AI Policy Wiki
Dashboard

RAND — How AGI Could Affect the Rise and Fall of Nations

high confidence · updated 2026-06-06

RAND Research Report (July 2025) presenting eight scenarios at the intersection of AGI-development centralization and geopolitical power shifts.

How AGI Could Affect the Rise and Fall of Nations is a RAND Corporation research report published in July 2025 (Document ID RR-A3034-2) that presents eight illustrative scenarios for how the development of artificial general intelligence could reshape geopolitical power. It is the product of RAND's Technology and Security Policy Center (TASP), part of RAND Global and Emerging Risks (GER), and is authored by Barry Pavel, Ivana Ke, Gregory Smith, Sophia Brown-Heidenreich, Lea Sabbag, Ashwin Acharya, and Yusuf Mahmood.

The report frames its scenarios as tools to stimulate policymaker thinking about AGI's geopolitical impacts rather than as predictions. It describes the work as scenario-based "assumption-based planning" intended to prepare for low-probability, high-impact outcomes.

Framework and scenarios

The report organizes its scenarios along two axes: the centralization of AGI development (multiple actors racing versus one actor leading) and the geopolitical outcome (empowering the United States, empowering its adversaries, disempowering both, or development being halted).

The eight scenarios map onto these axes as follows:

#ScenarioCentralizationOutcome
1Multilateral Coalition of Democracies LeadsDecentralizedUS-empowered
2Cold War 2DecentralizedUS-adversaries empowered
3The Wild FrontierDecentralizedBoth disempowered (non-state actors rise)
4The Corked BottleDecentralizedAGI halted
5The New '90sCentralized (US leads)US-empowered
6Authoritarian AdvantageCentralized (PRC leads)US-adversaries empowered
7The AGI CoupCentralized (any actor)Both disempowered (private actor seizes power)
8Mushroom Cloud ComputingCentralizedAGI halted via war

The AGI Coup scenario (Scenario 7) describes a single private actor seizing geopolitically disruptive power via AGI, a concern the report's authors note had previously been articulated chiefly within AI-safety discourse rather than in official analysis.

Methodology

The report draws on a literature review of the intersection of AI and geopolitics, and on semi-structured interviews with AI researchers and geopolitical experts, including former policymakers. It applies assumption-based planning (ABP), a RAND method, to identify the key assumptions underlying each scenario. It explicitly references prior work, including Aschenbrenner's Situational Awareness, Kokotajlo et al.'s AI 2027, and Hendrycks, Schmidt, and Wang's Superintelligence Strategy.

Key framing points

The report defines AGI as "powerful AI that is capable of performing any intellectual task that humans can perform" and ASI as the "ability of machines to far outperform humans in every field." It references a survey finding that AI researchers place the probability of machines automating all human tasks by 2047 at 50%, and identifies an "intelligence explosion" as a non-trivial possibility that industry participants take seriously.

On current AI use, the report notes that the United States was cited as already using AI for "one-half to three hours of work per week" in many jobs, and cites Brynjolfsson's 14% productivity-gain figure. As a biosecurity signal, it characterizes AI as "on the cusp of being able to meaningfully help novices create known biological threats."

The authors position the report as covering the scenario space rather than advocating a preferred outcome, distinguishing it from Aschenbrenner's Situational Awareness, which advances a single scenario of US nationalization, and from Hendrycks, Schmidt, and Wang's Superintelligence Strategy, which is policy-prescriptive around MAIM, nonproliferation, and competitiveness. The report instead presents the range of possible outcomes for US government audiences.

Relationships

Sources