model safety governance

Tag "model safety governance" — all articles

102026-11OpenAI launches Control Room to track and halt model workloadsOpenAI released its Control Room system on Nov 9, 2026, to track model workloads across its research infrastructure and rapidly stop them when safety or security concerns arise. The company introduced the mechanism in two official posts emphasizing human control over AI workloads, but disclosed no technical details, deployment scope or cases where it has intervened.OpenAI News