Lenovo and NVIDIA Strengthen Security for the Agentic AI Era

AI agents are expanding enterprise capabilities but require robust security and control across all operational layers. Lenovo and NVIDIA are addressing this need by supporting the NVIDIA Open Agent Safety Platform, an open reference design designed to enforce governance and security for AI agents in cloud, hybrid, on-premises, and air-gapped environments. The platform integrates with Lenovo Hybrid AI Advantage solutions to provide full-stack oversight, ensuring consistent security policies regardless of deployment location.

The NVIDIA Open Agent Safety Platform employs a zero-trust architecture to restrict agent access by default, enforcing boundaries on agent actions and interactions within enterprise systems. It combines NVIDIA OpenShell, a secure runtime for governing agent execution and inference routing, with NVIDIA Sentry, an out-of-band watchdog running on NVIDIA BlueField-4 DPUs. NVIDIA Sentry monitors agent activity and enforces security policies in silicon, capable of quarantining agents within milliseconds.

Lenovo Hybrid AI Factory solutions will integrate NVIDIA OpenShell, which operates on NVIDIA Vera CPU and other processor platforms, alongside NVIDIA Sentry on NVIDIA BlueField-4 DPUs. This collaboration builds on Lenovo and NVIDIA’s commitment to advancing AI security through openness and industry-wide cooperation, as demonstrated by their participation in the Open Secure AI Alliance.

The initiative aims to give organizations greater control over how AI agents interact with systems, data, and applications while maintaining flexibility across models and deployment environments. Lenovo emphasizes that security, governance, and customer control must be embedded from the start as AI agents take on more autonomous tasks, requiring scalable safeguards across the technology ecosystem.