere to the principles of inclusiveness, prudence and safety, risk-oriented and agile governance, technology and management combined with collaborative response, openness, cooperation, co-governance and sharing.
3.
- Endogenous security risks of framework structure: covering model algorithm security, data security, system security, etc.
- Application security risks: including network environment risks, real-world scenario risks, cognitive bias risks, ethical risks, etc.
- Countermeasures: Propose technical countermeasures and comprehensive governance measures, and provide security guidance for different entities such as algorithm developers, AI service providers, industry users, and the public.
4. Strategic significance
Although this framework is not legally enforceable, as our country's first systematic guidance document for artificial intelligence security governance, it provides a path reference for risk identification, technical response, and collaborative governance, and promotes international cooperation and consensus on artificial intelligence security governance.
See official website for details:
https://www.tc260.org.cn/front/postDetail.html?id=20240909102807
