0x4D31

0x4D31 / airt

Public

AIRT — A free, open-source AI Red Teaming course with 8 modules and hands-on Docker labs. Built with Perplexity Computer.

0x4d31.github.ioairt ai course prompt-injection redteam
106
14
100% credibility
Found Mar 12, 2026 at 93 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
HTML
AI Summary

A free open-source educational course providing hands-on labs to learn offensive security testing techniques for AI systems using local environments.

How It Works

1
🔍 Discover the free AI security course

You find an exciting online course that teaches how to test AI systems for weaknesses, like a fun security adventure with ready-to-play practice areas.

2
📥 Get the simple setup tools

You download a free tool called Docker that makes everything easy, like installing a game launcher on your computer.

3
🚀 Start your first practice lab

With one click, you launch a safe practice chatbot on your own computer to explore without any internet worries.

4
💻 Chat and test clever tricks

You talk to the practice AI, trying sneaky questions to see if you can make it reveal secrets, feeling like a detective cracking puzzles.

5
📚 Try more challenges step by step

You move through guided adventures learning tricks like fooling AI memory or teaming up virtual helpers, each one building your skills.

6
🏆 Master AI safety testing

After completing the hands-on labs, you gain confidence spotting and fixing AI weak spots, ready to protect real systems.

🎉 Become an AI security expert

You finish the course with practical know-how, proud of your new ability to keep AI safe from clever attacks.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 93 to 106 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is airt?

AIRT delivers a free, open-source course on AI red teaming, spanning 8 modules with 60+ hours of content and hands-on Docker labs that spin up vulnerable AI systems locally using Ollama. Developers query chatbots, exploit RAG pipelines, hijack multi-agent setups, poison supply chains, steal models, automate attacks, and analyze impacts—all without cloud API keys. Deployed as an HTML site via GitHub Pages and built with Perplexity Computer, it covers MITRE ATLAS and OWASP LLM Top 10 for offensive AI testing.

Why is it gaining traction?

In a sea of theoretical AI security talks, AIRT stands out with self-contained Docker labs that let you break real systems like prompt-injected chatbots or backdoored models in minutes. No setup hell—just clone, docker compose up, and attack. Early adopters praise the progression from basics to automation tools like garak and PyRIT, making complex topics like RAG exploitation or membership inference instantly playable.

Who should use this?

AI/ML engineers hardening production models, security pros pivoting to adversarial ML, or pentesters targeting LLM apps will get the most value. Ideal for teams running internal red teams on RAG or agentic systems, or devs auditing their own Docker-deployed AI stacks before launch.

Verdict

Grab it if AI security is your jam—the labs deliver practical exploits that stick better than slides. At 43 stars and 1.0% credibility score, it's raw and unpolished (light docs, no tests), but the free hands-on modules make it a no-brainer starter over paid courses. Run the labs today; iterate on your own airt-inspired tests tomorrow. (187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.