Nvidia launches open platform to keep AI agents inside safety boundaries
Nvidia announced the Open Agent Safety Platform, an open software platform and reference system design for governing AI agents from testing to deployment. Nvidia says it includes OpenShell secure runtime software that traces actions and enforces policy, plus a Sentry watchdog that runs on BlueField-4 DPUs to monitor behavior and quarantine agents that break boundaries in milliseconds. The effort is backed by more than 100 organizations including Anthropic, Microsoft, CrowdStrike, Hugging Face and SpaceXAI, and the software is available through Nvidia developer resources and GitHub.