NVIDIA has launched OpenShell, a tool designed to secure autonomous AI agents. This tool addresses security challenges by implementing security at the infrastructure level.
OpenShell Provides a Secure Environment for Autonomous AI Agents
OpenShell is part of the NVIDIA Agent Toolkit and functions as an open-source runtime for autonomous agents. It isolates each agent in its own sandbox, separating application operations from policy enforcement at the infrastructure level. Security policies thus become inaccessible to the agents, preventing them from bypassing these policies or leaking sensitive data.
This is especially important as autonomous agents can perform complex tasks and continuously improve themselves. By using OpenShell, organizations can define and monitor how autonomous systems operate, simplifying compliance with security standards. OpenShell is designed to collaborate with security partners such as Cisco and Microsoft Security, making it easier to implement security across the entire enterprise.
Source: NVIDIA
Read also: Helion in Talks with OpenAI About Fusion Power
