Securing AI with CrowdStrike Lab

Welcome to the Securing AI with CrowdStrike Lab, a hands-on experience designed to explore CrowdStrike's AI security capabilities across real-world environments. As employees adopt generative AI tools and development teams build internal AI applications, new risks emerge. Prompt injection, data leakage, and misuse of AI systems create attack surfaces that traditional security controls were not designed to address.

In this lab, you will see how CrowdStrike AI Detection & Response (AIDR) and Falcon Cloud Security provide unified visibility, enforce guardrails, and detect AI-specific threats across both SaaS-accessed LLMs and self-hosted models deployed on NVIDIA AI infrastructure.

The lab focuses on three key use cases demonstrating how CrowdStrike's Falcon AIDR can protect prompts and agents:

• AI Guardrails for Workforce: Implementing access rules to specific AI applications, models, and prompts
• AI Guardrails for Development: Protecting customer-facing AI applications from prompt injection attacks
• AI Container Image Assessment: Identifying security risks in containerized AI workloads running on NVIDIA GPU infrastructure using CrowdStrike Falcon.

The goal of this lab is to introduce users to practical AI security concepts and demonstrate how CrowdStrike AIDR protects AI systems across workforce, development, and runtime environments. By the end of the lab, you should have a clear understanding of how CrowdStrike AIDR delivers visibility into AI activity, enforces governance policies, detects AI-specific threats such as prompt injection and data leakage, and integrates these findings into the broader Falcon platform to strengthen enterprise security operations.

The lab walks the user through accomplishing the following:

1. Understand key AI security risks, including prompt injection, data exposure, and agent misuse, aligned to frameworks such as MITRE ATLAS and the OWASP Top 10 for LLM Applications.

2. Gain unified visibility into AI*usage across users, applications, SaaS-based LLMs, and self-hosted models.

3. Implement AI Guardrails for Workforce, ensuring appropriate access controls, usage policies, and monitoring are enforced for employee interactions with AI applications.

4. Implement AI Guardrails for Development, protecting internally developed and customer-facing AI applications from prompt injection, data leakage, and misuse.

5. Identify security risks within AI factory deployments by analyzing image assessments, vulnerabilities, misconfigurations, and runtime findings for containerized AI workloads running on NVIDIA GPU infrastructure.

Hardware:

• Windows 11 Jumpbox
• Dell PowerEdge R760xa Server
• 2 x NVIDIA L40s GPUs

Software:

• CrowdStrike Falcon AI Detection and Response (AIDR)
• CrowdStrike Falcon Cloud Security (FCS)
• NVIDIA Nemotron 3 nano model