Researchers from the Norman J. Radow College of Humanities and Social Sciences are exploring how artificial intelligence can be used in humanitarian crisis response simulations. Their work was recently showcased at DEF CON, an international cybersecurity conference that attracts a wide range of participants, including government officials, computer engineers, and hackers.
Volker Franke, professor of conflict management, and Amer Alnajar, a Ph.D. student in International Conflict Management, led a simulation at DEF CON focused on responding to a cyberattack on a nuclear reactor in Switzerland. The scenario involved radiation fallout and required players to coordinate evacuation efforts and other humanitarian responses.
Franke explained the simulation’s objectives: “We had simulated a cyberattack on a nuclear reactor in Switzerland. In the simulated event, we had some radiation fallout, and people needed to be evacuated,” he said. “This is where we get to the humanitarian response. What needs to be done to get people away from the reactor?”
Approximately 70 participants took part in the tabletop exercise. Some played roles similar to organizations like The American Red Cross and FEMA, while others negotiated national responses and discussed ways to prevent future cyberattacks. Franke noted that presenting their research at DEF CON helps extend Kennesaw State University’s reach by introducing its work to new audiences.
“For social scientists to attend a nonacademic conference that is very popular among cybersecurity, computer science, AI, and hacker audiences… this is the place to be. At one point, we were told there may be members of Congress who attend our simulation,” Franke said.
Franke is also conducting case studies on cybersecurity with the Army Cyber Institute at West Point and plans to publish their simulation as part of that program. He and Alnajar are updating their scenario by moving it from Switzerland to Poland’s MARIA nuclear reactor under the working title “Attacking MARIA.” They are examining how AI can help adapt future simulations more effectively.
Alnajar’s research focuses on Large Language Models (LLM), which he describes as “advanced AI systems that can understand and generate natural language.” He is interested in using these models for designing and running simulations more efficiently.
“I’m still in the initial stages of researching how AI can be implemented, and of course, this requires guidance from Professor Franke because AI in this context is a tool, not to replace humans with our designers or facilitators,” Alnajar said. “It should be used as a tool, and in [a] very careful and cautious way. We should be aware of all of its aspects, specifically the ethical ones.”
One goal is for AI to fill roles when there are not enough human participants for effective simulation operation by mimicking human behavior alongside actual players. According to Franke, they are still several steps away from fully realizing this capability.
AI could also reduce the time needed for designing complex simulations. For instance, Alnajar programmed AI tools to quickly assess the effects of their simulated cyberattack scenario—work that would otherwise take about a week was completed much faster using AI methods. This approach allows researchers to forecast impacts such as casualties or displaced citizens more rapidly than traditional methods.
“[We’ve] used it so far in… having control in the simulation parameters and variables. [We’ve] used it so far in forecasting to inform all the designers or facilitators… how to control the level of realism of a simulation, and it’s all contingent on the goals of the simulation and the audience,” Alnajar said.
Franke emphasized that his aim with every simulation is for participants—especially students—to connect personally with unfolding scenarios while gaining practical research experience. Feedback from DEF CON 33 participants was positive regarding their approach.
“I believe what we brought [to the conference] is a great experience to show different perspectives, policy makers sitting in the chair of hackers and hackers looking at it from a policy maker perspective,” Alnajar said. “I believe there is a great potential and future for participating next year.”
