国家网信办起草AI拟人化互动服务管理办法，向社会征求意见

Core Viewpoint - The National Internet Information Office has released a draft regulation for managing AI humanoid interactive services, emphasizing the need for safety, ethical standards, and user protection in the development and application of these services [1]. Group 1: General Principles - The regulation aims to promote the healthy development and standardized application of AI humanoid interactive services while safeguarding national security and public interests [3]. - The state encourages innovation in humanoid interactive services and implements a cautious and classified regulatory approach to prevent abuse and loss of control [4]. Group 2: Service Standards - Providers are encouraged to expand application scenarios safely and reliably, particularly in cultural dissemination and elderly companionship, while aligning with socialist core values [6]. - Providers must adhere to legal and ethical standards, avoiding activities that harm national security, spread false information, or negatively impact user well-being [7][8]. Group 3: Safety Responsibilities - Providers are required to establish comprehensive management systems for algorithm auditing, ethical review, information release, cybersecurity, data security, and personal information protection [8]. - Providers must ensure safety measures are integrated throughout the service lifecycle, including design, operation, and termination phases [9]. Group 4: User Protection - Providers must implement mechanisms to identify user states and intervene when extreme emotions or dependencies are detected, especially for vulnerable groups like minors and the elderly [11][10]. - Providers are mandated to create a minor mode that includes safety settings and requires parental consent for emotional companionship services [12]. Group 5: Compliance and Oversight - Providers must conduct safety assessments under specific conditions, such as reaching a user base of over 1 million or experiencing significant service changes [21]. - The regulation outlines penalties for non-compliance, including warnings, mandatory corrections, or service suspension [29].