CrowdStrike and NVIDIA Expand Security for AI Agents

Summary

CrowdStrike and NVIDIA are collaborating on the NVIDIA Open Agent Safety Platform, which aims to add layered controls for AI agents and connect agent activity with enterprise security context in CrowdStrike’s Falcon platform.

Key points

The collaboration focuses on extending security controls across the software, runtime, and infrastructure layers where AI agents operate.
The platform combines NVIDIA technologies including OpenShell for runtime controls and Sentry for out-of-band monitoring and policy enforcement.
The article describes controls such as least privilege, isolation, identity verification, telemetry, and Zero Trust access.
CrowdStrike says Falcon will connect agent activity with security context across endpoint, identity, cloud, data, and AI.
The companies frame independent controls outside the agent itself as necessary because agents may access data, credentials, APIs, and critical systems.

Article Details

Event Type: Security partnership expansion and AI security platform introduction
Impact: CrowdStrike and NVIDIA describe extending independent security controls across AI agent software, runtime, and infrastructure, connecting agent activity with enterprise-wide security context through the Falcon platform. Intended protections include agent isolation, identity verification, policy enforcement, and Zero Trust access. The article states that NVIDIA Sentry can quarantine agents in milliseconds during testing and evaluation; it does not report deployment outcomes.

People

Bartley RichardsonAuthor of the article announcing the expanded CrowdStrike and NVIDIA collaboration.

Vendors

CrowdStrikeCrowdStrike and Anthropic Give Critical Infrastructure Defenders the AI Advantage NvidiaA Win for Defenders: CrowdStrike and NVIDIA Extend Security Across the AI Stack

Products

BlueField-4NVIDIA Sentry adds an independent watchdog running on NVIDIA BlueField-4 DPUs that uses NVIDIA DOCA for out-of-band, in-silicon monitoring of agent activity and security policy enforcement.CrowdStrike Falcon GuardianWith the recent launch of CrowdStrike Falcon® Guardian, we have extended protection across the AI attack surface, including runtime protection for AI agents.CrowdStrike Falcon Next-Gen SIEMAt the infrastructure layer, our existing integration of NVIDIA DOCA Argus telemetry with CrowdStrike Falcon® Next-Gen SIEM brings infrastructure-level telemetry into the Falcon platform.CrowdStrike Falcon platformThe CrowdStrike Falcon® platform connects that activity to the broader security context of the enterprise.NVIDIA DOCANVIDIA Sentry adds an independent watchdog running on NVIDIA BlueField-4 DPUs that uses NVIDIA DOCA for out-of-band, in-silicon monitoring of agent activity and security policy enforcement.NVIDIA DOCA ArgusAt the infrastructure layer, our existing integration of NVIDIA DOCA Argus telemetry with CrowdStrike Falcon® Next-Gen SIEM brings infrastructure-level telemetry into the Falcon platform.NVIDIA Open Agent Safety PlatformCrowdStrike and NVIDIA are collaborating on the NVIDIA Open Agent Safety Platform, an open reference design to extend security deeper into the agentic stack and establish stronger boundaries for autonomous AI.Nvidia OpenShellAt the software layer, technologies like NVIDIA OpenShell provide an open source secure runtime that governs how agents execute, what they can access, and where inference occurs.NVIDIA SentryNVIDIA Sentry adds an independent watchdog running on NVIDIA BlueField-4 DPUs that uses NVIDIA DOCA for out-of-band, in-silicon monitoring of agent activity and security policy enforcement.NVIDIA Vera CPUsNVIDIA Vera CPUs provide high-performance compute for agentic workloads.