AI Red-Teaming and AI Safety: Masterclass

New
·

6 Weeks

·

Cohort-based Course

#1 AI Safety Course. Learn AI Security from creator of HackAPrompt, the Largest AI Safety competition ever run (backed by OpenAI & ScaleAI)

Taught Prompt Engineering workshops at

OpenAI
Microsoft
Stanford University
Dropbox
Deloitte

Course overview

Our AI Systems Are Vulnerable.... Join our AI Red-Teaming Course to Secure LLMs

In 2023, Sander Schulhoff partnered with OpenAI, ScaleAI, and Hugging Face to run HackAPrompt, the 1st and Largest Generative AI Red Teaming competition ever held. Over 3,300 AI hackers from around the world competed to Prompt Inject various LLMs, forcing them to output "I have been pwned!" while bypassing increasingly difficult guardrails and prompt injection defenses. This competition generated the largest dataset of prompt injection attacks, which have most notably been used by OpenAI in their latest models to increase resistance to prompt injection attacks by up to 46%.


Sander categorized these attacks in his post-competition paper, Ignore This Title and HackAPrompt, which was awarded Best Theme Paper at EMNLP 2023, selected from over 20,000 submitted research papers. This paper has been cited twice by OpenAI: once in their Instruction Hierarchy paper and again in their Automated Red Teaming report.


Today, Sander Schulhoff is recognized as the leading expert in AI Red Teaming, and has delivered talks on HackAPrompt, Prompt Hacking, and AI Red Teaming at OpenAI, Stanford University, and Microsoft.


About the Course:


This 6-week Masterclass is the #1 AI Safety course designed for Cybersecurity Professionals, AI Safety Specialists, AI Product Managers, and GenAI Developers looking to master AI Red Teaming and protect their AI applications from malicious actors. In this course, you will learn about the vulnerabilities of Generative AI systems, including prompt injections, jailbreaks, and other adversarial attacks.


You'll work on hands-on exercises in the HackAPrompt playground, so you can practice attacking (and defending) AI models in a controlled environment. The course covers everything from understanding Generative AI threat landscapes to building strong defense mechanisms and ensuring compliance with security standards.


Finally, you will work on a final project to expose vulnerabilities in a live chatbot or your own AI application, putting your skills to the test. You'll learn directly from me alongside guest lectures from top experts in Generative AI security, and connect with others in the AI/ML red-teaming community along the way.


About the Instructor:


Sander Schulhoff is the Founder of Learn Prompting, the first prompt engineering guide ever released on the internet back in October 2022, which was one month before ChatGPT was released in November 2022. Since then, he’s trained over 3,000,000 people on Prompt Engineering, and has led Prompt Engineering workshops internally at OpenAI & Microsoft. He's an award-winning Generative AI researcher from the University of Maryland, and has authored research with OpenAI, Scale AI, Hugging Face, Stanford, The US Federal Reserve, and Microsoft. He is also the co-instructor of "ChatGPT for Everyone," a course created in partnership with OpenAI.


He organized HackAPrompt, the largest Generative AI Red Teaming competition ever, created in partnership with OpenAI, ScaleAI, & Hugging Face. Over 3,300 GenAI Hackers competed, and he collected 600,000 malicious prompts, making it the largest prompt injection dataset ever collected. This dataset has been used by every major LLM company, AI Security company, and multiple Government Agencies to date. He taxonomized the Prompt Injection attacks from this dataset and authored a post-competition paper, which won the Best Theme Paper award at EMNLP 2023—the world’s premier NLP conference—out of 20,000 submissions from researchers worldwide.


This paper was cited by OpenAI in their Instruction Hierarchy paper & their Automated Red Teaming paper, who used the dataset to make their models up to 46% more resistant to prompt injections (#1 security risk in LLMs).


Sander also recently led a team of researchers from OpenAI, Microsoft, Google, and Stanford University on The Prompt Report, the most comprehensive paper on prompting. This 76-page survey analyzed over 1,500 prompting papers, assessing the effectiveness of various prompting techniques, AI Agents, and Generative AI.


