Team Atlanta has claimed the top spot in the Defense Advanced Research Projects Agency’s AI Cyber Challenge, or AIxCC, securing $4 million in prize money for its artificial intelligence-driven cyber reasoning system.
The team, including experts from Georgia Tech, Samsung Research and the Korea Advanced Institute of Science & Technology, outperformed six other finalists in the two-year competition aimed at creating AI systems capable of autonomously detecting and patching vulnerabilities of open-source software, particularly those used in critical infrastructure, including financial systems, public utilities and the health care ecosystem, DARPA said. The agency partnered with AI industry leaders Anthropic, Google, Microsoft and OpenAI to provide competitors with technical support and large language model credits.
“Finding vulnerabilities and patching codebases using current methods is slow, expensive, and depends on a limited workforce – especially as adversaries use AI to amplify their exploits,” DARPA Director Stephen Winchell explained. “AIxCC-developed technology will give defenders a much-needed edge in identifying and patching vulnerabilities at speed and scale.”
DARPA Accelerating Adoption of Cyber Reasoning Systems
Second-place winner Trail of Bits, a New York City-based small business, took home $3 million, while third-place winner Theori, comprising AI researchers and security professionals in the United States and South Korea, received $1.5 million. Other finalists are all_you_need_is_a_fuzzing_brain, Shellphish, 42-beyond-bug and Lacrosse.
The agency will make all seven cyber reasoning systems available as open-source software under a license approved by the Open Source Initiative. On Friday, DARPA released four of the seven AI-powered systems, with the remaining set to be available in the coming weeks.
DARPA and its partner, Advanced Research Projects Agency for Health, added $1.4 million in prizes for the competing teams to integrate AIxCC technology into real-world critical infrastructure software and accelerate the adoption of the systems.
