NVIDIA Launches In-Silicon Security Platform to Monitor and Control Autonomous AI Agents
NVIDIA launched Open Agent Safety Platform, pairing OpenShell sandboxing with BlueField-4 hardware controls for AI agents.
NVIDIA launched the Open Agent Safety Platform to monitor and constrain autonomous AI agents that can write code, call APIs, and use credentials. The design pairs the open-source OpenShell runtime—Gateway, Supervisor, and isolated Sandbox—with NVIDIA Sentry on BlueField-4 DPUs for out-of-band enforcement. OpenShell keeps credentials outside the agent, routes traffic through policy points, and verifies permission changes before they expand access. NVIDIA says OpenShell works with Codex, Claude Code, Pi, and Hermes, with early use at Cadence, Slack, and Gecko Robotics.
- OpenShell sandboxes agents with kernel isolation and limits files, APIs, credentials, and destinations.
- Policy changes are checked against operator boundaries and can require human review.
- NVIDIA Sentry on BlueField-4 DPUs enforces controls outside the host and the agent.
- NVIDIA cites early use at Cadence, Slack, and Gecko Robotics.
From the original article
3 min read at gbhackers.com
NVIDIA has launched its Open Agent Safety Platform, a security architecture designed for out-of-band monitoring, runtime policy enforcement, and hardware-backed control for autonomous AI agents. This platform combines the open-source NVIDIA OpenShell runtime with NVIDIA Sentry protections on BlueField-4 data processing units (DPUs), aiming to prevent agents from exceeding their authorized access or operating limits.
By Divya · gbhackers.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.