The Fog of AI: What the Technology Means for Deterrence and War

Artificial intelligence is becoming crucial for national security, initially seeming to bolster deterrence by improving intelligence, speeding assessments, and clarifying signals of capability and resolve. However, AI also creates new avenues for adversaries to undermine deterrence through two primary methods: AI-enabled influence operations and model poisoning. Influence operations, leveraging generative AI to create highly tailored synthetic content and personas, can manipulate public opinion and decision-makers' perceptions, eroding the domestic support vital for democratic resolve. Simultaneously, model poisoning corrupts the training data or live inputs of critical AI systems, leading to skewed intelligence assessments that can cause hesitation or miscalculation during a crisis. The article warns that these vulnerabilities could lead to a 'cognitive fog of war,' where even capable nations fail to deter aggression because their leaders cannot trust their own intelligence or gauge public sentiment accurately. To maintain deterrence in the AI age, policymakers must rapidly develop digital defenses, harden analytic systems against poisoning, and counter influence operations through collaboration between governments, platforms, and researchers.

(Source：Foreign Affairs)