CrowdStrike (NASDAQ: CRWD) announced a Secure-by-Design AI Blueprint developed with NVIDIA that integrates security protection directly into autonomous AI agent systems. The architecture combines CrowdStrike's Falcon platform with NVIDIA's OpenShell runtime to provide security controls for AI agents running locally on DGX Spark systems and in cloud environments.

The blueprint addresses security challenges as organizations transition from AI copilots to autonomous agents that operate independently. According to the announcement, AI agents present different security risks as they have privileged access to data, applications, and compute resources while operating at machine speed.

The integration embeds CrowdStrike's Falcon platform into NVIDIA's OpenShell runtime, which provides isolated sandboxes and policy enforcement for autonomous systems. The OpenShell runtime is part of NVIDIA's Agent Toolkit and supports both local deployment on NVIDIA DGX hardware and cloud-based implementations using NVIDIA's AI-Q Blueprint.

Key components include AI policy enforcement across agent operations, endpoint protection for local AI agents, cloud runtime protection for agent deployments, and identity-based governance for agent access controls. The system provides continuous monitoring and real-time enforcement to prevent unsafe behavior and unauthorized access.

"As we enter the agentic era, agents no longer simply assist – they act," said Daniel Bernard, Chief Business Officer at CrowdStrike. "This shift fundamentally changes the security equation, and security must be embedded into the AI stack itself."

Justin Boitano, Vice President of Enterprise Platforms at NVIDIA, stated that the integration enables enterprises to build and scale autonomous AI agents while maintaining security controls.

The companies plan to develop intent-aware controls that govern how agents plan and execute tasks while limiting potential damage from unintended actions.