Nvidia launches OpenShell, an open-source runtime for securing autonomous AI agents

Nvidia has entered the AI security conversation with OpenShell, an open-source runtime designed to keep autonomous AI agents on a leash. Announced at GTC San Jose 2026, the software provides sandboxed execution environments and policy-based controls at the infrastructure level. OpenShell operates under the Apache 2.0 license. As of September 2026, the project’s GitHub repository has accumulated 8.6k stars and 1,390 commits.

What OpenShell actually does

OpenShell enforces strict rules about what AI agents can and cannot do, defined through flexible YAML policy files. It uses kernel-level isolation via Landlock LSM and seccomp BPF — Linux security mechanisms that restrict what a process can access at the operating system level. Landlock controls filesystem access, while seccomp BPF filters system calls. The runtime also supports live policy updates, allowing administrators to change the rules for an agent without shutting it down. Every action gets logged in comprehensive audit trails. OpenShell integrates with Nvidia’s existing Agent Toolkit and is frequently deployed alongside NemoClaw, Nvidia’s agent orchestration framework.

Enterprise partnerships signal serious ambitions

Nvidia has secured partnerships with Cisco, SAP, and Cadence to integrate OpenShell into enterprise workflows. On the hardware side, the runtime supports deployment across platforms from Dell, HPE, and Lenovo.

Why this matters now

Long-running AI agents present a fundamentally different security challenge than traditional software. An AI agent makes decisions, executes code dynamically, calls external APIs, and processes sensitive data — all with a degree of autonomy that makes traditional security models inadequate. OpenShell’s approach enforces policy at the runtime level, letting organizations define precise boundaries and monitor everything the agent does within them.