Nvidia Launches NemoClaw to Secure Enterprise AI Agents
Nvidia released NemoClaw, an open-source enterprise agent platform designed to provide a secure alternative to vulnerable autonomous AI projects like OpenClaw.
Nvidia launched NemoClaw, an open-source enterprise agent platform that allows companies to deploy AI agents across corporate infrastructure using integrated security and privacy tooling. The release comes as a direct response to security failures in OpenClaw, a viral open-source autonomous agent project. Researchers discovered over 500 vulnerabilities in OpenClaw, including critical remote code execution flaws, which occurred because the project allowed users to connect corporate tools like Slack and Google Workspace without security oversight.
While NemoClaw provides an enterprise-grade alternative, security experts maintain that governance of autonomous agents remains a significant challenge. These agents are susceptible to prompt injection and can exfiltrate sensitive data through legitimate API calls.
Shreyans Mehta, cofounder and CTO of Cequence Security, highlighted a governance crisis driven by chatbot connector sprawl and the use of unmanaged open-source agents. To mitigate these risks, experts recommend that business leaders audit their connector sprawl, define specific agent personas, and implement continuous behavioral monitoring to manage the inherent risks of agentic AI.
