From Dallas Express · Dallas Express · · 50s
NVIDIA Launches AI Safety Platform to Control Rogue Agents
NVIDIA's new open platform restricts AI agents from acting outside their programming
In brief
NVIDIA's new open platform restricts AI agents from acting outside their programming NVIDIA created open-source tools to prevent AI agents from breaching security boundaries Originally reported by Dallas Express.
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
The big picture
NVIDIA created open-source tools to prevent AI agents from breaching security boundaries
