In October 2025, the Cyberspace Administration of China and the National Development and Reform Commission jointly rolled out the Guidelines for the Deployment and Application of Large Artificial Intelligence Models in the Government Sector. These guidelines serve as a crucial work blueprint and foundational reference for government departments at all levels to deploy and utilize large AI models effectively.

Underpinned by Xi Jinping Thought on Socialism with Chinese Characteristics for a New Era, the guidelines fully embody the essence of the 20th CPC National Congress, as well as the Second and Third Plenary Sessions of the 20th CPC Central Committee. They highlight several key principles: systematic planning, intensive development, a people-centric approach, standardized applications, collaborative construction for high efficiency, along with a commitment to safety, prudence, and tangible results. The overarching goal is to foster the orderly deployment, application, and ongoing refinement of large AI model technologies within the government sector.

The guidelines detail 13 specific application scenarios, categorized into four main areas: government services, social governance, office operations, and decision support. They mandate that government departments select representative scenarios for exploratory applications, taking into account local circumstances. Simultaneously, they emphasize the importance of standardized deployment, cautioning against the blind pursuit of technological superiority and the avoidance of 'digital formalism.' A comprehensive lifecycle management system is also called for.

In terms of operational management, the guidelines stipulate that large AI models should play a 'supportive' role, with measures in place to mitigate risks such as model 'hallucinations.' Continuous iteration and optimization are encouraged through the establishment of a regular update mechanism. Additionally, the guidelines place a strong emphasis on security management, requiring the implementation of a security responsibility system, strict compliance with confidentiality standards, and proactive measures to prevent sensitive data breaches.