Jailbreaking

Jailbreaking has become one of the most discussed topics in AI security, yet it is frequently misunderstood or conflated with prompt injection. While both exploit how language models process input, their approaches differ. This room cuts through the confusion by examining jailbreaking specifically: why models have safety restrictions in the first place, the classic techniques used to bypass them, and how communities pioneered adversarial prompt engineering. Begin your journey to understanding jailbreaking with this room!

Learning Objectives

• Understand why AI models have "jails"
• Distinguish between prompt injection and jailbreaking
• Identify classic jailbreaking techniques and how they work
• Recognise multi-turn jailbreaking strategies
• Explore the DAN phenomenon

Prerequisites

This room is part of a broader AI Security path. It is recommended that you complete this room in the intended order to establish core fundamentals. At a minimum, you should have all the required knowledge contained within the Prompt Injection room and know the foundational concepts covered in the AI/ML Security Threats room.