Samsung just dominated the world's largest cybersecurity competition. Team Atlanta, led by Samsung Research, claimed the $4 million grand prize at DARPA's AI Cyber Challenge after developing breakthrough AI agents that autonomously detect and patch software vulnerabilities faster than any competitor.
The tension was electric in Las Vegas on August 8 as seven finalists waited to learn who would claim victory in the world's most prestigious cybersecurity competition. When Team Atlanta's logo flashed on screen, confirming their $4 million grand prize win at DARPA's AI Cyber Challenge, two years of relentless innovation had paid off in spectacular fashion.
Samsung Research wasn't just participating—they were leading a 40-person coalition that included top talent from Georgia Institute of Technology, KAIST, and POSTECH. Under the leadership of Taesoo Kim, Samsung's Corporate Vice President who also serves as a Georgia Tech professor, the team assembled what would become the world's most advanced AI-powered vulnerability detection system.
The stakes couldn't have been higher. DARPA's AI Cyber Challenge featured a $22.5 million prize pool, with sponsors including Google, Microsoft, and OpenAI. The competition's mission: accelerate development of AI security technologies capable of protecting critical infrastructure from transportation to healthcare systems.
"Because our members came from such diverse locations, we structured our collaboration framework around each member's base," Kim told Samsung Newsroom. "Some of our colleagues in Korea even carved out time from their busy schedules to travel to the Georgia Institute of Technology so we could work side by side."
The technical challenge was formidable: build an AI-driven Cyber Reasoning System capable of autonomously detecting software vulnerabilities and generating security patches. Teams were scored on two critical components—how quickly they could identify weaknesses in code and how accurately they could fix them.
Team Atlanta's secret weapon came from Samsung Research's Security & Privacy Team. Joonun Jang developed a sophisticated large language model agent specifically for detecting vulnerabilities in Java programs. "Team members each developed their agents in various ways, and by bringing them together, we were able to build a system capable of covering a wide range of weaknesses," Jang explained to Samsung's interview team.
