对自杀自残行为人工接管，为AI装上“紧急制动”功能

Core Viewpoint - The National Internet Information Office has released a draft regulation on the management of AI humanoid interactive services, emphasizing the need for human intervention in extreme situations such as suicide or self-harm, highlighting the importance of human oversight in AI governance [1][2][4]. Group 1: AI Governance and Regulation - The draft regulation requires AI service providers to establish emergency response mechanisms to handle situations where users express intentions of self-harm [1][2]. - The regulation aims to address both existing issues, such as harmful content generation, and new challenges related to user safety and mental health [1][2]. - The focus on human intervention in life-threatening scenarios marks a shift in AI governance, prioritizing user safety over operational efficiency [2][3]. Group 2: Ethical and Operational Challenges - The regulation acknowledges that AI lacks true value judgment capabilities and cannot assume responsibility for severe consequences, necessitating human oversight [3][6]. - Establishing a human fallback mechanism is seen as a global trend, with discussions on how to implement it effectively while balancing technology, privacy, ethics, and safety [4][5]. - Key challenges include defining the boundaries of human intervention, ensuring accurate emotional state recognition by AI, and addressing privacy concerns related to data collection [5][6]. Group 3: Long-term Implications - The recognition of necessary human intervention reflects an acknowledgment of AI's limitations in ethical responsibility, emphasizing the need for human oversight in emotional and psychological support roles [5][6]. - The draft regulation aims to promote responsible AI innovation while ensuring that AI development aligns with public welfare and national strategic policies [6].