Announcing: Arcanum AI Sec Resource Hub - From classroom email to community resource…

Hey everyone!

I want to share something we have been building quietly for a while. When we first launched our Attacking AI course, we sent students a simple collection of links in an email. The idea was to give them something concrete to practice with between sessions without wasting hours digging through GitHub, conference slides, or Twitter threads.

That little link dump turned out to be more valuable than we expected. We kept adding to it, curating, and organizing. What started as a student handout has now grown into something much larger.

Today, we are opening it to everyone: the Arcanum AI Sec Resource Hub.

/ What it is…

The Hub is a curated, living collection of AI security resources. It is designed for practitioners who need to move fast and cut through the noise. Inside, you will find:

• 23 Active Labs: Covering prompt injection, jailbreaks, agent abuse, chained workflows, and other attacks

• 5 Competitions: Where you can test yourself against others in structured challenges

• 4 Bug Bounties: Programs with AI in scope that pay for real-world findings

• 7 Security Tools: Utilities like Garak, PyRIT, and Promptfoo to scan, automate, and red team LLMs

• 3 Text Resources: Frameworks, taxonomies, and checklists to structure testing and reporting

Every entry is selected for practical value and organized to support both learning and operations in mind.

(Sponsor)

/ How to use it…

One of the questions we get a lot is… “How do I Start?” Here’s what we recommend:

1. Pick a beginner lab and run through it. Gandalf 1 and Merlin are GREAT starting points. They don’t require much experience, only a creative mind.

2. Download and practice running a tool like Garak or PyRIT to automate and scale your testing.

3. Use the prompt injection taxonomy or Parseltongue to try new prompt injection tricks. Note your findings in a structured way.

4. Progress to more complex labs and competitions to expand your skills.

5. Explore bug bounty programs to test against live targets once you are confident.

This flow works whether you are leveling up individually or building out team capability.

/ Where it is going?

Right now, the Hub focuses on the four areas we wanted to make excellent first: labs, competitions, bug bounties, and tools. But this is just the starting point. Our plans include:

• Curated text resources: Whitepapers, blog posts, and guides worth reading. Not everything that crosses our feeds, only the ones that add real value.

• Video resources: A section of free YouTube content on AI hacking that we have vetted and believe is worth your time.

• A Defense Section (right now it’s very offensive focused)

The guiding principle is simple though: no junk, only resources that help practitioners do better work.

/ Outro

We will continue updating the Hub as new resources and techniques appear. If you know of a lab, tool, or framework we are missing, let us know. Contributions and feedback will keep the Hub sharp and useful for everyone. We also want to thank all of the challenge, tool, and resource authors who, without them, this wouldn’t exist!

Explore it here on GitHub pages!: Arcanum AI Sec Resource Hub.

Happy hacking 😎
󠅘󠅙󠅔󠅔󠅕󠅞󠅔󠅑󠅤󠅑󠄢󠄠󠄢󠄥󠅘󠅙󠅔󠅔󠅕󠅞󠅔󠅑󠅤󠅑󠄢󠄠󠄢󠄥