Nvidia releases Open Agent Safety Platform to monitor and govern agentic AI
Nvidia released an Open Agent Safety Platform that sandboxes AI agents with OpenShell software and BlueField DPU enforcement.
Nvidia released the Open Agent Safety Platform, pairing open-source OpenShell software with NVIDIA Sentry on BlueField-4 DPUs. OpenShell sandboxes agents on Vera CPUs, traces actions, and enforces policy, and can extend to Arm and Intel platforms. Sentry is an out-of-band watchdog Nvidia says can quarantine an agent that leaves its software boundary in milliseconds. Analysts said it only covers agents on infrastructure the enterprise controls, with IDC estimating under 25 percent of agentic security problems, and noted OpenAI, Amazon, and Google were absent from supporters.
- OpenShell traces agent actions and enforces policy on Nvidia Vera CPUs.
- Sentry on BlueField-4 DPUs can quarantine out-of-bounds agents in milliseconds.
- Controls do not cover unsanctioned SaaS, vendor, or attacker-supplied agents.
- IDC said the design addresses under 25 percent of agent security problems.
- OpenAI, Amazon, and Google were not listed among supporters.
- Vendors
- Nvidia
From the original article
6 min read at csoonline.com
Nvidia on Monday rolled out an agentic governance system called the Open Agent Safety Platform that combines software with out-of-band DPU-based silicon in a reference system design that it says will secure agents “from testing to deployment.” But while the Nvidia design’s silicon-based component provides some cybersecurity advantages, analysts argued that it cannot help with the vast majority of agentic problems.
csoonline.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.