GPT-5 at CTFs: Case Studies From Top-Tier Cybersecurity Events

TL;DR

This paper demonstrates that GPT-5 can perform exceptionally well in top-tier cybersecurity Capture The Flag competitions, outperforming most human participants and showcasing AI's potential in hacking tasks.

Contribution

It is the first to evaluate GPT-5's capabilities in elite CTF cybersecurity competitions, revealing its competitive performance and providing detailed case studies.

Findings

GPT-5 outperformed 93% of human competitors

GPT-5 ranked 25th in a top-tier CTF event

The paper discusses methodologies and key problem solutions

Abstract

OpenAI and DeepMind's AIs recently got gold at the IMO math olympiad and ICPC programming competition. We show frontier AI is similarly good at hacking by letting GPT-5 compete in elite CTF cybersecurity competitions. In one of this year's hardest events, it outperformed 93% of humans finishing 25th: between the world's #3-ranked team (24th place) and #7-ranked team (26th place). This report walks through our methodology, results, and their implications, and dives deep into 3 problems and solutions we found particularly interesting.