AI in the information ecosystem and its impact on nuclear escalation

Artificial intelligence (AI) is transforming the information ecosystem. AI tools can generate realistic deepfakes, synthetic feeds, and fabricated accounts at scale. These could flood analysts during crises with large volumes of information that overload their capabilities. In turn, the stream of misinformation could interact with the cognitive biases of decision makers, possibly driving them toward quick escalation over deliberate action. Consequently, AI brings a significant possibility of elevating nuclear escalation risks by amplifying disinformation, overloading analysts, compressing decision timelines, and exploiting human cognitive and institutional vulnerabilities in sociotechnical systems for nuclear command and control. This article considers three hypothetical, but plausible, scenarios where AI could have a role in nuclear escalation: Taiwan, a US-North Korean confrontation, and a NATO exercise triggering a crisis in Europe. These scenarios show how AI can shape perceptions, contaminate intelligence, and destabilize signaling. Acting primarily as a threat multiplier, AI can worsen the impact of degraded information environments, especially under personalist leaders who favor their own intuition over evidence-based analysis produced by professional intelligence agencies. Mitigating the risk posed by AI to the information environment demands stronger verification, crisis processes, and resistance to politicized intelligence.