This project analyzes historical “warning shots” in high-risk domains to identify why some near-miss events trigger meaningful institutional reforms while others are ignored, then applies these insights to contemporary AI governance to assess when major powers might alter or slow AI development in response to an AI warning shot.
About the project
This project examines how institutions respond to “warning shots”—significant near-miss events that signal emerging dangers—and investigates why some warnings lead to meaningful reforms while others are ignored or misinterpreted. Drawing on Guest (2023) and the literature on organizational learning and high-risk system failures, the project analyzes two historically important warning shots (such as the 1983 Petrov false alarm, the Challenger O-ring precursor warnings, or biosafety lab containment failures) to identify the factors that shape institutional reactions, including signal clarity, political incentives, and bureaucratic constraints. Using insights extracted from these cases, the project develops a general “Warning Shot Response Framework” that explains the conditions under which organizations learn effectively from near-misses. The framework is then applied to contemporary AI governance to explore how future AI-related warning shots—such as a misalignment incident, a contained AI-enabled cyberattack, or a model-weight leak—might be interpreted by the United States, China, and frontier AI labs. The goal is to assess when major powers might slow, halt, or reform their AI development in response to a warning shot, and what governance interventions (e.g., reporting protocols, investigative bodies, crisis communication channels) would make constructive learning more likely. The project is primarily qualitative and structured around a small number of case studies, a cross-case factor analysis, and scenario-based reasoning, making it tractable for a 12-week SPAR cycle while still producing highly policy-relevant insights.
Theory of change
This project advances AI safety by identifying the kinds of events that could realistically cause major powers to slow or halt frontier AI development in a world where AGI poses a potential existential threat and effective governance mechanisms do not yet exist. Many scholars, including those associated with MIRI, have argued that halting or pausing frontier AI development may be necessary to prevent catastrophic outcomes. However, implementing such a halt is extraordinarily difficult because the United States, China, and other actors are locked in a competitive dynamic in which unilateral restraint carries real strategic disadvantages. In practice, a slowdown is most plausible only if a sufficiently severe warning shot occurs—an incident serious enough to shift the political incentives on both sides, similar to the “positive” branch in scenarios like AI 2027. By examining historical near-misses in other high-risk domains, this project identifies the factors that determine when institutions meaningfully change course in response to danger. These insights help clarify which types of AI warning shots might be strong enough to alter the strategic calculus of major powers and what governance mechanisms—such as incident reporting, joint investigations, or crisis communication channels—could increase the likelihood that a warning shot leads to constructive action rather than escalation.
Your role
Mentees will take on a substantial and highly autonomous role in this project, but within a framework where the mentor provides crucial intellectual guidance and overall direction. Compared to my first SPAR project, mentees here will have greater ownership: they analyze the historical warning shots, construct the factor comparisons, propose mechanisms, and potentially build the initial version of the “Warning Shot Response Framework” themselves. They will also help apply the framework to contemporary AI warning-shot scenarios and draft the policy brief and presentation materials. At the same time, the mentor will play an essential role in shaping the research agenda, providing the conceptual structure, steering the interpretation of cases, ensuring rigor in the mechanisms and framework, and guiding the synthesis into AI-relevant insights. The mentor will review each stage of the work closely, refine the analytical structure where needed, and help ensure the final outputs meet a publishable standard. The overall goal is a true collaboration: mentees will drive much of the analytical development, while the mentor provides the strategic, conceptual, and scholarly oversight necessary to produce a coherent and high-quality result.
Prerequisites
Mentee Must-Have Prerequisites • Bachelor’s degree • Strong interest in AI governance, institutional decision-making, historical analysis, or international security • Ability to read and synthesize qualitative research in political science, history, organizational behavior, or risk studies • Comfort working with structured analytical tools (e.g., case study templates, factor coding, scenario frameworks) • Strong analytical writing skills and the ability to produce clear, well-reasoned summaries
Desired Team Composition • At least one mentee with a qualitative or policy background (political science, history, international relations, public policy, or organizational studies) • At least one mentee with strong conceptual reasoning skills (philosophy, psychology, economics, or decision science) who enjoys identifying mechanisms, incentives, and causal pathways
Nice-to-Have Qualifications (Optional but Beneficial) • Master’s degree in a relevant field • Prior coursework in political science, IR, decision theory, organizational behavior, or qualitative methods • Background in qualitative research or comparative case study work • Familiarity with AI governance literature, catastrophic risk studies, or institutional failure analysis • Experience conducting graduate-level or independent research • Ability to create visualizations (e.g., causal diagrams, factor comparison tables, conceptual frameworks)
Location preference
Anywhere on Earth
Application question(s)
- Choose one historical “warning shot” (a near-miss or early sign of danger) from any domain and briefly explain (a) why it qualifies as a warning shot, and (b) what the relevant decision-makers should have learned from it but did not. (150–250 words)
- Some argue that AGI does not pose a major existential threat to humanity. Provide 3 arguments to support this claim (150-250 words).
- Provide one writing sample (link or PDF) that demonstrates your analytical or research skills (up to 5 pages)
About the mentor
Zhamilia (goes by Jama) has seven years of experience at the intersection of data analytics, international policy, and strategy, helping global clients translate complex challenges into clear, actionable solutions. Her work bridges data science, software, and international strategy with a regional focus on Central Asia, the MENA region, China, and Southeast Asia. She has contributed to early-warning forecasting technology, decision-making under uncertainty, and horizon-scanning methodologies, earning multiple scholarships, awards, and co-inventor credits on several patents at Acertas.
She brings extensive experience in agent-based modeling, stakeholder and scenario analysis, and has advised public- and private-sector clients across MENA, India, Indonesia, the ROK, the UK, the US, and Latin America. Her research interests include AI governance, emerging technologies and global order, and AGI-related strategic competition. Jama serves as a Community Reviewer for Frontiers in Political Science, works as a Business Intelligence Analyst at Acertas, and is a Fellow at the TransResearch Consortium. She holds a Doctorate in International Politics and a Master’s in Applied Data Science from Claremont Graduate University, and a Bachelor’s in Business Administration from the American University of Central Asia.