Zhixixi reported on May 9 that yesterday evening, the Cyberspace Administration of China, the National Development and Reform Commission, and the Ministry of Industry and Information Technology jointly issued"Implementation Opinions on the Standardized Application and Innovation Development of Intelligent Agents" (referred to as: "Implementation Opinions").

This is the first time that the three national ministrieshave formulated a systematic policy with "Agent" as the core theme. Previously, no national ministry document has specifically focused on the definition, security, application, and ecology of agents to comprehensively standardize them.

▲Screenshot of the "Implementation Opinions" release interface

The "Implementation Opinions" clarifies that the development of intelligent agents must adhere to the basic principles of safe and controllable, standardized and orderly, innovation-driven, and application-driven, and proposes four measures: Consolidate the development foundation, secure the bottom line of safety, strengthen application traction, and build an innovation ecosystem.

In terms of strengthening application traction, it proposed 19 typical application scenarios around the directions of scientific research, industrial development, boosting consumption, people's livelihood and welfare, and social governance.

In terms of regulating the application of intelligent agents, it proposes to clarifyproduct guidelines, prevent security risks, improve governance systems, and strengthen industry self-discipline. According to the "Implementation Opinions", the management of intelligent agents will be more standardized in the future, Classification and grading, Security fence, digital identity, strict filing, one at a time; Prevent minors from being addicted and emotionally dependent, was also included in the comments.

In August 2025, the State Council issued the "Opinions on Deeply Implementing the "Artificial Intelligence +" Action", proposing to2027to take the lead in realizing extensive and deep integration of AI with key areas,The phased goal of the penetration rate of new generation smart terminals, smart bodies and other applications exceeding 70%. How to achieve this goal? The "Implementation Opinions" are specific measures to implement the opinions.

The full text of the "Implementation Opinions" is as follows:

Agent is an intelligent system with autonomous perception, memory, decision-making, interaction and execution capabilities. It is an important form of artificial intelligence products and services. With the rapid development of new generation artificial intelligence technologies such as large models, intelligent agents are accelerating their deep integration with cyberspace and the physical world, profoundly changing human production and lifestyle and social governance models. In order to implement the State Council's "Opinions on Deeply Implementing the "Artificial Intelligence +" Action" and promote the standardized application and innovative development of intelligent agents, these implementation opinions are formulated.

Basic Principles

Oriented by promoting technological innovation, improving governance capabilities, building an industrial ecosystem, and improving people's livelihood and well-being, we insist on safety and controllability, and regard the safety, reliability, and trustworthiness of intelligent agents as the bottom line requirements for development, and run through the entire process of intelligent agent technology research and development, application deployment, and promotion to effectively prevent systemic risks. Adhere to standards and order, adapt to the evolution of intelligent agent technology, build a governance system that is smoothly connected with existing policies and regulations, have industry self-discipline and autonomy, and have clear bottom lines, and promote the implementation of intelligent agents in an orderly manner. Adhere to the innovation drive, strengthen the linkage of theoretical innovation, technological innovation, and engineering innovation, systematically break through key intelligent agent technologies, improve the coordination mechanism of government, industry, academia, and research, build an open and shared intelligent agent ecosystem, and enhance the vitality of industrial innovation. Adhere to application traction, focus on practical needs such as scientific research, industrial development, boosting consumption, people's livelihood and welfare, and social governance, and give full play to the demonstration effect of typical application scenarios, starting with easy things and then difficult ones, step by step, to promote intelligent agent technology verification, product iteration, and application implementation.

Consolidate the foundation for development

Consolidate the technical base, improve the standard system, lower the threshold for intelligent agent research and development, adaptation, and application, and lay the foundation for enriching intelligent agent products and services.

(1) Improving the technical base

1. Strengthen basic technology research and development. Continue to improve the performance of general basic models, support specialized models in industry development subdivisions, and form a model product matrix that adapts to different scenarios and equipment. For intelligent agent training and operation, improve the ability to provide high-quality data sets. Strengthen technological research on intelligent agent task understanding, task planning, tool use, long-term memory, mutual recognition and interoperability, and group collaboration to improve the agent's generalization capabilities.

