AI Cyber Challenge Code/Vulnerability Visuals by Mark Griffin

The Artificial Intelligence Cyber Challenge (AIxCC) is a two-year competition that brings together the best and brightest in AI and cybersecurity to design novel AI systems to secure critical software.

The challenge projects and vulnerabilities for AIxCC's semifinal competition varied in size, scale, and complexity. To communicate the game details to a variety of audiences—including competitors, hackers, policymakers, and students—our team developed code visualization videos to illustrate the scope of the code involved, the impact of vulnerability exposures, and the difficulty of the tasks faced by competitors.

The team performed analysis on the specific challenge code to generate technically accurate three-dimensional call graphs, a standard visual representation of code. Each video also contains a description of the challenge project or vulnerability. The overall flow of the videos shows a gradual narrowing that represents the scope and the difficult process for competitors, essentially a "needle in a haystack" problem.

Green nodes and connections show the full extent of the code in a particular project; the yellow portion focuses on code reachable from an interface exposed to competitors; and red narrows down to the path where a particular vulnerability is triggered. For videos focused on specific vulnerabilities, the camera follows the path to the point of a crash, showing detailed diagnostic information typically used to patch or exploit such vulnerabilities.

The videos were created by custom analysis and 3D rendering/choreography components created by our team with a total of 64 videos generated. This included one video for each of the 59 vulnerabilities created for the semifinal competition as well as one video for each of the five challenge projects, which were all large Open Source software projects.