How Frontier AI Is Governed Today

A map of the institutions, actors and mechanisms that shape frontier AI — and the dependencies between them. Each node is rated by teeth: how much it can actually force a frontier lab to do something.
Lens: organized by governance layer, stacked by real enforcement power. The punchline of the map: the layer doing most of the day-to-day safety work (corporate & voluntary) has the least binding power, while the hardest power (executive / national-security) is fast but unaccountable — and almost every mechanism depends on a weaker one beneath it.
Corporate & voluntary Standards, evals & evidence National law & executive power International coordination Cross-cutting chokepoints
Click any node → see what it depends on (blue) and what depends on it (orange).
Layer

Teeth
What it is
Depends on
What depends on it
The gap / weakness
In the readings