NVIDIA Unveils Open Agent Safety Platform to Prevent Rogue AI Agents

NVIDIA has launched the Open Agent Safety Platform to ensure AI agents stay within their designated boundaries.

The platform includes NVIDIA OpenShell, an open-source runtime that creates a security boundary around autonomous agents and tracks their activities while enforcing policies.

NVIDIA Sentry is another key component of the platform, which runs on NVIDIA BlueField-4 DPUs and acts as an out-of-band watchdog to continuously monitor agent activity. If an agent attempts to move outside its software boundary, Sentry can quarantine and stop it in milliseconds.

The Open Agent Safety Platform has already gained the support of over 100 organizations, including Anthropic, Microsoft, Cisco, and Salesforce, which are integrating NVIDIA's technologies into their systems.