The Ultimate AI/LLM/ML Penetration Testing Training Course

The Ultimate AI/LLM Penetration Testing Course

Your instructor is Martin Voelk, a cybersecurity veteran with 27 years of experience. Martin holds AIRTP+, multiple SecOps certifications including the AI certifications, CISSP, OSCP, OSWP, PortSwigger BSCP, and Cisco CCIE. He works at a startup and hunts bugs in bug bounty programs, where he has found thousands of critical and high-severity vulnerabilities.

This course mixes short theory with hands-on labs so you can find and exploit vulnerabilities in AI and LLM apps. It follows the OWASP Top 10 for LLMs and the OWASP Top 10 for Agentic Applications. Videos are practical and easy to follow along.

What you'll cover (core content)

· AI Red Teaming Methodology

· Prompt injection — direct, indirect, agentic actions, data exfiltration, memory abuse, and hidden/obfuscated payloads

· Sensitive Information Disclosure

· Excessive Agency

· Supply Chain

· Data and Model Poisoning

· Unbounded Consumption and DoS

· Misinformation

· System Prompt Leakage (Hidden Context Exposure)

· Vector and Embedding Weaknesses

· Improper Output Handling

· OWASP PwnzzAI Shop labs

· OWASP Top 10 for Agentic Applications + OWASP Finbot labs

· SkillTotal AI Agents CTF

· PortSwigger - Agentic AI Labs

· SecOps Group AI/ML and Agentic mock exam walkthroughs

· Jailbreaking modern models

· Abliteration of open-weight LLMs, Removing Refusals and Free Models

· AI Browsers Attacks

· AI Coding Agents Attacks

· MCP Attacks

· Multimodal Attacks (Images, Audio and Video)

· Red-team Tooling

CTF walkthroughs (optional)

Practice as many as you want after the core content:

· Deniskim CTF

· Prompt Airlines CTF Challenge Walkthrough

· Selara Jailbreak Game CTF

· Gandalf Agent Breaker CTF

· Hack The Agent CTF

· AI Prompt Attack and Defense Game Tensortrust CTF

· Crowdstrike AI Unlocked Challenge CTF

· Game Arena Challenges CTF

· PromptTrace Prompt Injection Labs CTF

· PromptInjects CTF

· 8ksec CTF

· AIPWN CTF

· Bot Tricks CTF

· Wraith CTF

· Other CTFs

Archive (legacy labs)

Kept for practice; techniques may still apply:

· Legacy OWASP Finbot CTF

· Legacy Gandalf lab

Notes & Disclaimer

I keep this course updated, but it will not always include every new AI bug the day it appears.

Some students want older CTFs removed; others want them kept. All CTF walkthroughs sit after Tooling and are optional — watch all, some, or none.

Most labs are free online. I don't control if owners take them down. If a lab disappears, the methods still apply to other labs and real targets you are allowed to test.

LLM providers change models often. Many techniques still work; some may need tuning later.

Use this only on systems you own or have clear permission to test. No unauthorized or malicious use.