Extreme Power Concentration via AI
AI could enable unprecedented concentration of economic and political power in the hands of a small group — a corporation, government, or individual — foreclosing pluralistic futures.
Scale of the Problem
The top 5 AI labs control models with combined compute greater than all academic research combined.
Potentially existential in the sense of losing long-run option value. A stable global or near-global dictatorship enforced by AI could persist indefinitely, locking out the course corrections that have historically ended oppressive regimes. Toby Ord estimates risks of unrecoverable dystopia at roughly 1 percent this century, most of which runs through AI-enabled concentration.
Why This Is Pressing
Historically, autocratic regimes have been limited by their dependence on soldiers, officials, and workers whose cooperation could be withdrawn. Advanced AI could remove those constraints: AI-enabled surveillance, automated repression, and loyal autonomous militaries would let small groups maintain control without mass human cooperation. If a state, company, or small coalition controls the most advanced AI before governance catches up, the resulting concentration of power could become durable and extremely hard to reverse.
Why It's Neglected
Very neglected as a distinct problem. Some overlap with AI governance and democratic-resilience work, but dedicated capacity is limited. Forethought, RAND, IAPS, and a few academics contribute. Total dedicated funding is under 10 million dollars per year.
Can It Be Solved?
Concrete levers include compute governance, AI-coup-prevention research, preserving democratic checks on AI deployment, export controls on surveillance technology, and binding commitments by labs and governments against AI-enabled repression. Early policy windows exist in the US, UK, and EU.
Research & Solutions
Who Controls the Future? AI-Enabled Power Concentration and Democratic Pluralism
WorldProblems Solved
Sign in to read full documents, vote on solutions, and submit your own research.
What You Can Do
What this problem still needs
AI governance careers at RAND, Carnegie, Brookings, or national AI institutes. Research at Forethought, IAPS, or the Centre for the Governance of AI. Work on compute governance and surveillance-tech export controls. Support democratic-integrity organizations. Congressional and executive-branch staff roles.
Priority Score
20
Breakdown
Fund this problem
Support the researchers and organizations working on Extreme Power Concentration via AI.
Donate to this causeContribute to solving this
Create a free account to submit research, vote on solutions, and follow this problem.
Join for free