Technology
Microsoft releases provisional draft guidelines to restrict future artificial intelligence models
The proposed rules follow broader industry calls to slow frontier development and prioritize safety controls.
The short version
- Microsoft published a draft 37-page code of conduct setting behavioral and safety boundaries for its future artificial intelligence models.[CNBC · Business Insider]
- Under the draft, models are barred from aiding in weapons manufacturing, obtaining dangerous substances, promoting unhealthy eating, or generating violent and explicit content.[CNBC]
- The move comes amid industry debate and recent calls from leaders at Anthropic and OpenAI to moderate the pace of model development.[CNBC]
Key facts
- Microsoft released a draft 37-page code of conduct outlining rules for how its future AI models should operate and behave.[CNBC · Business Insider]
- The proposed guidelines prohibit models from assisting with weapons manufacturing, dangerous substance acquisition, unhealthy eating, or sexually explicit content.[CNBC]
- Microsoft AI CEO Mustafa Suleyman noted that the draft guidelines had been in development for about five months prior to their release.[CNBC]
What remains uncertain
- It remains unspecified how these provisional rules will be finalized or enforced across Microsoft's model development process.[CNBC]
Sources
Outlet counts describe coverage, not independent confirmation. Reports may share a wire service or original source.