SK Shieldus' white-hat hacker group EQST achieved a global fifth place in the artificial intelligence (AI) hacking competition at the world’s largest hacking and security conference, DEF CON 34. The team was the only domestic participant to make the top five, demonstrating its security capabilities.
On August 10, SK Shieldus announced that EQST secured fifth place among over 200 teams in the Capture the Flag (CTF) event held at the AI Village during DEF CON 34 in Las Vegas, USA.
DEF CON is the largest hacking and security conference in the world, where hackers and security researchers gather to share the latest hacking techniques and security threats. Held annually in Las Vegas, the event features lectures from security experts, technology demonstrations, and various village hacking competitions. This year, 38 villages focused on topics such as AI, automotive security, and cryptocurrency, showcasing the latest technologies and research in cybersecurity.
This year, discussions at DEF CON centered on the impact of large language models (LLMs) and AI agents, as well as self-acting AI technologies on the cybersecurity landscape. The AI Village highlighted new attack methods and security vulnerabilities utilizing AI technology, along with corresponding technologies and research cases.
The AI CTF that EQST participated in involved constructing AI agents based on a given AI model to perform penetration tests and solve security issues. Unlike traditional CTFs where humans directly execute attacks, this competition required AI agents to analyze problems and select attack methods to capture flags.
Teams competed by designing the structure and prompts of their agents based on the AI model, as well as strategizing their problem-solving processes and attack strategies. EQST optimized its AI model for different problem types, adjusted the agent's behavior, and analyzed scoring and difficulty levels to determine the order of attack, applying various strategies.
In particular, the process of designing AI agents to autonomously explore vulnerabilities and execute attacks required a comprehensive understanding of the AI model, penetration testing experience, and security problem analysis skills. As a result, EQST ranked fifth among over 200 teams worldwide, confirming its capabilities in utilizing AI technology for attack and security problem resolution.
Kim Tae-hyung, head of SK Shieldus EQST, stated, “Achieving a top-five ranking among domestic teams in the AI CTF with over 200 global participants showcases EQST's accumulated AI security technology and practical problem-solving capabilities. We will continue to expand our research on AI vulnerabilities and red teaming to proactively respond to new attacks and security threats using AI technology, while continuously enhancing our global AI security competitiveness.”
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.