Follow ZDNET: Add us as a preferred source on Google.
ZDNET’s key takeaways
- Nvidia’s NemoClaw aims to make OpenClaw agents more secure.
- OpenClaw agents are highly capable, but come with risks.
- The company also launched a multi-lab open-source model coalition.
“What’s your OpenClaw strategy?” Nvidia CEO Jensen Huang asked rhetorically to the crowd at Nvidia GTC, the company’s annual AI conference, on Monday.
The company is full steam ahead on AI agents — and it’s hoping its latest release can fix OpenClaw’s security problem. During the keynote, Huang announced Nvidia’s new NemoClaw stack, which is built to shore up the OpenClaw agent platform, the viral open-source assistant framework that has impressed users with its autonomous capabilities.
Also: Why buying into Moltbook and OpenClaw may be Big Tech’s most dangerous bet yet
OpenClaw does not run its own model; what sets it apart is how it leverages the sometimes-differing strengths of Anthropic’s Claude and OpenAI’s ChatGPT, while running locally on a user’s device to take action on its own. That level of autonomous capability and access to user information also poses a significant security risk, which has been its primary drawback.
Nvidia, however, believes OpenClaw is the foundation of personal AI. The company, which has been working with OpenClaw founder Peter Steinberger, referred to the agent platform as history’s most important software release during a Sunday briefing before the conference. Nvidia said NemoClaw can optimize OpenClaw for privacy and security using Nvidia’s Agent Toolkit, an open-source library for managing teams of AI agents.
How it works
NemoClaw installs Nvidia’s OpenShell, a new open-source runtime that keeps agents safer to use by enforcing an organization’s policy-based guardrails. OpenShell keeps models sandboxed, adds data privacy protections and additional security for agents, and makes them more scalable.
“This provides the missing infrastructure layer beneath claws to give them the access they need to be productive, while enforcing policy-based security, network, and privacy guardrails,” Nvidia said in the announcement. The company built OpenShell with security companies like CrowdStrike, Cisco, and Microsoft Security to ensure it is compatible with other cybersecurity tools.
Also: Is your AI agent a security risk? NanoClaw wants to put it in a virtual cage
Nvidia said NemoClaw can be installed in a single command, runs on any platform, and can use any coding agent, including Nvidia’s own Nemotron open model family, on a local system. Through a privacy router, it allows agents to access frontier models in the cloud, which unites local and cloud models to help teach agents how to complete tasks within privacy guardrails, Nvidia explained.
Automating work
Nvidia seems to be hoping that the additional security can make OpenClaw agents more popular and accessible, with less risk than they currently carry. The bigger picture here is how NemoClaw could give companies the added peace of mind to let AI agents complete actions for their employees, where they wouldn’t have previously.
In the release, Nvidia noted that advancing enterprise AI agents will “speed a generational shift in software and knowledge work,” and that the next phase of enterprise software will be all about specialized agents. As my colleague Tiernan Ray explains, Nvidia’s new Vera Rubin infrastructure is meant to back up this agentic AI drive, and will drive down costs in the process, according to the company.
Also: Nvidia wants to own your AI data center from end to end
Huang said during the keynote that he believes OpenClaw arrived at exactly the right time for the software industry, and that it spells a new path: agents-as-a-service rather than software-as-a-service (SaaS).
How to try NemoClaw
NemoClaw is currently in preview. Starting today, developers can access Nvidia’s Agent Toolkit and OpenShell here, use OpenShell with LangChain, or download it from GitHub directly to run locally. Enterprises can create and deploy AI agents via cloud providers like AWS, Google Cloud, and Microsoft Azure, among others.
A new open-source initiative
Nvidia also launched the Nemotron Coalition, a collaboration between several model developers and AI labs aimed at advancing open-source AI through shared resources and compute.
Also: Why enterprise AI agents could become the ultimate insider threat
The coalition includes Mira Murati’s startup Thinking Machines Lab, Perplexity, Cursor, and Mistral AI, among others. To start, Mistral and Nvidia will co-develop an open model trained on Nvidia DGXTM Cloud and open-source the result, which will also be the foundation for Nvidia’s forthcoming Nemotron 4 model family. Other coalition members will support the model with data and testing.
“By combining forces, the coalition aims to accelerate progress on AI models, expanding intelligence beyond any single model and
strengthening a vibrant open ecosystem while making model development more efficient so organizations can build, specialize, and innovate on a shared, open foundation,” the company said.
Also: Why AI is both a curse and a blessing to open-source software – according to developers
The move is an emphatic investment in making cutting-edge AI models available to everyone. As with all open-source projects, the initiative should pool expertise to effectively democratize competitive AI tools that individual developers can then adapt further to their local contexts or use cases.
“AI reaches its full potential when it works in every language and for every community,” said Pratyush Kumar, cofounder and CEO of Sarvam, another founding coalition member, in the release. “Open models make this possible by giving builders the freedom to adapt frontier capabilities to real-world needs.”