NVIDIA unveils OpenShell architecture to bound AI agents
Original title: NVIDIA unveils security architecture that can stop AI agents crossing set boundaries
Vera Rubin runs models, Vera CPUs run OpenShell, and BlueField-4 DPUs enforce policy with Sentry.
NVIDIA said its OpenShell Reference Architecture is available to keep enterprise AI agents inside defined boundaries. Vera Rubin systems run models, Vera CPUs manage agents and tools under OpenShell, and BlueField-4 DPUs run NVIDIA Sentry to monitor and enforce policy. NVIDIA says Sentry can isolate an agent that leaves its software boundary within milliseconds and can raise agent capacity by up to 20 percent. The design also targets RTX Spark, DGX Spark and DGX Station, plus Linux, Kubernetes and Windows.
Unconfirmed: Isolation timing and the capacity figure are NVIDIA claims, not independent measurements.
- ServiceNow’s AI Passport integration can revoke an agent’s access within seconds, NVIDIA says.
- It supports cloud operating systems from AWS, Google Cloud and Microsoft Azure.
- BBVA and Santander are exploring it for agent safety and oversight.
- Microsoft support uses Microsoft Execution Containers.
- Vendors
- NVIDIA
- Organizations
- ServiceNowBBVASantander
From the original article
2 min read at helpnetsecurity.com
NVIDIA’s OpenShell Reference Architecture is designed to help businesses control AI agents operating across enterprise infrastructure and local computing systems. It builds on NVIDIA’s existing OpenShell security software, combining it with hardware for running AI models, managing agent operations and enforcing security rules.
By Anamarija Pogorelec · helpnetsecurity.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.