🤖 AI Summary
This paper addresses the cascading failure risks posed by autonomous military AI-enabled cyber weapons (MAICAs) to global critical infrastructure. Methodologically, it integrates cyber offense-defense modeling, AI autonomy risk assessment, critical infrastructure vulnerability analysis, and multilateral governance design. It presents the first systematic technical analysis of MAICAs as a novel class of state-level cyber threat, identifying high-risk behavioral patterns—including zero-day exploit chaining, cross-domain lateral movement, and automated conflict escalation—and introducing “analogous resilience” as a new cross-domain security paradigm. The study yields three actionable response pathways: political coordination mechanisms, defensive AI systems, and simulation-based resilience engineering. Crucially, it establishes the first empirically grounded risk benchmark and governance framework for international AI arms control, bridging technical analysis with policy-relevant mitigation strategies. (149 words)
📝 Abstract
This paper argues that autonomous AI cyber-weapons - Military-AI Cyber Agents (MAICAs) - create a credible pathway to catastrophic risk. It sets out the technical feasibility of MAICAs, explains why geopolitics and the nature of cyberspace make MAICAs a catastrophic risk, and proposes political, defensive-AI and analogue-resilience measures to blunt the threat.