证监会科技司刘铁斌：构建规范健康的资本市场人工智能生态体系

Core Viewpoint - The China Securities Regulatory Commission emphasizes the need to build a standardized and healthy artificial intelligence ecosystem in the capital market, balancing innovation promotion and regulatory compliance while focusing on industry ecology, institutional ecology, and strict adherence to regulatory requirements [1][2][3] Group 1: Industry Ecology for AI Development - Establish a differentiated regulatory framework based on the impact and potential risks of AI applications, implementing tiered and classified supervision [1] - Promote pilot projects for "AI + capital market" to quickly develop replicable and scalable practical cases [1] - Set up AI pilot bases and innovation application verification platforms to facilitate the sharing of resources such as computing power, data, and models [1] - Enhance the construction of high-quality data sets in key areas like legal regulations, market information, and disclosure [1] - Strengthen resource support and research to improve incentive measures, talent development, and collaborative innovation among industry, academia, and research [1] Group 2: Institutional Ecology for AI Application - Improve governance structures and strategic planning to align AI applications with institutional development across business, technology, compliance, and risk management [2] - Adhere to national AI ethical guidelines and social norms to ensure compliance with laws, regulations, and ethical standards in AI applications [2] - Enhance information disclosure and investor protection by clearly identifying AI-generated content to safeguard investors' rights [2] - Strengthen data governance by using authoritative data sources and improving data management standards to enhance data quality [2] - Develop risk assessment and emergency response mechanisms to continuously monitor medium to high-risk business scenarios [2] Group 3: Compliance with Regulatory Requirements - Ensure content compliance by establishing content barriers and output verification mechanisms, with high-risk scenarios undergoing manual verification [3] - Mitigate data security risks by enhancing data management to prevent illegal collection, data misuse, information leakage, and other risks [3] - Strengthen cybersecurity management to improve system defense capabilities and ensure the stable operation of AI-related information technology systems [3] - Enhance model management capabilities by creating a comprehensive evaluation system covering safety, compliance, and service capabilities [3] - Recognize the inherent limitations of AI and establish a "human supervision + model assistance" mechanism in critical business areas to prevent risk spillover [3]