The Magnitsky Model for AI — Policy Simulator
A hands-on companion to the Phase II theory paper. Choose how an AI system is used to circumvent international law, watch the double veil defeat attribution, then enact Magnitsky-style targeted designations and tune the market and financial levers. The model scores the cascade and returns a verdict — Contained, Circumvented, or Cascaded-to-compliance — with plain reasons.
Choose a circumvention scenario
Each scenario spans a different register of the “double veil” — a corporate form stacked on an algorithmic one. Pick one, then choose who is behind the veil.
The double veil defeats attribution
Before any designation, the ordinary route fails. The corporate veil hides the principal; the algorithmic responsibility gap means no human authored the specific act. Attribution to a person or state cannot be made — so the model routes around the veil rather than trying to pierce it.
Why the ordinary route fails
Enact targeted designations & tune the cascade
Select which jurisdictions enact a Magnitsky-style designation. Their combined market size, banking centrality (dollar-clearing / correspondent access) and emulation propensity drive the cascade. Then tune the four levers.
The cascade
A designation in a central jurisdiction does not stay there. It propagates through market exclusion, the dollar-clearing chokepoint, over-compliant de-risking, and — if the norm is shared — legislative emulation by peer states.
