James Lupo's thesis
Line of Sight: Designing AI to Detect and Disrupt Active Shooter Events
– Executive Summary –
Despite decades of active shooter incidents in the United States—from Columbine High School in 1999 to Virginia Tech in 2007, through the Century 16 movie theater in Aurora, Colorado, in 2012, and Robb Elementary School in Uvalde, Texas, in 2022—there remains no coordinated, technology-driven strategy for early detection and mitigation.[1] While security systems have advanced significantly in recent years, with artificial intelligence (AI) and object recognition technologies becoming increasingly sophisticated and ubiquitous in everyday applications, these technologies have not been purposefully applied to the active shooter problem. This thesis identifies a critical gap in homeland security: the absence of an automated early warning system capable of identifying an active shooter at the onset of an attack, potentially preventing or limiting casualties. It contends that existing technology, properly implemented, could detect weapons, recognize threatening behaviors, and trigger autonomous security responses before significant harm occurs.
Three inquiries guided the research approach for this thesis. First, the research examined whether common factors exist across active shooter scenarios that could be consistently identified by an AI system, creating a pattern recognition framework for detection. Second, the research explored potential system designs that could effectively control an active shooter situation at its onset and mitigate outcomes favorably, focusing on early intervention through automated security responses. Finally, it investigated the potential engineering challenges and social implications associated with implementing such a system, including technical reliability concerns, privacy considerations, and public acceptance of AI-driven security measures in public spaces.
The research employed a case study method examining historical active shooter events to identify common patterns and create a “template” for active shooter scenarios. This template informed the design parameters of an AI surveillance system focused on object recognition. The proposed system was evaluated using scenario-based thought experiments and computer simulations, testing the concept against various active shooter scenarios to assess its effectiveness and identify potential improvements.
This thesis presents a novel approach to the persistent problem of active shooter incidents by leveraging existing technologies in a new application. By focusing on object recognition rather than facial recognition, the system can navigate privacy concerns while providing critical early detection capabilities. The research aims to provide homeland security practitioners with a technological framework that could significantly reduce casualties in active shooter scenarios through early detection and automated response. This comprehensive approach—combining historical case analysis, technological assessment, and scenario-based testing—offers a practical solution for a persistent homeland security challenge.
[1] Ralph W. Larkin, “Armageddon (Well, Almost),” in Comprehending Columbine, ed. Ralph Larkin (Philadelphia: Temple University Press, 2007), 1–16; Timothy W. Luke, “April 16, 2007 at Virginia Tech—To: Multiple Recipients: ‘A Gunman Is Loose on Campus…,’” in There Is a Gunman on Campus: Tragedy and Terror at Virginia Tech, ed. Ben Agger and Timothy Luke (Lanham, MD: Rowman and Littlefield, 2008), 1–25; Julia Jacobo, “A Look Back at the Aurora, Colorado, Movie Theater Shooting 5 Years Later,” ABC News,July 20, 2017, https://abcnews.go.com/US/back-aurora-colorado-movie-theater-shooting-years/story?id=48730066; Julia Jacobo and Nadine El-Bawab, “Timeline: How the Shooting at a Texas Elementary School Unfolded,” ABC News, December 12, 2022, https://abcnews.go.com/US/timeline-shooting-texas-elementary-school-unfolded/story?id=84966910.
