构建生成式人工智能的安全治理新机制

Group 1 - Generative artificial intelligence (AI) is deeply integrated into various sectors such as government information processing and media content creation, becoming a core support for the development of new productive forces [1] - The 20th Central Committee's Fourth Plenary Session emphasizes the importance of strengthening national security capabilities in emerging fields like AI, highlighting the significance of digital technologies in national security [1] - The traditional approach of "develop first, govern later" may lead to unmanageable risks and missed opportunities in global AI competition, necessitating a governance mechanism that balances resilient defense and innovation [1] Group 2 - The assessment of national security risks associated with generative AI reveals multi-dimensional and cross-sectoral challenges, particularly in political, economic, and social security [2] - In the political security domain, generative AI poses real threats through the industrialized production of false information and the subtle erosion of cultural values, with low barriers for generating misleading content [2] - Economic security concerns include job displacement due to AI and reliance on imported high-end CPUs and GPUs, which poses risks to the AI industry's supply chain and technological sovereignty [3] Group 3 - Social security risks are highlighted by AI-related fraud and privacy breaches, with AI technologies enabling new forms of crime and the potential leakage of sensitive data from large models [3] - A comprehensive governance framework for generative AI should be guided by an overall national security perspective, integrating various aspects such as ideology, social order, and cultural heritage into risk assessments [4] - The governance mechanism should focus on clear responsibilities among government, enterprises, and the public, promoting collaboration and enhancing safety awareness [6] Group 4 - The governance approach should move away from unilateral administrative regulation to a model that encourages technological solutions and industry-driven governance [6] - Legal frameworks must be established to address the unique challenges posed by generative AI, including copyright issues and specific regulations for high-risk scenarios [7] - Ethical guidelines should be practical and enforceable, with companies required to conduct risk assessments and establish ethics review committees for AI applications [7]