Nvidia Launches AI Agent Safety Platform with Hardware-Based Monitoring

Nvidia has launched a new security platform called the Nvidia Open Agent Safety Platform to monitor and control AI agents. The platform combines software-level controls with hardware-based monitoring to restrict permissions, monitor activity, and isolate systems that behave outside defined rules.

The platform is designed for AI agents that can write code, use tools, access data, and operate across enterprise systems with limited human supervision.

Nvidia CEO Jensen Huang has framed the challenge around practical security controls rather than relying solely on an AI model to regulate itself. He believes it's an engineering problem and not solvable otherwise.

The platform is centered on two components: Nvidia OpenShell and Nvidia Sentry. OpenShell provides a sandboxed runtime where developers can establish policies governing an agent’s access to files, credentials, processes, and external networks.

Nvidia has made OpenShell available as open-source software. The company has also announced the platform alongside support from more than 100 organizations across the technology and enterprise sectors.