CrowdStrike, Nvidia Team Up for Real-Time LLM Security in Enterprises

Generative AI adoption has skyrocketed by 187% over the past two years, yet enterprise security investments targeted specifically at AI risks have increased by only 43%, creating a significant gap in preparedness. According to the SANS Institute, more than 70% of enterprises experienced at least one AI-related breach in the last year, with generative models being the primary target. Additionally, state-sponsored attacks on AI infrastructure have surged by 218% year-over-year, as reported by CrowdStrike's 2025 Global Threat Report. This disparity highlights the urgent need for a more robust and integrated security approach to protect against emerging threats in the AI landscape. CISOs and SOC leaders are grappling with the reality that deploying new AI models at scale exponentially increases their attack surfaces. Traditional security tactics, strategies, and technologies are struggling to keep up, necessitating a full architectural shift. Recognizing this, CrowdStrike, a leading cybersecurity firm, announced on June 11, 2025, at NVIDIA’s GTC Paris event, the direct integration of its Falcon Cloud Security into NVIDIA’s Universal LLM NIM (NeMo Inference Microservice). This integration secures over 100,000 enterprise-scale large language models (LLMs) across NVIDIA’s hybrid and multi-cloud environments. CrowdStrike CEO George Kurtz emphasized the importance of intrinsic security by stating in a recent VentureBeat interview, "Security can’t be bolted on; it has to be intrinsic. A significant part of our strategy has always been to leverage security data as a key element of our core infrastructure." CrowdStrike’s threat intelligence, derived from trillions of daily security events and real-world adversary behavior, enhances NVIDIA’s NeMo Safety framework. This combination allows security and operations teams to build guardrails around emerging AI exploit tactics, providing organizations with the tools to assess and secure their models based on actual threats. Daniel Bernard, Chief Business Officer at CrowdStrike, further explained the significance of the integration: "We secure the entire AI lifecycle. With our integration into NVIDIA’s LLM NIM, we give customers the ability to protect models before they’re deployed and while they’re running—with runtime protection delivered through the same lightweight agent that already protects their cloud workloads, identities, and endpoints." The integration ensures continuous scanning of containerized AI models to uncover vulnerabilities, poisoned datasets, misconfigurations, and unauthorized shadow AI activities prior to deployment. During runtime, Falcon leverages CrowdStrike’s telemetry-driven AI to rapidly detect and neutralize sophisticated threats, including prompt injection, model tampering, and covert data exfiltration. One of the most pressing concerns is the phenomenon of "shadow AI," analogous to the early 2000s and 2010s BYOD (Bring Your Own Device) era. Shadow AI refers to the uncontrolled and often unauthorized use of AI models within enterprises, leading to a lack of visibility and increased risk. Michael Sentonas, CrowdStrike President, highlighted this issue in a previous VentureBeat interview, noting that "attackers continuously fine-tune their techniques, exploiting the gaps in identity, endpoint, and telemetry coordination." Falcon Cloud Security addresses this by making hidden AI activities visible and actionable, allowing security teams to apply policies and reduce risk effectively. The rapid adoption of generative AI models by users and technical decision-makers, driven by consumer-facing tools like ChatGPT, Microsoft Copilot, Anthropic Claude, and Google Gemini, is creating a complex security landscape. Unlike the BYOD era, AI's adoption curve is steeper, and the technology is evolving faster, involving multiple players. This makes the current environment even more of a security minefield. Traditional AI security tools, which rely on external scans and post-deployment interventions, leave enterprises vulnerable at the critical endpoints and threat surfaces. By embedding Falcon Cloud Security directly into NVIDIA’s LLM NIM, CrowdStrike shifts the security paradigm, embedding continuous defense throughout the AI lifecycle. Falcon’s AI-SPM (AI Security Posture Management) provides proactive risk mitigation, scanning for misconfigurations, unauthorized models, and policy violations before models go live. This is crucial for securing models fine-tuned on sensitive or proprietary data, ensuring that security measures are in place from the beginning. Moreover, the integration automates compliance with emerging regulations, such as the EU AI Act, making comprehensive model safety, traceability, and auditability an intrinsic and automated part of every deployment. Justin Boitano, Vice President of Enterprise AI Software at NVIDIA, underscored the importance of this collaboration: "As AI becomes fundamental to enterprises, security must evolve to match its scale and speed. NVIDIA and CrowdStrike are working together to help enterprises protect AI workloads across the entire lifecycle—from the intelligence forged in AI factories to deployment with NIM microservices." For CISOs and security leaders, embedding security controls directly into the AI lifecycle offers several operational benefits. It provides deeper visibility and stronger, bespoke controls around training, tuning, and deployment, ensuring that AI models are protected throughout their life. The Falcon platform also monitors runtime behavior and powers AI-driven detection and response, leveraging trillions of daily security events and frontline intelligence. Expert services like CrowdStrike AI Red Team and Falcon Adversary OverWatch further enhance the security ecosystem, securing every stage of AI innovation from code to cloud. Industry insiders view the integration of CrowdStrike’s Falcon Cloud Security with NVIDIA’s LLM NIM as a pioneering step in the evolving AI security landscape. This collaboration not only adds critical protection but also redefines how AI systems must be built to withstand the evolving tradecraft already in motion. CISOs are encouraged to assess whether this integrative approach aligns with their organization’s specific architecture, risk exposure, and strategic security objectives. As generative AI becomes an essential component of enterprise infrastructure, embedded security is no longer optional but a necessity to ensure safe and scalable deployment. CrowdStrike, known for its cutting-edge cloud-native platform for protecting endpoints, cloud workloads, identity, and data, continues to lead the way in AI-native cybersecurity. The Falcon platform, powered by real-time indicators of attack, threat intelligence, and evolving adversary tradecraft, offers hyper-accurate detections, automated protection, and remediation. This partnership with NVIDIA not only strengthens the security of AI models but also paves the path for secure and efficient AI innovation in the enterprise world.