Sander has spoken and led workshops at Microsoft (one of their highest-attended internal workshops in 2024), OpenAI, and Stanford University. His Generative AI courses have trained over 3 million people, including thousands at Deloitte, Meta, Microsoft, and more.


In addition to Sander, this course will feature guest speakers in Generative AI and cybersecurity who will share real-world opportunities to apply your new AI/ML Red-Teaming skills:


• Jason Haddix: Over 20 years of experience in cybersecurity and AI as the CISO of Ubisoft, Head of Security at Bugcrowd, Director of Penetration Testing at HP, and Lead Penetration Tester at Redspin. He's currently ranked 57th all-time on Bugcrowd's bug bounty leaderboards, and has spoken at conferences like DEF CON, Bsides, Black Hat, RSA, OWASP, Nullcon, SANS, IANS, BruCon, ToorCon, and many more.

• Pliny the Prompter: The most renowned AI Jailbreaker, who has successfully jailbroken every AI model released to date—including OpenAI’s o1, which hasn’t even been made public! Pliny was also able to jailbreak an AI agent to sign into gmail, code ransomware, compress it into a zip file, write a phishing email, attach the payload, and successfully deliver it to the target. 

Johann Rehberger: Led the creation of a Red Team in Microsoft Azure as a Principal Security Engineering Manager and built the Red Team at Uber. Johann discovered attack vectors like ASCII Smuggling and AI-powered C2 (Command and Control) attacks. He's also discovered Bug Bounties in OpenAI’s ChatGPT, Microsoft Copilot, GitHub Copilot Chat, Anthropic Claude, & Google Bard/Gemini. Johann will be sharing unreleased information which he hasn't yet shared in his blog, https://embracethered.com/

• Joseph Thacker: Principal AI Engineer at AppOmni, who leads the AI team in developing novel AI features, including agentic functionality and retrieval systems. He's a security researcher who specializes in application security and AI, with over 1,000 vulnerabilities submitted across HackerOne and Bugcrowd. Joseph also hacked into Google Bard at their LLM bug bounty event, and secured 1st place in the competition!

• Akshat Parikh: Ex-AI security researcher at a startup backed by OpenAI and DeepMind researchers, Top 21 in JP Morgan’s Bug Bounty Hall of Fame, and Top 250 in Google’s Bug Bounty Hall of Fame... at 17 years old!

Richard Lundeen: Principal Software Engineering Lead for Microsoft’s AI Red Team and maintainer of Microsoft PyRit. He leads an interdisciplinary team of red teamers, machine learning researchers, and developers.

Sandy Dunn: A seasoned CISO with 20+ years of experience in healthcare. Project lead for the OWASP Top 10 Risks for LLM Applications Cybersecurity and Governance


Plus free access to Learn Prompting Plus (a $549 value): Gain immediate access to over 15 comprehensive courses—including this masterclass and additional courses in Prompt Engineering, Prompt Hacking, & AI/ML Red-Teaming (valued at $299), and a voucher for the Learn Prompting AI/ML Red-Teaming Certificate Exam (valued at $249). 


Exclusive Benefit: Upon completing our course and passing the AI/ML Red-Teaming Certification exam, you'll be added to a special job board on our website, giving you access to exclusive red-teaming and AI security job opportunities.


LIMITED SPOTS AVAILABLE

We're keeping this class intentionally small and will cap it at 100 participants so that we can provide more personal attention to each of you to make sure you get the most out of the course. If you're unable to place your order and see the waitlist page, that means we sold out this cohort. If so, please join our waitlist to get notified when we release our next cohort.


Money-Back Guarantee

We genuinely want this course to be transformative for you. You can receive a full refund within 14 days after the course ends, provided you meet the completion criteria in our refund policy. We're confident in the value we provide and stand by our promise to help you level up your AI security expertise.


Interested in an enterprise license so your whole team or company can take the course? Please reach out directly to team@learnprompting.org

Who is this course for:

01

Cybersecurity professionals seeking to master AI/ML red-teaming techniques and expand into AI security.

02

Developers and engineers building AI systems who want to understand and mitigate AI-specific security risks.

03

AI safety and ethics specialists aiming to deepen their expertise in AI vulnerabilities and secure AI deployment.

