ISI’s Yinzhi Cao participating in DARPA competition

April 8, 2024

Yinzhi Cao, assistant professor of computer science and technical director of the Johns Hopkins Information Security Institute, is part of a team selected by the Defense Advanced Research Projects Agency to develop cyber reasoning systems using artificial intelligence technology for the AI Cyber Challenge (AIxCC). The teams will test their solutions at DEFCON, set for August in Las Vegas.

Yinzhi Cao.

Cao’s team, known as 42-b3yond-6ug, is one of seven teams chosen by DARPA, a research arm of the US Department of Defense, to participate in this challenge aimed at revolutionizing software security. Each team will receive $1 million in initial funding from DARPA.

The event is designed to promote innovation among the security and AI communities using a competition model. Each group is led by small businesses. Net Shield LLC, an Illinois-based IT security company, is leading the 42-b3yond-6ug team.

“We believe that the participation will advance the state-of-the-art cyber reasoning system and bridge the gap between the AI and security communities to better detect and exploit real-world vulnerabilities,” says Cao.

Cao’s team comprises experts from Northwestern University, the University of Utah, the University of Waterloo, the University of New Hampshire, the University of Colorado, and Net Shield LLC. Several of Cao’s students including computer science graduate students Zengyu Liu, Mingqing Kang, Zifeng Kang, and Yichao Xu; Leon Gonzalez, an engineering graduate student; Phil Matthew, a PhD student in computer science; and Neil Fendley, associate professional staff at the Johns Hopkins Applied Physics Laboratory will also be involved in this work. Gonzalez, Fendley, and Matthew also currently work with the JHU Applied Physics Laboratory.

The seven teams—including Cao’s—competing in the Small Business Track at DEF CON 32 will receive $2 million and advance to the final competition in 2025. Of those seven, three will be recognized as winners of the competition at DEF CON 2025.

There are two tracks within the AI Cyber Challenge—the Small Business Track and the Open Track. The Small Business Track teams receive funding to support their work at the beginning of the competition, whereas only the winning Open Track teams are eligible for funding at the end of the competition.


JHU Information Security Institute