I Forced an AI to Give Me Its Password | Prompt Injection 101

Author: zSecurity - Bewertung: 35x - Views:210

Learn how to use Prompt Injection to Jailbreak and hack AI through a series of gamified practical labs.

🔗 Material Security: https://material.security/product/oauth-agent



🧠 Learn how to use AI for Hacking and Hack AI in my Hacking Masterclass 👇

https://zsecurity.org/courses/masterclass-membership/



🧠 My other hacking courses 👇

https://zsecurity.org/courses/



🔗 Gandalf Labs: https://gandalf.lakera.ai

🔗 Pliny P4RS3LT0NG: https://elder-plinius.github.io/P4RS3LT0NGV3/



---------------------------------------------------------------

zSecurity Company - https://zsecurity.com/

Community - https://zsecurity.org/

Facebook - https://www.facebook.com/ZSecurity-1453250781458287/

Twitter - https://twitter.com/_zSecurity_

Instagram - https://www.instagram.com/zsecurity_org/

Linkedin - https://www.linkedin.com/company/zsecurity-org/

TikTok - https://www.tiktok.com/@zsecurity_org

---------------------------------------------------------------



📌 Timestamps:



0:00 – What is prompt injection?

1:41 – OWASP Top 10 risks for LLMs

2:28 – Why hack AI? Jailbreaking & data theft explained

7:08 – Getting started with the Gandalf prompt injection labs

8:20 – Level 1: Basic prompt injection

9:29 – Level 2: Bypassing system prompt restrictions

11:45 – Level 3: Defeating thinking models

15:20 – Level 4: Prompt Chaining

17:42 – Level 5: Harder reflection models

19:17 – Level 6: Bypassing GPT-powered filters

22:23 – Levels 7 & 8: Expert level!