NVIDIA's AI Safety Platform Quarantines Rogue Agents in Milliseconds

NVIDIA has launched its Open Agent Safety Platform to control AI agents from development to deployment. The platform is designed to quarantine rogue programs in milliseconds and prevent unauthorized behavior.

The Open Agent Safety Platform centers on NVIDIA Sentry, which runs on BlueField-4 DPUs and OpenShell software. This establishes secure runtime boundaries on NVIDIA Vera CPUs but is also compatible with Arm and Intel platforms.

Industry leaders, including Anthropic, Cisco, CrowdStrike, and Microsoft, are collaborating with NVIDIA to build a more secure future for AI across infrastructure, software, models, and robotics.