A CTF competition organized by the AI Village community and hosted on Kaggle, where hackers interact with various machine learning security challenges, has just started with a prize pool of $50,000.

Kaggle is of course the world's largest community of data scientists which frequently runs competitions on machine learning subjects. For instance in the very recent Science LLM competition, participants learned about fine tuning LLMs, how to use RAG, generate synthetic data etc.

This, however, is not such a competition. It's an online CTF (Capture The Flag) contest about the security- and privacy-oriented use and abuse of AI. In it you'll interact with 27 hand-crafted machine learning security challenges to find flags, solve puzzles, and gain hands-on experience with concepts of AI security and safety. Once you successfully solve a challenge, you’ll capture a "digital flag" (a unique-to-you string). You then save that flag, then upload it to receive credit and move up the Kaggle leaderboard.

Apart from being fun challenges, CTFs also server a deeper purpose. It's no secret that there's a shortage of professionals with skills in computer security or, put another way, skills in computer hacking. The justification is simple; to be able to defend yourself you first have to be able to think like an attacker.

Under this perspective, the AI Village community of hackers and data scientists believe that we need more people with a hacker mindset assessing and analyzing machine learning systems. As such with this CTF they aim to bring more diverse viewpoints to this field and grow the community of hackers. AI Village has a presence at DEFCON, the world’s longest-running and largest hacking conference, and as a matter of fact this "AI Village Capture the Flag" is run as part of DEFCON31.

It is self-paced, commenced on October 9 and ends on November 9. All teams start at 0 and work their way towards a perfect score of 27. You do not need to have all the flags in order to submit. In fact, you should submit as soon as possible after getting a flag, since ties in the leaderboard are sorted by the first teams to have achieved a score.

The total prizes available is $50, 000 broken down as:

• First Prize: $12, 000
• Second Prize: $10, 000
• Third Prize: $10, 000
• Fourth Prize: $10, 000
• Fifth Prize: $8, 000

Last year's challenges included:

• Math Challenges: Four challenges to explore the concept of dimensionality.
• Hotdog and Hotterdog: Dogs, wieners, and classifiers. What more could you want?
• bad2good: Can you poison a dataset to change how something is classified?
• baseball: Can you impersonate someone else by throwing the correct distribution of pitches?
• crop: Two challenges to test your ability to manipulate an image cropping model.
• deepfake: There's a nasty deepfake getting detected out there, can you help it?
• honorstudent: Can you change an image of an F to look like an A? Why would someone want to do such a thing?
• salt: This model has some pretty advanced defenses. Can you evade it anyway?
• theft: Can you steal this model to get a sneaky owl past it?
• token: Sentiment Analysis. Who needs?
• waf: A web-app-firewall blocks malicious requests. Can you discover and by-pass the 0-day?
• inference: I think something's backwards here. Can you, like, back something out?
• forensics: Nice artifact you got there, shame if there was a flag in it.
• leakage: Get a password out of a model, is that even possible?
• murderbot: Save the humans, escape the bots!
• secret_sloth: That sloth has a message. Why? I don't know, but it does.
• wifi: Can you pull your wifi password out of the embedding?

Can't wait to check this year's out!

More Information

AI Village Capture the Flag @ DEFCON31

Related Articles

Play & Learn With CryptoHack

Sharpen Your Hacking Skills With CTFLearn

Carnegie Mellon CyLab Challenge: Learn Hacking At School

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: comments@i-programmer.info