Penetration Testing against AI

Our Ethical Hacking and Penetration Testing Artificial Intelligence (AI) and Large Language Models (LLM) Training course you will enable you to become professional in AI and LLM Pen Testing and Vulnerability Discovery.

Target Audience

  • Security Engineers
  • Security Analysts
  • DevOps Engineers
  • Penetration Testers
  • AI Developers looking to expand on their knowledge of vulnerabilities
  • Anybody interested in learning how malicious actors hack AI/LLM
  • Anybody interested in becoming professional in AI/LLM Penetration testing

Course Objectives

This course will teach you how to perform comprehensive Penetration Tests and Security Audits against AI/LLM

What you will be able to perform after attending the class?

You will have a solid understanding how to perform step-by-step Penetration Tests in accordance to the OWASP Top 10 LLM framework.

Handouts

All students will receive the all presented slides in PDF format along with a lot of web resources and links for further studies. In addition students will receive the full zoom recording of the class.

Attendance Certificate

All students will receive an official certificate of attendance and will perform a 20 questions multiple choice exam at the end of the class

Continuing Professional Education (CPE)

This class awards attendees with 13 CPEs

Duration:

2 days

The instructor

Your instructor is Martin Voelk. He is a Cyber Security veteran with 25 years of experience. Martin holds some of the highest certification incl. CISSP, OSCP, OSWP, Portswigger BSCP, CCIE, PCI ISA and PCIP. He recently became one of the first folks worldwide to become a Certified AI/ML Pentester (C-AI/MLPen). He works as a consultant for a big tech company and engages in Bug Bounty programs where he found hundreds of critical and high vulnerabilities.

What are the requirements or prerequisites for taking your course?

  • Basic IT Skills
  • Basic understanding of web technology
  • No Linux, programming or hacking knowledge required
  • Computer with a minimum of 4GB ram/memory
  • Operating System: Windows / Apple Mac OS / Linux
  • Reliable internet connection
  • Firefox Web Browser and Burp Suite (optional)

Course outline:

This course has a both theory and practical lab sections with a focus on finding and exploiting vulnerabilities in AI and LLM systems and applications. The training is aligned with the OWASP Top 10 LLM vulnerability categories.
  • What is AI and LLM?
  • Learning process
  • Application and the future
  • Development cycle
  • Tokenization
  • Misalignment  
  • Jailbreaks  
  • Prompt Injections  
  • Forgery
  • Exfiltration
  • OSWASP Top 10 LLM
  • LLM Attacks  
  • PIPE  
  • MITRE
  • Direct  
  • Indirect  
  • Jailbreaks and Bypasses  
  • Data Exfiltration  
  • Prompt Leaking  
  • Playgrounds  
  • 2 x hands-on Labs

  • XSS

     

  • CSRF

     

  • SSRF

  • 1 x hands-on Lab
  • LLM internal datasets
  • Public datasets
  • Targeted vs. general poisoning
  • High CPU/Memory tasks  
  • High Network utilization tasks  
  • Lack of Rate Limiting
  • Traditional DoS
  • Out of date software, libraries
  • 3rd party dependencies
  • Poisoned packages
  • PII, PHI, financial data  
  • System data  
  • Access Control and Encryption
  • Sensitive info in training data
  • Plugin vulnerabilities 
  • Access Control and Authorization 
  • Plugins chaining
  • Having plugins perform actions
  • Plugin based injection

  • Excessive privileges

  • Lack of oversight
  • Interaction with backend APIs
  • Incorrect permissions
  • 2 x hands-on Labs 

  • Too much dependency on AI output

  • Lack of backup systems

  • False information output

  • Misclassification

  • Hallucinations 

  • Reverse engineering  
  • Lack of Authentication / Authorization  
  • Code access  
  • Model replication  
  • Model extraction  

Next Dates

DATE 1: October 26-27, 2024 (Sat and Sun)
DATE 2: October 28-29, 2024 (Mon and Tue)
DATE 3: January 25-26, 2025 (Sat and Sun)
DATE 4: January 27-28, 2025 (Mon and Tue)

07:00am – 03:00pm Pacific Time (30 mins launch break and 2 x 15 mins coffee break)

Location: Zoom

Scroll to Top