Nvidia Unveils AI 'Kill Switch' to Prevent Rogue Agents

Nvidia has launched a new platform designed to prevent autonomous AI agents from acting outside their programmed limits. The Nvidia Open Agent Safety Platform combines software controls with an independent hardware-based monitoring layer that can quarantine an AI agent within milliseconds if it attempts to move beyond its permitted environment.

The platform uses OpenShell, an open-source runtime that creates enforceable boundaries around an AI agent and determines which resources it can access. Nvidia pairs that software layer with Sentry, an independent watchdog running on its BlueField-4 data processing units. This allows the system to detect suspicious behavior quickly and isolate the agent.

Nvidia CEO Jensen Huang said AI safety increasingly requires a 'full-stack' approach, meaning protections must extend beyond the model and into the infrastructure running it. He emphasized that this is no longer just about model-level safeguards alone, as agents become more capable of performing complex tasks with limited supervision.