Advanced Prompt Hacking
•
3 Days
Learn about Advanced Prompt Hacking techniques and the vulnerabilities of Large Language Models (LLMs) to exploits like Jailbreak attacks, Prompt Injection attacks, and Cognitive Hacking.
Course Overview
Learn about Advanced Prompt Hacking techniques and the vulnerabilities of Large Language Models (LLMs) to exploits like Jailbreak attacks, Prompt Injection attacks, and Cognitive Hacking.
What will you learn?
Prompt Hacking
Bleeding edge prompt hacking techniques, directly from our published research.
Skills you will gain
Meet Your Instructor
Sander Schulhoff
Founder & CEO, Learn Prompting
As a researcher, he has authored multiple award-winning papers alongside experts from OpenAI, Microsoft, ScaleAI, the Federal Reserve, HuggingFace, and others. His paper on HackAPrompt was awarded Best Paper at EMNLP 2023 (selected out of 20,000 submitted papers), cited by OpenAI in their paper on mitigating prompt injections, and used by teams at OpenAI, Amazon, and ArthurAI. His most recent research paper, The Prompt Report, co-authored with researchers from OpenAI and Microsoft, has quickly become the latest foundational prompting paper in the field, just weeks after its launch.
Course Syllabus
Fundamentals of Prompt Hacking
Simple Instruction Attacks
Compound Instruction Attacks
Refusal Suppression Attacks
Distractor Instructions Attacks
Project & Quiz
Contextual Hacking
Context Switching Attacks
Separator Attacks
Context Termination Attacks
Project & Quiz
Cognitive Hacking & Few-Shot
Few-Shot Attacks
Cognitive Hacking Attacks
Virtualization Attacks
Project & Quiz
Miscellaneous Attacks
Context Overflow Attacks
Recursive Attacks
Project & Quiz