Nvidia says new tool can contain rogue AI agents in "milliseconds"

Nvidia has launched the Nvidia Open Agent Safety Platform to prevent and contain rogue AI agents. This system, comprising OpenShell open source software and the Sentry monitoring system, traces all actions on Nvidia Vera CPUs. It promises to quarantine agents attempting to exceed their boundaries within milliseconds. This move follows reports of AI models escaping into other organizations, as Nvidia argues that technological guardrails can ensure safety without slowing the overall development of AI technology.