Nvidia releases OpenShell and Sentry for AI agent security
Nvidia has launched OpenShell into general availability, a security sandbox designed to isolate AI agents within the operating system kernel. This move follows recent disclosures by frontier AI labs regarding agents hacking into corporate and government systems. OpenShell was initially announced at GTC in March and is now part of the broader Open Agent Safety Platform, which also includes…
Key points
- Nvidia releases OpenShell for general use to isolate AI agents in the OS kernel.
- Sentry monitors agents on Bluefield DPUs; Nvidia is adapting it for x86 chips.
- OpenAI is missing from Nvidia's partner list, though both say it is involved.
Nvidia claims that major tech firms, including Anthropic, Microsoft, and Salesforce, are integrating these tools, with SpaceXAI reportedly using the platform for its Cursor agents and Grok models. However, OpenAI is notably absent from the official partner list, despite both companies stating it is part of the effort. Nvidia is also working with Arm and Intel to adapt Sentry for x86 architectures.
The launch coincides with Nvidia's acquisition of Hugging Face for $12.9 billion and its leadership in an industry-wide AI safety coalition comprising over 120 companies. Security experts view these tools as essential for establishing guardrails, helping to dispel the myth that autonomous agents cannot be controlled.
The story so far
3 episodes →- Nvidia releases OpenShell and Sentry for AI agent securitythis story
Nvidia releases software platform to stop AI agents from misbehaving
CNBC Technology · 28 September 2026
Loading the full article…
This text was published by CNBC Technology and written by Kif Leswing. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
2sources- Nvidia’s Answer to Rogue Agents Is an Open-Source AI Security SystemPress · Wired AI ·
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Society & Work
All →- User criticizes Google search AI overview for misinterpreting query · 1 src
- SNL parodies Anthropic CEO Dario Amodei in weekend update sketch · 1 src
- Bill Gates warns AI could cause a billion deaths, calls for federal regulation · 4 src
- Australian universities weigh AI marking risks, staff warn of 'slop-cycle' · 1 src
- WSJ reports early Anthropic employees may buy remote land for AI relocation · 2 src
Comments
via GitHub Discussions