Threat Modeling for GenAI: LLM Security & Risk Scoring

About this course

This course contains the use of artificial intelligence.

Generative AI applications present entirely new attack surfaces that traditional application security frameworks fail to address.

Who this course is for:

• Application Security Engineers designing controls for GenAI deployments.
• Cybersecurity Architects evaluating risk in RAG and Agentic systems.
• AI/ML Engineers building LLM-powered applications in production.
• CISOs and IT Risk Managers needing clear residual risk evaluation metrics.
• Penetration Testers transitioning into AI red teaming and security reviews.

What you will learn:

• Conduct end-to-end threat modeling tailored specifically to Generative AI architectures
• Identify and prevent direct and indirect prompt injection attacks
• Secure Retrieval-Augmented Generation pipelines and vector storage layers
• Apply OWASP Top 10 for LLMs and STRIDE-AI to production environments
• Implement context-aware input/output guardrails and model firewalls
• Calculate risk scores for non-deterministic model failure modes using DREAD-AI
• Design secure trust boundaries between LLMs, APIs, and autonomous tools
• Quantify residual risk and create executive-ready compliance documentation

Requirements:

• Basic understanding of application security and threat modeling concepts.
• Familiarity with general LLM terminology (e.g., prompts, embeddings, RAG, agents).
• No prior machine learning or red teaming experience is required.

Final Project Description: You will perform a complete, end-to-end security assessment on "Enterprise SupportBot AI," an autonomous agent platform connected to internal knowledge bases and backend execution APIs.

Read the full description on Udemy on Udemy (opens in a new tab)

Prices, discounts and availability are set by Udemy. We may earn a commission when you purchase through links on this site.