Nvidia Launches Open Agent Safety Platform With 100-Plus Partners to Control AI Agent Behavior

Nvidia unveiled the Open Agent Safety Platform on September 28, 2026, an open software initiative backed by more than 100 organizations aimed at securing AI agents throughout their lifecycle, from development testing through production deployment. The platform combines two core components: OpenShell, an open-source runtime enforcing agent boundaries on standard CPUs, and Sentry, a hardware-based watchdog running on Nvidia BlueField-4 data processing units (DPUs) that monitors and quarantines non-compliant agents in milliseconds.