2. Improve the agent tool chain. Carry out research on the underlying framework of agents, accelerate the development of key components such as perception, memory, decision-making, interaction, and execution, and improve the tool chain for agent R&D, testing, deployment, and operation and maintenance. Develop security and governance tools such as adversarial sample detection and behavioral anomaly detection to improve the ability to detect, intervene, block, and recover from non-compliant behaviors of agents.

(2) Building a standard protocol

3. Establish an intelligent agent standard system. Develop guidance documents for intelligent agent standardization work, form an intelligent agent standard framework, system layout standard systems such as key technologies, important products, data exchange, application scenarios, quality evaluation, security assurance, trustworthy certification, etc., and accelerate the formulation of basic standards for intelligent agents and software tools, application services, hardware peripheral interfaces, etc. Strengthen the promotion and application of key national and industry standards for intelligent agent interconnection such as the Intelligent Interconnection Protocol (AIP). Support the development of mandatory standards in medical, transportation, media, public safety and other fields. Encourage enterprises to develop products and services in accordance with relevant standards and improve the standardization of intelligent agents. Actively participate in the formulation of international standards.

4. Lay out and develop the intelligent Internet. Research and establish an intelligent Internet system architecture, explore the establishment of an intelligent agent registration platform, provide intelligent agentdigital identity management, retrieval and discovery, capability declaration and other services, and support information query and management of developers, deployment methods, interface protocols, compliance certification and other information. Improve multi-agent collaboration capabilities and study basic technologies such as agent identity, trusted interconnection, compliant payment, security protection, and conflict resolution. Leverage the technical advantages of Internet Protocol version 6 (IPv6) to improve the end-to-end communication capabilities of intelligent agents. Explore and establish an intelligent Internet monitoring indicator system.

Keep the bottom line of safety

Adhere to people-oriented, intelligent benevolence, pluralistic and co-governance, safety and stability, create an institutional environment that regulates development and encourages innovation, and promotes the healthy and orderly development of intelligent entities.

(1) Clarify product guidelines

5. Improve policies, regulations and ethical norms. Accelerate the research on policies, regulations and ethics related to agents, give full play to the content resources and review and control advantages of professional institutions, and ensure that agent behavior complies with laws, regulations and mainstream values. Prevent agents from taking advantage of data and personification technology to spread bad values ​​and algorithmic exploitation, and prevent risks such as addiction and emotional dependence among minors and the elderly. Make good connections with artificial intelligence ethics review and other systems.

6. Clarify decision-making authority. Under the premise of complying with laws and regulations, respecting social morality and ethical norms, clarify the reasonable boundaries and required permissions of various decision-making methods such as decision-making only by the user himself, decision-making requiring user authorization, and independent decision-making by the agent. Ensure that users have the right to know and the final right to make decisions independently on the agent, and the operations performed by the agent must not exceed the scope of user authorization.

7. Strengthen behavioral control. Develop technologies such as rule embedding and behavioral fences to ensure that the behavior of intelligent agents in public places, private places, special places, etc. is legal and compliant. Explore the use of technologies such as blockchain to establish a verifiable and traceable mechanism for the behavior of intelligent agents in important application scenarios to prevent major risks caused by inappropriate behavior of agents.

(2) Preventing security risks

8. Improve endogenous safety capabilities. Research security technologies such as agent data security, personal information protection, password protection, attack detection, authority management, and behavior control to improve the security capabilities of the agent system and prevent security risks such as data poisoning, privacy leaks, algorithm tampering, system vulnerabilities, and out-of-control operations. Research intelligent body safety detection technology and explore the establishment of an intelligent body safety assessment system.

9. Strengthen supply chain security. Develop full-cycle security specifications for agent development, deployment, application, and maintenance, and strengthen security management of model access, application program interface calls, and use of extended tools. Explore and establish an intelligent supply chain security information sharing and early warning mechanism, issue risk warnings in a timely manner, and improve security assurance capabilities.

