About

Plato 21 is a scenario about what artificial intelligence in the wrong hands means for self-government. The name is a claim: that the oldest recorded warning about democracy (a jury of five hundred, tipped by professional persuaders, voting to kill the man who asked it questions) comes due again in the twenty-first century. It was made by a small group of people who work on these systems from both sides: some who train large models, some who write the rules meant to govern them. We built it because the public conversation about AI and democracy keeps sliding between two useless poles, breathless doom and bored dismissal, and neither one tells a policymaker what to do on Monday.

We chose a scenario, keyed month by month, for a specific reason. Abstraction is where bad thinking hides. It is easy to say a technology "poses risks to democratic institutions" and commit to nothing. It is much harder, and much more honest, to claim that a particular thing can happen by a particular date through a particular chain of cause and effect. Fixing events to a calendar forces that discipline on the writer and offers it to the reader: you can check the mechanism, find the weak link, and argue with the timing. A date is a claim you can be wrong about, which is what makes it worth making.

The four parts are built from different materials, and we want that to be clear. The prologue and Part I are documented and sourced history: the prologue reaching from Athens to Weimar to the social-choice theorists, Part I covering the modern record. Part II is a forecast. Part III is a counterfactual told from 2034. The seam between what happened and what we imagine is deliberately visible, because blurring it would be exactly the kind of thing this project is against.

On the fictional actors. Parts II and III use invented names to keep the argument about mechanisms rather than personalities. Prometheus, Aegis, and Tianji are fictional AI labs and resemble no specific company. Veldova is a fictional mid-size European democracy and resembles no specific country. Most social-media posts rendered through the site, and the "Readers added context" notes attached to them, are fictional composites written for this project: their accounts, handles, and metrics are invented, and no real person or real account is depicted. Each such fictional post deliberately carries one common misconception about democracy (that a majority may do anything, that one manipulated election proves all are fake, that whatever is legal is legitimate, and so on), with the correction attached beneath it in the note. The exceptions are clearly marked. A small number of posts carry an "Archived · documented" badge: those are real, public statements by named people, quoted verbatim, dated, and footnoted to reputable coverage, included because the historical record needs them. Anything without that badge is invented. The notes are part of the argument, not decoration: they are healthy in the documented parts, decay as Part II drifts, and return in the epilogue. Where we name a real instrument, such as the EU AI Act, the Digital Services Act, or the C2PA provenance standard, we mean the real thing. Where we name an actor, we do not.

What we want from you. Share it with someone who can act on it. Argue with it, out loud, in the places where these decisions are actually made. Steelman it before you dismiss it: assume for a moment that the mechanisms are sound and ask what you would do, then decide whether your objection is to the reasoning or only to the discomfort. And if you find something wrong, an event misremembered in Part I or a step in Part II that would not hold, tell us. This is meant to be a document people improve, not a prophecy people admire.

Reach us at hello@plato21.ai. We read everything, and we would rather be corrected than agreed with.