NVIDIA Launches Open Platform to Put Safety Controls Around AI Agents
The NVIDIA Open Agent Safety Platform combines open-source software with hardware-based security to provide governance and control across the full agent stack.
NVIDIA has launched an open software platform and reference system designed to secure AI agents from testing through deployment, as companies increasingly give autonomous systems access to sensitive data, applications and real-world infrastructure.
The NVIDIA Open Agent Safety Platform combines open-source software with hardware-based security to provide governance and control across the full agent stack.
The Left Shift Bureau
The platform includes NVIDIA OpenShell, a secure runtime environment for AI agents, and NVIDIA Sentry, a reference system designed to continuously monitor agent behaviour.
OpenShell creates a runtime boundary around agents, allowing organisations to trace their actions and enforce policies while they operate. The software runs on NVIDIA Vera CPUs but is open source and can be extended to third-party compute platforms, including those based on Arm and Intel architectures.
Sentry adds another layer of protection by operating independently of the agent. Running on NVIDIA BlueField-4 data processing units, the system continuously monitors agent activity and can quarantine an agent attempting to move outside its defined boundaries within milliseconds.
“As we continue to discover the frontier of AI capabilities, we must accelerate discovery at the frontier of AI safety. Safety and security require full-stack engineering. NVIDIA Open Agent Safety Platform brings together industry, researchers and public-sector organizations to share best practices, align on evaluation methods and foster international cooperation,” said Jensen Huang, NVIDIA Founder and CEO.
The platform is also gaining support across the technology industry from Anthropic, Salesforce, SAP, Scale AI, Microsoft, Hugging Face, CrowdStrike, Cisco and more than 100 other organisations.
Pritam Bordoloi
Anthropic is integrating OpenShell and BlueField capabilities with its Claude Managed Agents, while Salesforce has integrated OpenShell with Slack to give teams visibility into agent activity and audit events, as well as the ability to approve or reject requests for additional permissions.
“Companies are giving AI agents more of their most important work, and they need to direct and verify what those agents do, especially in sensitive environments. Claude Managed Agents gives companies a clear view of what each agent is doing, and NVIDIA’s platform adds another layer of governance and control across hardware and software,” said Paul Smith, Anthropic Chief Commercial Officer.
The Left Shift Bureau
The Left Shift Bureau
The Left Shift Bureau
