NVIDIA Launches NemoClaw to Add Security and Privacy Controls to OpenClaw

NVIDIA

Mar 19, 2026 · Updated Apr 25, 2026

NVIDIA NemoClaw, an open-source stack announced at GTC, adds policy-based privacy and security guardrails to OpenClaw. It bundles Nemotron models and the OpenShell runtime in a single install for running safer, always-on AI agents locally.

NVIDIA NemoClaw bundles Nemotron local models and the OpenShell runtime into a single install that layers privacy and security controls on top of OpenClaw, the open-source coding agent platform. OpenShell enforces policy-based guardrails that control how agents behave and handle data, while NemoClaw evaluates available compute resources to run Nemotron models locally for enhanced privacy. The stack is open source and available now in early preview.

NemoClaw uses NVIDIA Agent Toolkit software to secure OpenClaw and includes a privacy router that connects agents to cloud-based frontier models within defined guardrails. NVIDIA describes OpenClaw as the operating system for personal AI, and NemoClaw layers security and privacy controls on top to support secure, always-on operation.

Point an existing OpenClaw setup at the NemoClaw stack to add guardrails and local model inference across GeForce RTX PCs, RTX PRO workstations, DGX Station, and DGX Spark.

View the full update on nvidia.com

NVIDIA Newsroom

@nvidianewsroomMar 16

#NVIDIAGTC news: NVIDIA announces NemoClaw for the OpenClaw agent platform. NVIDIA NemoClaw installs NVIDIA Nemotron models and the NVIDIA OpenShell runtime in a single command, adding privacy and security controls to run secure, always-on AI assistants. https://t.co/v1LtP3c8ZW

160

View on X

Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →

See all AI news & updates from NVIDIA →

Keep reading

NVIDIA Open-Sources OpenShell to Secure Enterprise AI Agents with Kernel-Level Isolation

NVIDIA released OpenShell, an open-source security runtime that wraps autonomous AI agents in isolated sandboxes to prevent unauthorized data access or system actions. By using kernel-level isolation and declarative YAML policies, the tool provides the security layer required for production-grade enterprise agent deployments.

OpenClaw and NVIDIA release security dataset for 67,000 agent skills

OpenClawJun 1

OpenClaw and NVIDIA release security dataset for 67,000 agent skills

OpenClaw and NVIDIA have open-sourced a dataset of security scans for 67,453 skills on the ClawHub registry. The findings reveal that traditional malware scanners and new agentic-risk tools rarely agree on what makes a skill dangerous, highlighting a critical verification gap for autonomous agents.

Zhipu AIMar 31

Z.ai Launches AutoClaw for One Click Local AI Agent Deployment

Z.ai released AutoClaw, a local setup tool for the OpenClaw assistant framework that requires no API keys and keeps all data on-device. By pairing the tool with the new GLM-5-Turbo model, users can run complex multi-step agentic workflows entirely within their own infrastructure.