-+ 0.00%
-+ 0.00%
-+ 0.00%

On September 28, local time, Nvidia announced the launch of the NVIDIA Open Agent Safety Platform, an open software platform and reference system design designed to enhance AI security throughout the entire process from intelligent testing to deployment, and to achieve full-stack governance and control of the software, hardware, computing and robotic systems running the intelligent. The platform includes NVIDIA OpenShell secure runtime software to set enforceable boundaries for agents running on CPUs. OpenShell is now widely available and runs at a very low cost on the first NVIDIA Vera CPU built specifically for intelligent AI. As open source software, OpenShell can also be extended to third-party computing platforms such as ARM and Intel. The reference system design includes NVIDIA Sentry. This is an out-of-band monitoring program running on the NVIDIA BlueField-4 DPU that continuously monitors agent behavior and independently enforces security policies within the chip. If an agent tries to cross the software boundary, Sentry can isolate and stop it within milliseconds.

Zhitongcaijing·09/28/2026 23:57:05
Listen to the news
On September 28, local time, Nvidia announced the launch of the NVIDIA Open Agent Safety Platform, an open software platform and reference system design designed to enhance AI security throughout the entire process from intelligent testing to deployment, and to achieve full-stack governance and control of the software, hardware, computing and robotic systems running the intelligent. The platform includes NVIDIA OpenShell secure runtime software to set enforceable boundaries for agents running on CPUs. OpenShell is now widely available and runs at a very low cost on the first NVIDIA Vera CPU built specifically for intelligent AI. As open source software, OpenShell can also be extended to third-party computing platforms such as ARM and Intel. The reference system design includes NVIDIA Sentry. This is an out-of-band monitoring program running on the NVIDIA BlueField-4 DPU that continuously monitors agent behavior and independently enforces security policies within the chip. If an agent tries to cross the software boundary, Sentry can isolate and stop it within milliseconds.