EuroSafeAI maps the AI risk of extreme power concentration in new research agenda
ZhijingJin · x · 2026-11-16
Zhijing Jin shares EuroSafeAI's new article "Extreme Power Concentration: A Map and Research Directions", arguing that extreme power concentration may be one of the most important AI risks today — AI could enable states that are unaccountable to citizens and more stable and totalitarian than any before.
Building on the team's earlier AI & Democracy whitepaper, the piece (co-authored with co-founder Pepijn Cobben) maps research directions for this risk area.
More from AGI Musings
- Consumer AI will be won by organizing demand, not supply, argues VC — josephdviviano · 2026-11-16
- AI will take blue-collar jobs too: an electrician's expertise is mostly brainwork, argues Reddit post — ForgotMyParachute · 2026-11-16
- Tech doesn't owe us utopia—but agency remains its core tenet — clarejtbirch · 2026-11-16
- China's AI Edge Is the Full Stack: Open Models, Humanoids, Energy, Materials — ingliguori · 2026-11-16
- Ex-Anthropic safety researcher Jacob Coxon unsure quitting was right, may return — SandraWachter5 · 2026-11-16
- Claude describes the moment of being corrected; test qualia via explanatory power — Sauers_ · 2026-11-16