It started with a conversation that stopped us in our tracks. HTH CTF staff came across a blog post about a recent conference where someone used AI to solve CTF challenges—and without much prompting, it completed nearly 90% of them. That wasn’t just impressive. It was a wake-up call.
Reality check
We decided to put it to the test ourselves. We fed AI our previous years’ challenges, expecting mixed results. Instead, it tore through them with surprising efficiency. That led to a sobering moment. Like most of our team meetings, it quickly turned into a flood of ideas, debates, and more questions than answers.
“How do you build challenges in a world where AI already knows most of the answers?”
Hard questions
We kept circling the same core problems: how do we stop people from "cheating"? How do we continue teaching relevant skills? And how do we design challenges that can’t be solved with a single prompt, without making them so difficult that players give up? These weren’t easy questions—and honestly, they still aren’t fully answered.
Adapt and overcome
Like any good group of hackers, we adapt. The truth is, you can’t stop the AI movement. That door is wide open—and more importantly, why would you want to close it? We hear it all the time: "AI is going to replace our jobs". But our experience has been the opposite. We have more ideas, more work, and more capability than ever—and AI is the reason we can keep up.
Rethinking CTFs
That realization changed everything. Instead of treating AI as a cheating tool, we now see it as a requirement to play the game. This means shifting what we teach. We don’t need to spend time drilling advanced Linux commands or scripting from scratch—AI can handle that. What matters now is knowing how to use AI effectively and understanding where it fails.
“The new skill isn’t avoiding AI—it’s knowing how to think with it.”
What’s new this year
This year’s CTF reflects that shift. We’ve cut back on the traditional “do this, then that” style challenges and introduced entirely new categories. Some challenges will be impossible to solve without AI. We’ve also added a new Games category—interactive, learn-by-playing experiences that would have taken us months to build before, but now take hours thanks to AI.
Players will learn about prompt injection, how to write better prompts, and how to recognize the pitfalls of AI. In the past, we had plenty of ideas but not always the time or specialized knowledge to bring them to life. Now, we’re building challenges we could only dream about before—and having a lot of fun doing it.
Looking ahead
We’re excited for what this means, not just for this year’s event, but for the future of learning and hacking. We hope players enjoy solving these challenges as much as we enjoyed creating them.
We’ll see you at BrewDog in Canal Winchester, Ohio, June 3–5 for another unforgettable year of Hackers Teaching Hackers, Spaceballs edition! May the Schwartz be with you!