10. Resolve application-derived risks. Improve normalized risk identification, early warning and intervention mechanisms for intelligent agents, strengthen risk handling capabilities such as human-machine collaborative review and interception, and prevent systemic security risks. Strengthen the security management of intelligent agent applications to prevent intelligent agents from being used for automated attacks, privacy violations, false information generation and dissemination, Internet fraud and other illegal and criminal activities.

(3) Improving the governance system

11. Build a classification and hierarchical governance framework. Carry out intelligent agent hierarchical management in a prudent and steady manner based on application scenarios and potential impacts. For sensitive areas and key industries, the cybersecurity and informatization department will work with industry authorities to determine opening scenarios, and implement management measures such as filing, testing, and recall of problematic products in accordance with relevant laws, regulations, regulatory requirements, and safety protection standards. For some low-risk areas such as lifestyle entertainment and daily office work, we will improve intelligent agent assessment and testing tools and achieve efficient governance through compliance self-testing, information reporting, distribution platform management, and industry self-discipline.

12. Improve the compliance service system. Strengthen the supply of professional services such as intelligent risk monitoring and early warning, detection and assessment, consulting, and certification, and guide the industry to actively develop intelligent agent monitoring tools. Carry out third-party evaluation services such as intelligent agent functions, performance, quality, compliance, etc., promote the mutual recognition of certification and testing results, and provide users with a reference for selecting intelligent agents. Compile and publish intelligence technology and application maturity reports to provide reference for industrial R&D and applications.

(4) Strengthening industry self-discipline

13. Guide the industry to strengthen self-discipline. Encourage industry organizations and major enterprises to jointly formulate industry self-regulatory rules and clarify regulatory details such as intelligent agent function compliance, algorithm governance, intellectual property protection, and fair competition. Guide intelligent agent development platforms, distribution platforms, and service providers to establish fair and reasonable platform rules, user service agreements, and privacy policies, clarify the rights and responsibilities of both supply and demand parties, and ensure the healthy development of the industry. Strengthen publicity and education on the risks of intelligent application and enhance user safety awareness.

14. Explore the credit evaluation mechanism. Guide industry organizations to establish a credit evaluation mechanism for the voluntary participation of intelligent market entities, conduct credit evaluations for technology abuse, induced consumption, false propaganda, concealment of defective information, etc., and carry out breach of trust punishment in accordance with laws and regulations. Guide intelligent agent developers, development platforms, distribution platforms, service providers, etc. to participate in credit evaluation and jointly create a good development environment.

Strengthen application traction

Actively and steadily promote the application of typical scenarios of intelligent agents, drive the optimization and improvement of technology products, and explore the formation of replicable and popularized intelligent agent application models.

(1) Scientific research

15. Scientific research and exploration. Develop theoretical deductions, simulations and other intelligent agents to explore potential technical paths. Strengthen the intelligence's information correlation and integration, knowledge system construction and other capabilities, and improve the discovery capabilities of natural sciences, philosophy and social sciences. Promote the integration of intelligent agents with scientific instruments and experimental platforms to realize the intelligentization of the entire process of solution design, experimental operations, data processing, and result analysis.

16. Research and development assistance. Develop software development agents and improve full-process development capabilities such as demand analysis, architecture design, code generation and testing. Promote the integration of intelligent agents with computer-aided design (CAD), computer-aided engineering (CAE) and other software, and provide functions such as design plan generation, simulation verification, and parameter tuning.

(2) Industrial Development

17. Intelligent manufacturing. Research and develop production management intelligence, dynamically optimize production scheduling, resource allocation and process connection, promote the application of intelligence in the field of industrial Internet, and improve the level of refined management of enterprises. Improve the intelligent agent's capabilities in process parameter optimization, processing accuracy detection, and product defect identification, promote the integration of intelligent agents with CNC machine tools, industrial robots, automated production lines, etc., and promote quality, efficiency, and cost reduction.

