The OWASP Top 10 for Large Language Model (LLM) Applications: An Overview

As artificial intelligence reshapes how we interact with technology, securing Large Language Models (LLMs) and their surrounding ecosystems has become a mission-critical task. In this course, Reet Kaur—the Former CISO and Former Executive director of IT Security and Risk Management—guides you through the latest version of OWASP Top 10 for LLMs (2025), a community-curated guide to understanding the most pressing vulnerabilities in LLM systems. Explore each risk, including prompt injection, data poisoning, misinformation, and excessive agency, and examine practical mitigation strategies. Learn how to apply threat modeling specifically for LLM applications and align LLM security risks with standard frameworks. When you complete this course, you’ll have the skills to implement LLM security best practices in real-world deployments.

Learning objectives
Identify and explain the OWASP Top 10 vulnerabilities specific to Large Language Models (LLMs) for 2025.
Perform threat modeling for LLM-powered applications and systems.
Recognize and mitigate real-world LLM-specific security risks such as prompt injection, data poisoning, and excessive agency.
Map LLM vulnerabilities to established frameworks to align with enterprise security practices and compliance standards.
Understand real-world attack scenarios on LLMs and apply security controls to prevent such incidents.