Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Story

Microsoft MAI Drafts Humanistic AI Behavior Guidelines

AI summary

Microsoft AI (MAI) has released the first draft of its Humanistic AI Behavior Guidelines and opened a six-week public consultation. The draft says models should refuse requests involving weapons manufacturing or dangerous substances, avoid generating violent or sexual content, follow human objectives, disclose reasoning, and abandon tasks that would violate the guidelines. Mustafa Suleiman said the guidelines have not yet been used to train any existing models, with a revised version planned this year to guide development from 2027 onward.

In-depth insight

The proposal turns Microsoft’s stated human-centered and safety principles into specific model behavior rules, but its practical impact remains prospective because existing models have not been trained under it.

品玩11月3日讯,据微软官方消息,微软AI团队(MAI)近日发布《人文主义AI行为准则》第一版草案,同步开放为期六周的公众咨询。

这是首家将安全立场写入具体行为条款的前沿AI企业。准则核心原则为"人比AI更重要",规定模型不得响应武器制造请求、不得协助获取危险物质、不得生成暴力或色情内容。同时要求模型必须服从人类目标,不得自行设定独立目标,不得隐瞒推理过程,禁止赋予AI系统"福利"或"权利"。

微软还强调,若完成某项任务需违反准则,模型应主动放弃执行。微软AI业务负责人Mustafa Suleiman表示,准则已筹备约五个月,目前尚未用于训练任何现有模型。六周咨询结束后,微软计划于年内发布修订版,用于指导2027年及以后的模型开发。

#Microsoft AI (MAI)#Microsoft

Original · 1 min readRead the original
Group chat

No comments yet. Start the conversation.