Civic institutions—the rule of law, universities, and a free press—are the backbone of democratic life.
AI’s most dangerous effect is “destructive affordances”: things like speed, scale, automation, and the ability to overpower human intelligence that allow even small actors with minimal resources to challenge large institutions that historically kept society stable.
Institutions are fragile, and AI makes them weaker. The paper argues AI will cause institutional failure & not necessarily out of malevolence. The paper emphasizes that AI does not need agency or intent to cause destruction.
The good news? Human institutions can adapt. They need to be redesigned for AI-scale speed and complexity, be able to verify information in real time, coordinate across borders, govern AI capabilities and deployment & handle systemic risks rather than specific threats.
To me, the EU seems most likely to have a handle on this. It’s also the place that in 2026 is rapidly realising it’s under attack from authoritarians & anti-democratic forces. Some viewed the EU’s AI regulation through the lens of innovation, now it seems a smart move from the point of view of national security.
It’s possible that multiple things can be going on at once, I’m sure globalization & neoliberalism and the lowering of living standards in the Western world is partially responsible too. This research however is about a new threat and the future.