2025五道口金融论坛｜莫万贵谈AI风险，金融机构过度依赖几家科技公司，要避免羊群效应

Core Viewpoint - The integration of AI in the financial sector presents significant risks, including model hallucination, algorithmic opacity, and over-reliance on a few tech companies, necessitating a governance framework involving multiple stakeholders [1][3][4]. Group 1: AI Risks in Finance - Model hallucination poses a challenge due to the high data accuracy required in finance, which can lead to inappropriate applications in certain areas [3]. - Algorithmic opacity complicates regulatory oversight and risk management, making it difficult to trace accountability [3]. - The increasing reliance on a few large tech companies by financial institutions may amplify traditional risks and create systemic vulnerabilities [3][4]. Group 2: Governance Framework - A governance ecosystem should include six key stakeholders: financial institutions, consumers, tech companies, industry governance organizations, regulatory bodies, and financial professionals [4][5]. - Financial institutions must ensure that AI technologies are suitable for their specific business scenarios to avoid unnecessary complexities and risks [4]. - There is a need for human intervention in critical decision-making processes to enhance controllability and risk management [5]. Group 3: Regulatory Approaches - Regulatory bodies should adopt a balanced approach, allowing for innovation while ensuring risk management through trial and error in a controlled environment [5][6]. - Enhanced regulatory measures should be implemented for systemically important financial institutions, including additional stress testing and higher capital requirements [7][8]. - Cross-market and cross-institutional risk monitoring should be strengthened to address the interconnected risks posed by AI technologies [8].