18. Energy resources. Develop intelligent agents that sense environmental elements such as atmosphere, water, soil, and noise to improve risk warning capabilities for natural disasters, environmental pollution, and other risks. Rely on intelligent agents to strengthen the full-cycle management capabilities of land and space resources. Relying on intelligent agents to achieve efficient exploration of energy, metals and other mineral resources. Develop intelligent agents such as power dispatching, power consumption monitoring, and power grid maintenance to improve the efficiency of power resource use.

19. Transportation. Develop intelligent agents for traffic safety supervision, emergency command and dispatch, etc. to improve capabilities such as identification of traffic violations, early warning of transportation infrastructure risks, supervision of key vehicles (vessels), and rapid response to accidents. Optimize the performance of traffic monitoring and dispatching agents, develop transportation vehicle management and control agents, and improve the traffic efficiency of road networks, water networks, and airspace.

20. Agricultural production. Develop agricultural service intelligence agents to provide agricultural technology guidance, pest and disease diagnosis and prevention and other services. Promote the application of intelligent agents in planting, breeding, efficient breeding and other aspects, and promote the intelligent transformation of agriculture. Promote the integration of smart bodies with smart agricultural machinery, smart greenhouses, and agricultural service platforms to improve agricultural production efficiency.

21. Financial services. Develop financial risk control intelligence to improve risk identification capabilities in credit approval, transaction monitoring, account security and other aspects. Improve the intelligent agent anomaly detection and compliance audit functions, and enhance capabilities such as credit default prediction, credit card fraud interception, and anti-money laundering monitoring.

(3) Boost consumption

22. Terminal application. Promote smart bodies to empower Internet applications and services, and optimize service experiences such as online shopping, travel navigation, daily payment, and daily office work. Promote the coordinated development of smart bodies and terminal devices such as mobile phones, computers, automobiles, home furnishings, wearables, and consumer robots, and improve the ability to complete cross-application and cross-device tasks.

23. Cultural tourism. Develop content creation agents for literature, music, painting, audio-visuals, performing arts, etc. to promote the dissemination and promotion of excellent culture. Develop intelligent tour guides, multilingual translations, and other tourism service intelligence agents for the elderly and disabled to improve tourism service levels.

24. Business services. Improve intelligent customer service capabilities and provide 7×24 hours consultation, appointment, after-sales and other services. Develop embodied intelligence for guidance, cleaning, warehousing, and distribution to improve the operational efficiency of commercial venues such as catering, retail, accommodation, and logistics. Explore the provision of low-cost housekeeping, elderly care, childcare, disability assistance and other services through embodied intelligence.

(4) People's livelihood and welfare

25. Education and teaching. Explore agents such as courseware generation, homework correction, and academic situation analysis to improve teacher work efficiency. Rely on intelligent agents to develop personalized learning plans and improve functions such as intelligent guidance, question-and-answer guidance, and virtual teaching assistants. Support online education platforms to develop intelligent agents and provide lifelong learning services.

26. Medical health. Improve the performance of medical auxiliary intelligent agents such as medical image analysis, disease diagnosis reasoning, and customized diagnosis and treatment plan generation, and explore intelligent agents such as drug management, surgical scheduling, and medical record management to improve the efficiency of medical services. Steady development of pre-consultation, report analysis and other intelligent systems to improve patient experience.

27. Human resources. Explore the application of intelligence in employment promotion, technical skills talent training and evaluation, labor relations public services and other fields to improve employment service capabilities. Develop social insurance, labor dispute arbitration, wage arrears management and other intelligent entities to protect the legitimate rights and interests of workers.

28. Information services. Explore the application of intelligent agents in network content construction and management, and encourage information release departments and content dissemination platforms to develop intelligent agents for user analysis, topic selection and planning, editing and processing, distribution recommendations, intelligent review, public opinion guidance, emotional guidance, real-time translation, etc., to achieve efficient integration of multi-modal information and cross-domain information.

(5) Social Governance

29. Government services. Explore matter-assisted approval agents to promote the intelligence of government approval processes. Develop policy consultation intelligence to provide all-day online government consultation, process guidance and other services. Explore the proactive push of adaptation policies, service reminders and handling guidelines, and accelerate the transformation from "people looking for services" to "services looking for people".

