This module provides a deep dive into adversarial prompt injection techniques and their implications on large language models (LLMs). Learners will explore various attack methods, including prompt override, impersonation, and gradient-based attacks, along with defensive strategies to secure LLM-based applications.

This module explores the concept of poisoning attacks in the context of Retrieval-Augmented Generation (RAG) and fine-tuning large language models (LLMs). It covers how malicious data can compromise model behavior and provides practical strategies for detecting and mitigating such threats. Learners will gain a clear understanding of attack scenarios and defensive measures in AI training.

This module explores advanced threats and vulnerabilities in generative AI systems, with a focus on supply-chain risks, privacy attacks, and model security. Learners will gain insights into how malicious actors can exploit open-access LLMs and learn strategies to mitigate these risks. The content also covers practical techniques for identifying and responding to adversarial attacks in AI deployments.

This module covers the principles of secure by design AI, including threat modeling, risk assessment, and the use of industry-standard taxonomies like MITRE ATLAS and AI Exchange. Learners will gain skills in identifying and mitigating AI threats, prioritizing risks, and ensuring trustworthy AI systems throughout the development lifecycle.

This module delves into the integration of MLSecOps with MLOps and LLMOps, focusing on secure AI development through workflows, patterns, and platform implementation. Learners will gain insights into building and applying security orchestration frameworks, evaluating data and models, and integrating pipelines with development workflows.

This module focuses on aligning AI security practices with enterprise strategies using the NIST CSF framework. It covers foundational elements such as secure-by-design principles, AI risk management integration, and the operational aspects of detecting, responding to, and recovering from security threats in AI systems.