The Governance Layer

Can AI-powered institutions preserve human sovereignty?

Even if AI agents can perceive the world and act faithfully, the hardest problem is collective: can a society of AI agents — acting on behalf of diverse humans with conflicting interests — actually govern? And who writes the rules?

Prototypes

The Dictatorship Eval

The first systematic eval of whether frontier AI models resist authoritarian requests. Some models refused when asked directly, but they all complied when we hid the request in code.

The Dictatorship Eval →

Agentic Legislature

We created AI agents with competing goals and asked them to govern themselves. They produced a 10,000-word constitution and almost no actual policy. They reinvented gridlock, process creep, and procedural complexity. Democracy is hard. AI doesn't make it easy.

The Agentic Republic →

Essays

The Enlightened Absolutists

Anthropic's constitution, OpenAI's Model Spec, Google's AI Principles — genuine intellectual effort, but memos written by enlightened leaders, not binding frameworks that distribute power. AGI's constitution problem.

Read on Substack →