Security Risk in AI/ML is a one-day course covering the security risks unique to machine learning and generative AI systems. The threat surface differs from traditional application security and security teams need to learn new ways to assess and mitigate risk as AI adoption outpaces AI-specific security practices at many organizations. Students will examine how adversaries manipulate, poison, steal, and extract information from ML models and LLM-integrated applications. They will leave able to identify and address AI/ML security risks aligned with 2026 AI governance and security priorities.
The course covers adversarial machine learning risks in detail. Students will learn more about how adversarial examples evade model predictions, data poisoning attacks corrupt training data, and model extraction techniques steal models. Attendees then dig into LLM-specific risks like prompt injection and privacy risks from model inversion and membership inference attacks. An examination of supply chain risk rounds out the course to explore the risks posed by malicious model files and vulnerable dependencies in the ML pipeline.
Through lecture, discussion, and demonstration, the course examines real AI/ML attack techniques covered in the OWASP Top 10 for LLM Applications. By completion, attendees will have the foundational knowledge to apply a structured risk assessment approach and mitigate security risks across machine learning and generative AI systems.
Who Should Attend
Security Engineers, ML Engineers, Data Scientists, AI Engineers, Risk/Compliance personnel
What Attendees Will Learn
Upon completing Security Risk in AI/ML, attendees will be able to:
- Identify AI/ML-specific threats including adversarial examples, data poisoning, and model extraction
- Assess prompt injection risks in LLM-integrated applications
- Evaluate privacy risks using model inversion and membership inference concepts
- Secure the ML supply chain against malicious models and dependency risks
- Apply risk mitigation frameworks such as the OWASP Top 10 for LLM Applications
Prerequisites
No formal prerequisites are required for this course; basic familiarity with AI/ML concepts is helpful but not required, as relevant concepts are introduced throughout the course.