Red Hat and NVIDIA Expand AI Agent Security Partnership

Red Hat and NVIDIA expanded their collaboration to secure autonomous AI agents and govern access to AI models, reports SiliconANGLE. Announced at GTC Berlin, the work adds security and infrastructure features to Red Hat AI Factory with NVIDIA across hybrid cloud environments.

Red Hat plans to integrate NVIDIA's OpenShell Reference Architecture into its AI platforms. OpenShell isolates agents in a sandbox, while NVIDIA Sentry uses BlueField data processing units and DOCA Argus software to monitor processes, files and network activity from a separate hardware security domain.

Security systems can use Argus alerts to isolate affected workloads. Red Hat says Sentry can quarantine a compromised agent within milliseconds, while moving monitoring and networking tasks to BlueField processors reduces demand on host CPUs.

The companies also introduced a reference design for model discovery, access policies, rate limits and subscription tracking. Red Hat Enterprise Linux for NVIDIA now supports Vera CPUs and is prepared for the upcoming Vera Rubin platform, while Nemotron models will receive joint support in vLLM and Red Hat AI Factory with NVIDIA.