Nvidia recently launched an open software platform aimed at providing companies with tighter control over AI agents. This news comes at a time when industry concerns over the safety of autonomous models continue to be raised.
The Open Agent Safety Platform combines OpenShell, an open-source runtime software that sets boundaries around agents, with Sentry, a hardware-based monitoring system that is powered by Nvidia’s BlueField-4 DPUs computing platform to detect and quarantine agents that breach software limits “in milliseconds”.
The company explained Sentry enforces security policies and operates in real-time from an isolated trust domain that remains invisible to both AI agents and attackers.
According to the chipmaker, OpenShell tracks activity as agents run on its Vera CPUs. The software can also be extended to third-party processors from Arm and Intel.
Nvidia launched the tools with dozens of partners including Anthropic, Mistral, Microsoft, SpaceXAI, Hugging Face, Accenture, Palantir and Perplexity. It revealed Anthropic has already integrated OpenShell and BlueField with its Claude Managed Agents, while SpaceXAI is deploying the platform with Cursor coding agents and Grok AI models.
Infrastructure and energy players including Red Hat, Siemens Energy and Schneider Electric are also among those adopting the platform.
Source: Mobile World Live
Image Credit: Nvidia
Source: Tahawul Tech