AI Safety Governance Framework (Version 1.0)

Summary

Version 1.0 of the AI Safety Governance Framework, issued by TC260 under the Global AI Governance Initiative. Its principles: be inclusive and prudent to ensure safety; identify risks with agile governance; integrate technology and management for coordinated response; promote openness and cooperation. It classifies inherent safety risks (models and algorithms, data, systems) and application safety risks in the cyber, real-world, cognitive and ethical domains, including misuse for nuclear, biological and chemical weapons and uncontrollable AI. It sets technical and governance measures, such as tiered management and incident response, and safety guidelines for developers, providers and users.

Safety topics

Status notes

Released at the main forum of the 2024 National Cybersecurity Week to implement the Global AI Governance Initiative, and published by TC260 in Chinese and English. Version 2.0 (15 Sept 2025) was built on it and replaced it; version 3.0 followed on 14 Sept 2026.

Sources

Open the official source https://www.cac.gov.cn/2024-09/09/c_1727567886199789.htm

Secondary source: https://www.tc260.org.cn/front/postDetail.html?id=20240909102807

Local copy

Verification

History

1. 2026-10-14english_version: – → {"kind": "authentic", "url": "https://www.cac.gov.cn/cms/pub/interact/downloadfi — Return round (14 Oct 2026): set to TC260's English text, linked from the CAC notice and now stored. Kind 'authentic': the issuer published it under its own name with the same title block (V1.0); 'government translation' would also be defensible, since the Chinese is evidently the drafting text.2026-10-14-refresh-translation-fixes · Claude Code (claude-opus-5-5)
2. 2026-10-14notes: Released at the main forum of the 2024 National Cybersecurity Week to implement → Released at the main forum of the 2024 National Cybersecurity Week to implement — Return round (14 Oct 2026): says TC260 published both language versions and that version 2.0 was built on this one (stored notice).2026-10-14-refresh-translation-fixes · Claude Code (claude-opus-5-5)
3. 2026-10-14safety_topics: ["risk-based classification", "evaluation & testing", "international cooperation → ["risk-based classification", "evaluation & testing", "catastrophic risk", "inci — Return round (14 Oct 2026): added 'catastrophic risk' (3.2.2(c) nuclear, biological and chemical weapons; 3.2.4(c) AI becoming uncontrollable) and 'incident reporting' (5.7 risk and threat information sharing and emergency response; 6.2(h) providers report safety incidents and vulnerabilities to the competent authorities).2026-10-14-refresh-translation-fixes · Claude Code (claude-opus-5-5)
4. 2026-10-14summary: China's first national AI safety governance framework. Its principles are inclus → Version 1.0 of the AI Safety Governance Framework, issued by TC260 under the Glo — Return round (14 Oct 2026): rewritten from the framework's Chinese text, now stored, with the four principles in TC260's own English wording (1.1 'Be inclusive and prudent to ensure safety' for 包容审慎、确保安全, and so on). 'Guidelines for developers, providers and users' is now supported (section 6: model and algorithm developers, service providers, users in key areas, the public). Added the dual-use weapons risk (3.2.2(c)) and loss of control (3.2.4(c)); 5.1 分类分级管理 rendered 'tiered management' after TC260's 'tiered and category-based management'.2026-10-14-refresh-translation-fixes · Claude Code (claude-opus-5-5)
5. 2026-10-14original_title: 人工智能安全治理框架（1.0版） → 人工智能安全治理框架（V1.0） — Return round (14 Oct 2026): '人工智能安全治理框架（1.0版）' was not verbatim; the framework's own title block reads 人工智能安全治理框架 （V1.0）.2026-10-14-refresh-translation-fixes · Claude Code (claude-opus-5-5)

Related records

AI Safety Governance Framework 2.0 China · Superseded

AI Safety Governance Framework 3.0 China · Published (non-binding, current)

GAIGI China · Published (non-binding, current)

Citation

National Cybersecurity Standardization Technical Committee (TC260). AI Safety Governance Framework (Version 1.0). 2024-09-09. https://www.cac.gov.cn/2024-09/09/c_1727567886199789.htm. Status: Superseded; verified 2026-10-14 (level A). Local copy archive/CN-07/2026-10-11/official.html retrieved 2026-10-11 14:00:10 UTC, SHA-256 6198bb66c8ba74227e639c0921decb8c4659d210538d7a011d8c2e90b91f575b. Every AI Policy Ever, version 1.18 (snapshot 2026-10-18), record CN-07.

Improve this record

Know of an amendment, a status change or an error? Report news or an error on GitHub; the form opens with this record filled in.