CrowdStrike launches frontier AI models for cybersecurity in partnership with Nvidia

CrowdStrike just unveiled what it’s calling the first family of frontier AI models built specifically for cybersecurity defenders. The new platform, dubbed SafeMind, was announced at the company’s Fal.Con 2026 event on September 1 and represents the deepest integration yet between CrowdStrike’s security expertise and Nvidia’s AI infrastructure. At its core, SafeMind combines Nvidia’s open Nemotron models with CrowdStrike’s proprietary threat intelligence and telemetry data. The result is an agentic AI system, meaning it doesn’t just answer questions but actively performs security tasks with minimal human intervention. The early performance numbers are hard to ignore. Nemotron-powered workflows completed investigations up to 5x faster than previous methods during testing. Triage accuracy improved by more than 3x when models were fine-tuned on CrowdStrike’s data. And the system achieved 96% accuracy in converting natural language queries into CrowdStrike Query Language, the syntax security analysts use to hunt for threats. The platform deploys more than 50 specialized agents designed to automate various security tasks, from vulnerability assessments to remediation workflows. These ag...