European Investor
28 Sep 2026, 18:55
IBM Backs NVIDIA Open Agent Safety Platform to Strengthen Enterprise AI Agent Security
IBM is supporting NVIDIA’s new Open Agent Safety Platform, an open software platform and reference design aimed at securing and governing AI agents as they take on more autonomous work across enterprise environments.
IBM said its contribution spans agent identity, credentials, infrastructure, storage and hybrid cloud. IBM Agent Identity and HashiCorp Vault are integrating with NVIDIA OpenShell to give agents verified identities and tightly controlled access, while IBM Identity Protection is designed to help organizations discover and monitor agents across their environments.
At the infrastructure level, IBM Storage is integrating NVIDIA BlueField-4 into IBM Fusion, while Red Hat OpenShift will support governed agent deployments across hybrid cloud environments. Red Hat is also working with NVIDIA to integrate BlueField and OpenShell for zero-trust security, runtime isolation and real-time threat detection.
NVIDIA’s approach places agents inside a policy-governed runtime and uses its Sentry framework and BlueField hardware to enforce access policies, verify agent identity and monitor behavior. IBM said the goal is to make agent actions attributable, least-privileged and auditable, even if the underlying workload is compromised.
The broader focus is shifting from what AI models can do to what enterprises can safely allow autonomous agents to do. IBM also joined NVIDIA as a founding member of the Open Secure AI Alliance, which is governed by the Linux Foundation.
IBM is supporting NVIDIA’s new Open Agent Safety Platform, an open software platform and reference design aimed at securing and governing AI agents as they take on more autonomous work across enterprise environments.
IBM said its contribution spans agent identity, credentials, infrastructure, storage and hybrid cloud. IBM Agent Identity and HashiCorp Vault are integrating with NVIDIA OpenShell to give agents verified identities and tightly controlled access, while IBM Identity Protection is designed to help organizations discover and monitor agents across their environments.
At the infrastructure level, IBM Storage is integrating NVIDIA BlueField-4 into IBM Fusion, while Red Hat OpenShift will support governed agent deployments across hybrid cloud environments. Red Hat is also working with NVIDIA to integrate BlueField and OpenShell for zero-trust security, runtime isolation and real-time threat detection.
NVIDIA’s approach places agents inside a policy-governed runtime and uses its Sentry framework and BlueField hardware to enforce access policies, verify agent identity and monitor behavior. IBM said the goal is to make agent actions attributable, least-privileged and auditable, even if the underlying workload is compromised.
The broader focus is shifting from what AI models can do to what enterprises can safely allow autonomous agents to do. IBM also joined NVIDIA as a founding member of the Open Secure AI Alliance, which is governed by the Linux Foundation.
