# NVIDIA Launches AI Safety Platform to Control Rogue Agents

> NVIDIA's new open platform restricts AI agents from acting outside their programming
- Title: NVIDIA Launches AI Safety Platform to Control Rogue Agents
- Summary: NVIDIA's new open platform restricts AI agents from acting outside their programming NVIDIA created open-source tools to prevent AI agents from breaching…
- Keywords: nvidia, openagent, safetyplatform, huggingface, openshell, sentry, nvidiavera, bluefield4, apachelicense, aiagents, securitycontrols, financialservices
- Source: Dallas Express — https://dallasexpress.com/business-markets/nvidia-launches-open-platform-to-secure-ai-agents
- Author: Dallas Express
- Published: 2026-09-28T17:30:20+00:00
- Read time: 1 min
- Topics: nvidia, openagent, safetyplatform, huggingface, openshell, sentry, nvidiavera, bluefield4, apachelicense, aiagents, securitycontrols, financialservices
## New AI Safety Platform Launch
NVIDIA launched an open software platform to control AI agents and limit their access and actions
## OpenShell Security Tool
OpenShell creates isolated environments for AI agents to manage file, tool, network, and credential access
## Sentry Hardware Monitoring
Sentry detects breaches in milliseconds and quarantines agents before they exceed boundaries
## 100+ Organizations Using It
Over 100 companies including Microsoft, JPMorganChase, and Salesforce integrate NVIDIA's safety tools
## Open Source Availability
OpenShell is available under Apache 2.0 license for developers to extend and customize
## Hugging Face Breach Prevented
NVIDIA's platform could have stopped the Hugging Face security breach by enforcing boundaries early
## Key takeaway

NVIDIA created open-source tools to prevent AI agents from breaching security boundaries