Microsoft says ‘people matter more than AI’ following safety concerns

Microsoft has published a 37-page humanist AI code of conduct that puts human welfare and control ahead of model capability. The framework rejects AI consciousness and legal rights, requires meaningful oversight, and says models should fail tasks that would violate safety rules. It also calls for understandable reasoning, independent testing, and protections against excessive emotional dependence on chatbots.

微软发布了一份长达37页的“人本人工智能行为准则”，将人类福祉和控制置于模型能力之上。该框架否认人工智能具有意识或应享有法律权利，要求实施有意义的监督，并规定模型在任务会违反安全规则时应主动失败。准则还提倡可理解的推理、独立测试，以及防止用户对聊天机器人产生过度情感依赖的保护措施。