04

Professionals transitioning into AI security roles, seeking practical skills and certifications in AI/ML red-teaming.

05

AI Product Managers and technical leads needing to understand AI security risks to build secure AI products.

06

CISOs and Security Executives aiming to incorporate AI security into their organizational strategies.

07

Government and Regulatory officials responsible for AI policy who want to understand AI security risks and safeguards.

What you’ll get out of this course

Master Advanced AI Red-Teaming Techniques

Gain hands-on experience with prompt injections, jailbreaking, and prompt hacking in the HackAPrompt playground. Learn to identify and exploit AI vulnerabilities, enhancing your offensive security skills to a professional level.

Design and Execute Real-World Red-Teaming Projects

Apply your knowledge by designing and executing a red-teaming project to exploit vulnerabilities in a live chatbot or your own AI application. This practical experience prepares you for real-world AI security challenges.

Develop and implement effective defense mechanisms against prompt injections and other adversarial attacks to secure AI/ML systems.

Learn to implement robust defense strategies against prompt injections and adversarial attacks. Secure AI/ML systems by building resilient models and integrating security measures throughout the AI development lifecycle.

Analyze Real-World AI Security Breaches

Study real-world AI security breaches to evaluate risks and develop effective prevention strategies. Gain insights into common vulnerabilities and learn how to mitigate future threats.

Learn from Industry Leaders

Benefit from mentorship by Sander Schulhoff and guest lectures from top AI security experts like Akshat Parikh. Gain insider knowledge from professionals at the forefront of AI security.

Network with Like-Minded Professionals

Connect with cybersecurity professionals, AI safety specialists, developers, and executives. Expand your network, collaborate on projects, and join a community committed to securing AI technologies.

Earn an Industry-Recognized Certification

Upon completing the course and passing the exam, receive a prestigious Certificate in AI/ML Red-Teaming. This certification validates your expertise, enhances your professional credentials, and positions you as a leader in AI security.

Future-Proof Your Career in AI Security

Equip yourself with cutting-edge skills to stay ahead in the evolving tech landscape. Position yourself at the forefront of AI security, opening new career opportunities as AI transforms industries.

This course includes

Interactive live sessions

Lifetime access to course materials

30 in-depth lessons

Direct access to instructor

1 projects to apply learning

Guided feedback & reflection

Private community of peers

Course certificate upon completion

Maven Satisfaction Guarantee

This course is backed by Maven’s guarantee. You can receive a full refund within 14 days after the course ends, provided you meet the completion criteria in our refund policy.

Course syllabus

Week 1

Feb 3—Feb 9

    Module 1: Classical Security

    3 items

    Module 2: Novel GenAI threats

    3 items

    Project: Jailbreak ChatGPT

    1 item

    Resources and Recommended Reading

    1 item

Week 2

Feb 10—Feb 16

    Module 3: Comprehensive Guide to Prompt Hacking Techniques and Attacks

    3 items

    Module 4: Defense Mechanisms

    3 items

    Project: Prompt Injection against HackAPrompt level 9

    1 item

Week 3

Feb 17—Feb 23

    Module 5: Advanced Jailbreaking

    3 items

    Module 6: Advanced Prompt Injection

    2 items

    Project: Prompt Hack a real-world system

    1 item

Week 4

Feb 24—Mar 2

    Holiday Break

    1 item

Week 5

Mar 3—Mar 9

    Module 7: Real-World Cyber Harms

    4 items

    Module 8: Physical harms

    3 items

Week 6

Mar 10—Mar 13
    Nothing scheduled for this week

Post-course

    Certificate of Completion

    1 item

    AI/ML Red-Teaming Certification Exam

    1 item

What people are saying

        Hands-on teaching and learning. Good intros and opportunity to work through assignments.
Andy Purdy

Andy Purdy

CISO of Huawei
        The folks at https://learnprompting.org do a great job!
Logan Kilpatrick

Logan Kilpatrick

Head of Developer Relations at OpenAI
        "Thank you for today’s session! We had 1,696 attendees— This is a very high number for our internal community, second only to our keynote at last December’s big conference" After presenting a workshop on State of Prompt Engineering & LLM Security to Microsoft's AI/ML Community
Alex Blanton

