Microsoft published a provisional code of conduct restricting its AI models, responding to recent calls from Anthropic and OpenAI leaders to slow frontier development. The code prohibits models from assisting with weapons, dangerous substances, or unhealthy content, and requires transparency in reasoning. Microsoft will seek external input before finalizing rules that will guide development starting in 2027.
213 other tracked developments touching the same claims.