As of the end of July this year, there have been a total of 130 large artificial intelligence models released in China, of which 64 were released in the first seven months of this year alone. Not long ago, the first batch of 11 large model products registered through the "Interim Measures for the Management of Generative Artificial Intelligence Services" were approved for opening up services to the whole society, attracting attention. The large model has entered a period of rapid development. Currently, an important issue is how to coordinate innovative development and secure applications, promote high-quality development of generative artificial intelligence, and continuously stimulate the "head goose" effect of artificial intelligence. Benefiting from the vast amount of data, parameters, and good learning ability, large models enhance the universality of artificial intelligence and are expected to become a new foundation for artificial intelligence technology and applications, a fundamental tool for production and life, and bring profound changes to economic and social development. At the same time, we are constantly improving our regulatory methods and formulating corresponding classification and grading regulatory rules or guidelines. For example, the draft authorization of the EU Artificial Intelligence Act passed in June this year focuses on risk and strictly stipulates the pre review procedures and compliance obligations for high-risk artificial intelligence systems. For domestic large models, there are many but not strong issues that need to be addressed urgently. Due to the difficulty in breaking through the underlying technology, many companies choose to use mature large models to directly empower existing products and services, resulting in severe homogenization of technology and a lack of innovation at the source, resulting in fewer high-quality landing applications. The strong human-machine interaction paradigm of generative artificial intelligence poses challenges to the credibility, reliability, and controllability of generated content, and increases the uncertainty of ethical and security risks. Therefore, it is necessary to seize the key, integrate resources, further practice internal skills in source innovation, lay a solid foundation in risk management, and diligently cultivate external skills in building a large model business ecosystem. Adhere to innovation at the source and improve the core competitiveness of generative artificial intelligence. Develop a medium to long-term special plan for tackling fundamental and original technologies such as high-end computing chips, computing architectures, and large model algorithms, to achieve more theoretical and technological breakthroughs from "0" to "1". Introduce guidance and implementation rules for computing power infrastructure and data resources, accelerate the unified scheduling, openness, and operation process of computing power and data resources, and cultivate a green and low-carbon service model of "generative artificial intelligence+public cloud". Build a batch of national level open source platforms, code hosting, and large model development and testing platforms to ensure the autonomy and controllability of core algorithms. Establish a solid risk bottom line and improve the generative large model classification and grading supervision system. Improve the risk supervision system that integrates government, professional institutions, and society to ensure the credibility and controllability of large models. Establish a national generative artificial intelligence regulatory department, based on the entire lifecycle of large model development and application, establish a classification and grading standard system for endogenous and derivative risks, clarify regulatory objects, content, responsibilities, methods and principles, and implement a large model risk filing system; Focusing on hot areas such as finance, healthcare, and education, we will continue to release regulatory rules and implementation measures for industry models. Select a group of risk assessment professional institutions to carry out risk monitoring and large-scale model qualification assessment. Accelerate the implementation of applications and create a generative large model business ecosystem. Build a universal model element market ecosystem for resource collaboration among the three major platforms of computing power, data, and open source community, and support high-quality research and development by national innovation consortia