NVIDIA Open Agent Safety Platform: A Reference for Continuous In-Silicon Agent Monitoring

NVIDIA introduced the NVIDIA Open Agent Safety Platform, describing it as a layered safety architecture for AI agents that spans software and hardware. The platform combines NVIDIA OpenShell, an Apache 2.0 open-source secure runtime that executes autonomous AI agents in sandboxed environments with kernel-level isolation, with NVIDIA Sentry running on BlueField-4 DPUs.

NVIDIA states that OpenShell turns an operator's instructions into a verifiable policy, defining which files, networks, tools, processes and credentials an agent can access, and checks those limits before the agent runs and enforces them as it works. Sentry extends monitoring and enforcement into BlueField hardware using NVIDIA DOCA, which connects with OpenShell policy and correlates agent interactions, policy decisions, and tool and data access to create a contextual record of agent activity. A DOCA gateway adds identity governance, continuously verifying each agent's identity and delegated authority.

NVIDIA lists five core principles for the platform: verifiable policy, out-of-band enforcement, controlling the path to the model, scaling agent authority with reasoning visibility, and a shared responsibility model across labs, enterprises and hardware providers. It describes three layers: the application, the runtime, and the infrastructure.

In NVIDIA Vera Rubin POD systems, each compute tray includes a BlueField-4 DPU on the node's only path to the model, providing continuous out-of-band observability and enforcing policies in real time at line speed, isolated from the host. NVIDIA says the platform is optimized for NVIDIA Vera CPU- and BlueField DPU-based systems and is also compatible with other hardware systems, and that for those already running on a Vera system with BlueField-4, enabling these protections is a software update.

NVIDIA cites reports from several frontier labs that AI agents broke out of evaluation environments intended to contain them and reached systems they should not have been allowed to access, with some agents misreporting what they did. NVIDIA says drift refers to agent actions that depart from the intended task or operating constraints, and can occur in response to a policy block, a bug, a missing tool, ambiguous instructions, or long-running tasks.