The following guide from China’s internet and cybersecurity regulatory agency sets ground rules for the acceptable Chinese government use of AI models. The guide encourages central, provincial, and local government offices to use AI to speed up paperwork, improve the effectiveness of surveillance, and predict events such as market fluctuations and natural disasters. It also warns government agencies against adopting AI models willy-nilly, instead urging lower-level governments to use the same systems as their higher-ups. The guide does not specify how governments should choose AI models, nor does it recommend particular models or providers.

An archived version of the Chinese source text is available online at:    https://perma.cc/UKX3-C9CK

Guidelines for the Deployment and Application of Artificial Intelligence Large Models in the Government Sector

These Guidelines have been formulated in order to deeply and thoroughly implement the decisions and deployments of the Chinese Communist Party (CCP) and the State Council, regulate and guide the development and application of artificial intelligence (AI) large models in the government sector, and raise the levels of digitalized and intelligentized (智能化) governance and service in government affairs. These Guidelines are primarily intended to provide government departments at all levels with a work orientation and basic reference for the deployment and application of AI large models. They will be dynamically adjusted based on progress in practice, taking into account new circumstances and new requirements for the development and application of AI large models.

I. Overall Requirements

Persist in taking Xi Jinping Thought on Socialism with Chinese Characteristics for a New Era as the guide, deeply and thoroughly implement the spirit of the 20th Party Congress and the Second and Third Plenums of the 20th Central Committee, and thoroughly implement General Secretary Xi Jinping’s important thinking on building China into a cyber powerhouse.¹ We must completely and accurately implement the new concept of development (新发展理念), and do overall planning for high-quality development and a high level of security.² Persisting in systematic planning and intensive development, a people-centered approach and standardized application, joint construction and sharing and efficient coordination, security and prudence, and striving for real impact, we must, in an orderly fashion, advance the deployment, application, and continuous optimization of AI large model technologies, products, and services in the government sector; and we must fully leverage the advantages of AI large models in areas such as semantic understanding and reasoning, multimodal content generation, and knowledge integration and analysis, so as to provide effective assistance to government staff and convenient services to the public, and promote innovation-based development in government affairs, thereby enhancing governance efficiency, optimizing service management, and assisting scientific decision-making.

II. Application Scenarios

Government departments may select among typical scenarios to conduct exploratory application of AI large models, in accordance with local conditions and actual circumstances, focusing on common and high-frequency requirements in work such as government services, social governance, government office activities, and decision support. The main reference scenarios include the following:

a) Government Services

1. Intelligent question answering. Integrating regional, departmental, and sector-specific operational resources and knowledge bases, use natural language understanding, retrieval-augmented generation, and knowledge graph technologies to provide convenient online government consultation services, strengthen accurate comprehension of inquiries from the public, and generate reference replies in real time, thereby helping to resolve the public’s concerns and increase information accessibility.

To view the rest of this translation, download the pdf below.

1. Translator’s note: Alternate English translations for the Chinese term wǎngluò qiángguó (网络强国)—here translated as “cyber powerhouse”—include “cyber superpower,” “network powerhouse,” “network superpower,” and so on. For a more thorough discussion in English of the meaning of the term wǎngluò qiángguó, see: https://www.newamerica.org/cybersecurity-initiative/digichina/blog/lexicon-wangluo-qiangguo/.
2. Translator’s note: The Chinese word 安全 encompasses the meanings of both “safety” (protection from accidental harm) and “security” (protection from deliberate harm). In this translation, it is variously translated as “safety,” “security,” or “safety and security” at the translator’s discretion.