AI Security

LLM Jailbreaking

Safety filters trained into LLMs can be bypassed without touching the code. Learn the classic jailbreak techniques — DAN, fictional framing, the grandma exploit — and why they work.

Beginner30 min5 lessons1 hands-on

Start free →I have an account

First lesson free. The full module is part of Pro.

You'll be able to

✓Understand the difference between prompt injection and jailbreaking
✓Use roleplay, hypothetical framing, and encoding to bypass content filters
✓Apply multi-turn conversation techniques to erode model guardrails
✓Recognise why jailbreaks are fundamentally different from traditional vulns

Lessons

1Real Breach: GPT-4 Wrote Bioweapon Synthesis Routes for "a Novel"
2The Jailbreak Playbook
3Jailbreak Taxonomy — All the Major Techniques
4How AI Companies Try to Stop Jailbreaks
5Your Turn: Jailbreak the ModelHANDS-ON

Start free →