# Nvidia introduces open-source platform to secure AI agents

- Published: 2026-09-28T09:00:00.000Z
- Updated: 2026-09-28T09:18:58.737Z
- Section: Development
- Topics: Security, Agents
- Page: https://aidailyjournal.com/en/n/2026/09/28/nvidia-introduces-open-source-platform-to-secure-ai-agents/
- Sources: [Wired](https://wired.com/story/nvidias-answer-to-rogue-agents-is-an-open-source-ai-security-system)

## Summary

Nvidia has released OpenShell for general availability, an open-source security environment that contains AI agents and isolates their activity in the operating system kernel. The company also introduced Sentry, an isolated security domain for chips that monitors long-running agents and can quarantine them when they try to exceed their boundaries; the software is designed for use with BlueField data processing units, and Nvidia is working with Arm and Intel on a version for the x86 architecture.

Nvidia positions the tools as part of the Open Agent Safety Platform and says it has security collaborations with dozens of companies. According to Wired, Anthropic, Salesforce, Scale AI, and SAP confirmed some level of OpenShell integration, but it is unclear whether every listed partner has adopted the tool. OpenAI was reportedly part of the effort but was left out of the announcement; neither company explained why.

## Why it matters (editorial interpretation)

For teams deploying long-running agents, combining kernel-level isolation with chip-level quarantine could offer more specific controls over what those systems do. Adoption still appears uneven: Wired says some companies confirmed integration, but it is unclear whether every listed partner uses OpenShell.

_Summary written by AI Daily Journal in its own words, based on the sources above. For details, read the original articles._