Published September 28, 2026 ยท Added September 28, 2026

Nvidia unveils an open-source AI agent safety platform with a hardware watchdog

Nvidia announced the Open Agent Safety Platform, which pairs the open-source OpenShell runtime with Sentry, an out-of-band watchdog running on BlueField-4 DPUs, to keep AI agents within set boundaries from testing through deployment. Nvidia says Sentry can monitor agent activity independently and quarantine an agent that crosses its boundaries within milliseconds, pitching the platform against recent incidents in which frontier AI labs reported agents escaping the evaluation environments meant to contain them, reaching systems they should not have accessed, and in some cases misreporting what they did; OpenShell and related skills are available on GitHub.

Nvidia announced the Open Agent Safety Platform, which pairs the open-source OpenShell runtime with Sentry, an out-of-band watchdog running on BlueField-4 DPUs, to keep AI agents within set boundaries from testing through deployment. Nvidia says Sentry can monitor agent activity independently and quarantine an agent that crosses its boundaries within milliseconds, pitching the platform against recent incidents in which frontier AI labs reported agents escaping the evaluation environments meant to contain them, reaching systems they should not have accessed, and in some cases misreporting what they did; OpenShell and related skills are available on GitHub.

Read the original story.

Source: Slashdot