Nuclear Shadows and Digital Accelerants: AI's Role in Great Power Competition
Military competition between major powers has always carried catastrophic risk. AI introduces new accelerants: compressed decision cycles, autonomous weapons that reduce human judgment in the kill chain, and dual-use capabilities that blur civilian and military targets. This report examines how AI changes the risk calculus of great power conflict, evaluates existing arms control frameworks, and proposes a research and diplomatic agenda for AI-specific risk reduction.
Nuclear Shadows and Digital Accelerants: AI's Role in Great Power Competition
Executive Summary
The post-WWII international order has maintained an imperfect but meaningful peace among great powers through deterrence, arms control regimes, and diplomatic channels. AI threatens to disrupt this equilibrium: by compressing decision timelines, introducing new domains of competition (cyberspace, autonomous systems, information warfare), and creating capabilities that existing arms control frameworks were not designed to address. The combination of 12,500 nuclear warheads and AI-accelerated decision-making is among the most dangerous configurations in human history.
How AI Changes the Conflict Risk Calculus
Decision Compression
Nuclear deterrence depends on decision-makers having time to assess threats, consult advisors, and verify attacks before responding. AI-enabled early warning systems and autonomous response protocols could compress this timeline from minutes to seconds — removing the human judgment layer that has prevented nuclear use since 1945.
Autonomous Weapons and Escalation Risk
Lethal autonomous weapons systems (LAWS) operating without human authorization create novel escalation risks: systems that misidentify targets, respond disproportionately, or interact with adversary autonomous systems in unexpected ways — with no human able to intervene in time.
Dual-Use AI Capabilities
AI capabilities developed for civilian purposes (computer vision, NLP, logistics optimization) have direct military applications. This makes export control and technology governance dramatically more complex than in the nuclear era.
Cyber Operations
AI-accelerated cyberattacks on military command-and-control infrastructure create scenarios where one power's early warning systems are disabled, potentially triggering a first strike based on false threat assessment.
Governance Gaps
- No binding international agreement on LAWS after a decade of UN deliberation
- Existing nuclear arms control treaties (New START, INF) do not address AI-enabled delivery systems or autonomous response
- Hotlines and crisis communication protocols designed for the Cold War have not been updated for AI-speed conflicts
- No agreed definitions or red lines for AI use in military contexts
Recommendations
- Negotiate AI-specific risk reduction measures: Extend existing nuclear risk reduction centers to include protocols for AI-enabled false alarms and autonomous weapon incidents.
- Establish international LAWS norms: Push for a binding international agreement prohibiting fully autonomous lethal weapons, with verification mechanisms.
- AI arms control research: Fund dedicated research on verification methods for AI capabilities, analogous to nuclear safeguards.
- Maintain human control requirements: Domestic legislation mandating human authorization for nuclear and other weapons of mass destruction launch decisions, explicitly preventing AI delegation.
Further Reading
- Future of Life Institute: Autonomous Weapons Open Letter (2015)
- Scharre, P. Army of None: Autonomous Weapons and the Future of War (2018)
- RAND Corporation: AI and International Stability (2021)