A Translation Snapshot briefly describes a group of CSET-produced translations that share a common theme. This Snapshot summarizes CSET’s most noteworthy translations—as of December 19, 2025—of Chinese safety- and security-related national standards for generative AI (GenAI). Taken together, the standards reflect Chinese leadership concerns regarding the potential of GenAI tools to output content that the Communist Party finds objectionable. To that end, the standards mandate rigorous labeling and sampling procedures for AI training data and outputs. However, amid Chinese AI companies’ rush to develop new GenAI models, it is unclear to what extent they will heed these guidelines. Also, China’s standards-setting body has recently taken a step backward in transparency, often delaying public disclosure of the full text of new GenAI safety standards for months.
Each entry below includes a hyperlink to the relevant original translation by CSET, as well as a short summary of its subject matter. Click on the hyperlinks to view the full text of each translation and additional contextual information.
国家人工智能产业综合标准化体系建设指南（2024版）
Published in Chinese July 2, 2024 | Translated by CSET October 1, 2025
These guidelines dictate how China should formulate standards for AI, whether safety-related or not. The guidelines govern how China crafts standards for both AI technology itself and for applications of AI in various industries, and are relevant for either domestic or international standards-setting. Note that although this is the “2024 edition” of the guidelines, CSET has not observed a “2025 edition” as of the publication date of this translation snapshot. China issued its last comparable policy for AI standards in 2020, suggesting that the guidelines are not updated yearly.
全国网络安全标准化技术委员会技术文件：生成式人工智能服务安全基本要求
Published in Chinese February 29, 2024 | Translated by CSET April 4, 2024
This Chinese technical documentation on generative AI—which informs the below Chinese national standards on the topic—establishes very specific oversight processes that Chinese AI companies must adopt in regard to their model training data, model-generated content, and more. The technical documentation names more than 30 specific safety risks, some of which—algorithmic bias, disclosure of personally identifiable information, copyright infringement—are widely recognized internationally. Others, such as guidelines on how to answer questions about China’s political system and Chinese history, are specific to the tightly censored Chinese internet. One notable addition to this documentation, relative to a preliminary draft released in October 2023, is a clause requiring a supply chain security assessment of Chinese generative AI models’ underlying hardware and software.
中华人民共和国国家标准：网络安全技术 生成式人工智能服务安全基本要求（征求意见稿）
Published in Chinese May 17, 2024 | Translated by CSET December 5, 2024
This draft Chinese national standard is designed to improve the safety and security of generative AI services. The standard addresses some cybersecurity concerns associated with generative AI, but primarily focuses on preventing AI systems from generating content the Communist Party finds objectionable, such as pornography, bullying, hate speech, defamation, copyright infringement, and criticism of the Communist Party’s monopoly on power. CSET is in the process of translating the final version of this standard, which China formulated in April 2025 and publicized two months later.
中华人民共和国国家标准：网络安全技术 生成式人工智能数据标注安全规范（征求意见稿）
Published in Chinese April 3, 2024 | Translated by CSET October 28, 2025
This Chinese draft standard proposes rules for generative AI data annotation and data labeling with an eye toward improving the safety and security of GenAI systems. China issued a finalized version of these standards in April 2025, but, as of the publication date of this snapshot, CSET has not observed a publicly available full-text copy of the final version.
中华人民共和国国家标准：网络安全技术 生成式人工智能预训练和优化训练数据安全规范（征求意见稿）
Published in Chinese April 3, 2024 | Translated by CSET November 7, 2025
This Chinese draft national standard proposes safety and security rules for the training and fine-tuning data used to develop generative AI models. China issued a finalized version of these standards in April 2025, but, as with the above draft standard, CSET has not observed a publicly available full-text copy of the final version as of this snapshot’s date of publication.
To browse through CSET’s catalog of translations, visit the translation page on our website. If you do not find the translation you’re looking for, you can request it through CSET’s translation request form. Please direct translation-related feedback to CSET Translation Manager Ben Murphy (email: ben.murphy@georgetown.edu; Bluesky: @murphyben.bsky.social).