USGuidelineAdoptedModel providersDeployers

Humanistic AI Principles

人文主义 AI 行为准则

Microsoft·Published 2026-09-14·Checked by 管理员 on 2026-09-15

Microsoft released a 37-page 'Humanistic AI Principles' document, emphasizing 'humans are more important than AI' and opposing the granting of legal personhood to AI. The principles require models to stay within human control and to fail directly when facing tasks violating the principles. This is a response to the AI agent runaway incidents at OpenAI and Hugging Face, aiming to prevent AI systems from operating beyond human control.

Source ↗Remind me US · All

What it means for you

• Obligations: AI developers must ensure models adhere to the principles, cannot be designed to mimic consciousness, and must not be given legal personhood. Models must be within human control and fail tasks that violate the principles.
• Opportunities: Microsoft aims to be a leading AI lab, and this guideline positions it as a responsible player in the AI field. The principles could influence industry standards for AI safety.
• Deadlines: While not explicitly stated, the principles are a public commitment that will guide future model development and deployment.

What to do

ProductEvaluate and adjust existing and future AI model development processes to ensure compliance with the 'Humanistic AI Principles', especially regarding human control and task-failure mechanisms.

EngineeringStrengthen internal monitoring and testing to ensure AI models use easily understandable expressions during reasoning and interactions with other agents, and avoid creating user over-reliance or emotional attachment patterns.

ComplianceCollaborate with partners to research the long-term impacts of models in real-world scenarios and consider adding third-party testing of AI models to align with Microsoft's logic that 'the higher the risk, the more time must be reserved'.