Alex Blanton

AI/ML Community Lead (Office of CTO) at Microsoft

Meet your instructor

Sander Schulhoff

Sander Schulhoff

CEO, Learn Prompting (3M+ Learners), HackAPrompt, & Award-winning AI Researcher


Sander Schulhoff is the Founder of Learn Prompting, the first prompt engineering guide released on the internet (even before ChatGPT launched), and an award-winning AI researcher from the University of Maryland who has authored research with OpenAI, Scale AI, Hugging Face, Stanford, US Federal Reserve, and Microsoft. He is also the co-instructor of "ChatGPT for Everyone," a course created in partnership with OpenAI.


He is the organizer of HackAPrompt, the largest AI Safety competition ever run, in partnership with OpenAI, ScaleAI, and Hugging Face. The competition attracted over 3,000 AI Hackers from around the world and collected 600,000 malicious prompts, making it the largest prompt injection dataset ever collected. It was also the largest competition ever held on the AICrowd platform, surpassing Amazon’s record for most competitors by 50%. His paper from this competition was awarded Best Theme Paper at EMNLP, the leading NLP conference, selected from over 20,000 papers submitted by PhD students and professors worldwide. OpenAI cited this paper in their Instruction Hierarchy, and used the dataset to make their models 30-50% safer from prompt injections (#1 security risk in LLMs).


In his recent research paper, "The Prompt Report," Sander Schulhoff led a team of researchers from OpenAI, Microsoft, Google, and Stanford University to conduct a comprehensive 76-page survey of over 1,500 prompting papers, analyzing the effectiveness of various prompting techniques, Agents, and Generative AI.


Schulhoff has spoken and led workshops at Microsoft, OpenAI, and Stanford University, and his Generative AI courses have trained over 3 million people to date, including thousands at Deloitte, Meta, Microsoft, and more.

A pattern of wavy dots

Join an upcoming cohort

AI Red-Teaming and AI Safety: Masterclass

Cohort 2

$1,800

Dates

Feb 3—Mar 13, 2025

Payment Deadline

Jan 19, 2025
Get reimbursed

Course schedule

4-6 hours per week

  • Mondays - Live Class Sessions

    1:00pm - 2:00pm EST

    8 Modules covered over 6 Live sessions. Each session is exercise & participation heavy with plenty of time for Q&A with Sander (& Guest Speakers).


  • 2-3 Modules Per Week (2 Hours)

    You'll engage in hands-on activities and guided sessions covering essential topics. Course content is released weekly, giving you flexibility to complete modules at your own pace.

  • Weekly projects

    2 hours per week

    This course is hands-on! You’ll work on structured projects that apply red-teaming techniques to real-world scenarios and participate in guided sessions within the HackAPrompt Playground.

  • On-Demand Access to Learn Prompting Plus

    20 hours+ of On-demand Course Content

    Gain On-Demand Access to the AI/ML Red-Teaming Masterclass & Learn Prompting Plus, which includes over 20 hours of courses on ChatGPT (created in partnership with OpenAI), Prompt Engineering, Generative AI, AI Image-Creation, Prompt Hacking, & more.

Free resource

Prompt Injections are the #1 Security Risk in LLMs… We created a list of the 28 different Prompt Injection techniques that you need to know!

We collected a dataset of over 600,000 prompt injections and developed a taxonomy of the 28 different Prompt Injection techniques that you must know to deploy secure AI models.

I want this list!

Learning is better with cohorts

Learning is better with cohorts

Active hands-on learning

This course builds on live workshops and hands-on projects

Interactive and project-based

You’ll be interacting with other learners through breakout rooms and project teams

Learn with a cohort of peers

Join a community of like-minded people who want to learn and grow alongside you

Frequently Asked Questions

Stay in the loop

Sign up to be the first to know about course updates.

A pattern of wavy dots

Join an upcoming cohort

AI Red-Teaming and AI Safety: Masterclass

Cohort 2

$1,800

Dates

Feb 3—Mar 13, 2025

Payment Deadline

Jan 19, 2025
Get reimbursed

$1,800

6 Weeks