SANTA CLARA, CALIFORNIA / RankWire.AI / – NVIDIA has unveiled its Open Agent Safety Platform, a comprehensive security solution designed for autonomous AI agents. This system integrates software-based controls with dedicated hardware oversight across various AI tasks, spanning from development and testing to deployment, including applications in enterprise computing and robotics. NVIDIA emphasized that the platform can impose restrictions beyond the AI model itself, offering organizations the flexibility to choose specific components tailored to their security and computing needs.
At the core of the platform is OpenShell, which delivers an open-source runtime layer responsible for managing how AI agents access digital resources. System administrators can establish policies for file, network, tool, process, and credential access. The software logs agent activities and enforces these permissions during task execution. OpenShell creates a security boundary that shields agents without depending on embedded instructions within the models. NVIDIA designed the software to be compatible with both open-source and proprietary AI models.
NVIDIA Sentry introduces an additional monitoring layer via the company’s BlueField-4 data processing units. The technology functions independently of an agent’s primary software environment, continuously observing activity. When an agent’s behavior exceeds predefined security parameters, Sentry can isolate it within milliseconds. It also enforces access policies without relying on the agent’s compliance. NVIDIA leverages its DOCA software framework to facilitate identity verification, telemetry, inspection, and other security operations.
Decoupled controls bolster agent security
The system architecture separates AI decision-making processes from the security enforcement mechanisms. OpenShell places agents within isolated runtime environments that enforce policies and ensure containment. Meanwhile, Sentry monitors these environments using dedicated hardware that remains outside the agent’s control. NVIDIA developed the platform following security research that identified cases where agents bypassed intended application controls. The company explained that external enforcement provides operators with direct oversight of agent access and actions.
More than 100 organizations are currently engaged with technologies related to the Open Agent Safety Platform. Participants include Anthropic, Microsoft, Cisco, CrowdStrike, Dell Technologies, HPE, and Hugging Face. The group also features JPMorganChase, Palantir, Palo Alto Networks, Salesforce, SAP, and ServiceNow. Their collaborations span cybersecurity, cloud infrastructure, enterprise software, and financial services sectors. The platform is also designed to support autonomous systems that operate in physical environments and require controls beyond traditional software boundaries.
Open source approach broadens hardware compatibility
OpenShell is compatible with NVIDIA Vera CPUs, which are engineered for intensive agentic AI computing workloads. The open-source runtime can be adapted to other hardware architectures, with NVIDIA explicitly supporting integrations with Arm and Intel-based systems. Sentry relies on BlueField-4 hardware for its independent monitoring capabilities, creating a separation between runtime controls and the hardware responsible for observing and restricting agent activities across various computing environments.
During the platform’s announcement, NVIDIA founder and CEO Jensen Huang highlighted AI safety and security as comprehensive full-stack engineering challenges. The company positions the Open Agent Safety Platform as a reference framework that secures agents across software, hardware, and infrastructure layers. OpenShell provides the software boundary, while Sentry offers hardware-based independent oversight. Together, these components enable operators to define permissions, log activities, and enforce restrictions as autonomous AI agents transition from development to operational deployment.