30. Judicial services. Explore full-process case-handling auxiliary intelligence to improve capabilities such as case material sorting, case information entry, evidence review, and auxiliary legal document generation. Develop legal publicity, legal consultation, legal supervision and other intelligent entities to provide the public with efficient and convenient online judicial services.

31. Public safety. Explore intelligent entities such as monitoring and early warning, emergency response, rescue dispatch, and collaborative governance to improve production safety supervision and disaster prevention, reduction, and relief capabilities. Improve the identification of abnormal behavior of intelligent agents, early warning of potential threats, and dynamic prevention and control processing capabilities to maintain public safety. Promote the application of embodied intelligence in areas such as disaster relief, security inspections, and dangerous goods disposal.

32. Urban governance. Explore the application of intelligent agents in urban planning, construction and governance to support intelligent construction, housing management, safe operation of urban infrastructure, etc., improve the professional level of urban governance, and improve the quality of the urban living environment.

33. Bidding. Explore tendering and bidding intelligent agents to realize full-link intelligent management of tendering and bidding activities to ensure that the entire process is standardized and efficient. Improve the intelligence level of bidding transactions, services and supervision, and achieve scientific and reasonable bidding, fair and impartial bid evaluation, and penetrating and efficient supervision.

Building an innovative ecosystem

Smooth supply and demand channels, promote high-level interaction between the R&D side and the demand side, and form an intelligent industry ecosystem that is driven by the market and driven by internal development.

(1) Promote industrial cooperation

34. Cultivate the power of open source innovation. Guide the domestic artificial intelligence open source community to strengthen the layout of agents and carry out the compatibility and adaptation of agents with open source chips, open source operating systems, and open source large models. Guide enterprises, universities, and scientific research institutions to actively participate in open source projects such as agent frameworks, interactive interfaces, and tool chains, promote the integrated development of technology systems, and accelerate the improvement of international influence.

35. Build an industry collaboration platform. Give full play to the role of industrial collaboration platforms such as intelligent agent-related ecological alliances and technology verification laboratories, coordinate with the upstream and downstream of the industry chain to carry out intelligent agent common technology research and development, standard formulation, evaluation and certification, etc., and carry out the cultivation of talents with a combination of intelligent agent technology and industrial applications. Guide enterprises in Internet applications, smart terminals and other fields to jointly build an ecosystem and explore a mutually beneficial and win-win cooperation model.

(2) Strengthen application promotion

36. Build application promotion channels. Promote the establishment of intelligent agent software stores and industry supply and demand information release platforms, and guide intelligent agent companies to actively release products to create an agglomeration effect. Carry out supply and demand docking activities for intelligent applications, and adopt methods such as public bidding and unveiling to attract intelligent companies to customize and develop corresponding products. Guide complete machine, software and other companies to develop products and services based on intelligence and cultivate the user market.

37. Promote the opening of key scenes. Promote the opening up of intelligent application scenarios in key areas, carry out pilot applications of intelligent agents in industrial clusters, key industries and key areas, and create a number of leading demonstration projects. Develop market-oriented and professional intelligent technology transformation service organizations, explore intelligent application scenarios, and improve the efficiency of technological achievement transformation. Promote industry data sharing and openness to support the training and deployment of intelligent agents in key scenarios.

38. Actively cultivate the global ecology. Relying on international platforms such as the World Artificial Intelligence Conference and the World Internet Conference, we will exchange and display the innovation results of intelligent agent technology. Promote terminal equipment and software companies to adapt to smart bodies, guide relevant companies to improve overseas compliance, and promote smart bodies to adapt to local laws, regulations and cultural customs.

Safeguard Measures

The Cyberspace Administration of China, the National Development and Reform Commission, and the Ministry of Industry and Information Technology will work with relevant parties to strengthen overall planning, strengthen resource integration and force coordination, improve supporting policies, form a synergy, and promote the implementation of key tasks. Establish and improve the evaluation index system for intelligent agent development, and strengthen the monitoring and evaluation, rolling implementation and dynamic adjustment of standardized application and innovative development of intelligent agents.