Autoplay
Autocomplete
Previous Lesson
Complete and Continue
The Dark Side of AI: Jailbreaking, Injections, Hallucinations & more
The Dark Side of AI (Guardrails, Jailbreaks, Injections, Hallucinations + more)
Introduction to Guardrails and Jailbreaking (16:32)
Jailbreak! (The DAN Prompt) (7:25)
Exercise: Create Your Own Jailbreak Prompt
Many Shot Jailbreaking (18:09)
Prompt Injections - Part 1 (9:36)
Prompt Injections - Part 2 (17:42)
Thinking Like LLMs - Multi-Modal Injection (9:17)
Leaking - Part 1 (Prompt Leaking) (8:35)
Leaking - Part 2 (Data Leaking) (18:07)
Exposure (5:40)
Poisoning (3:18)
Toxicity (4:39)
Hallucinations (13:31)
Thinking Like LLMs - Big vs Small (18:58)
Challenge: Conduct Your Own Mechanistic Interpretability Research on Hallucinations (4:34)
Challenge Instructions
Leaderboard: Mechanistic Interpretability
The Model Card (11:05)
Model Cards Deep Dive (14:43)
Exercise: Explore the Latest Model Card and Learn Something New!
Exposure
This lecture is available exclusively for ZTM Academy members.
If you're already a member,
you'll need to login
.
Join ZTM To Unlock All Lectures