The Hypothesis: Government Without Territory
SIRDA is exploring a distributed transnational government whose jurisdiction would be defined by capability rather than territory and limited to catastrophic and existential AI risk. The proposed functions are specific: establish review obligations, assess contested evidence, adjudicate disputes, and protect the rights of people and organizations affected by its decisions.
The founding question concerns where its authority comes from. Three sources are available, and the design must keep them distinct. Participants can delegate authority by joining. Independent actors can lawfully condition access to what they control on the institution's determinations. States can grant or recognize authority through law and agreement. The people exposed to the risk need a separate route to representation, because none of the three sources speaks for them. The inquiry asks whether this combination can establish a legitimate basis of government that persists beyond continuing agreement among national administrations.
The word government brings concrete obligations into view. Who selects officials and removes them? Who pays? How does a decision become binding? Who can challenge it? What changes the institution's powers, and who authorizes those changes? Each claimed power needs a purpose, an operating mechanism, and an effective route for review.
Participation and the public record would be digital: evidence, decisions, dissents, amendments, and the reasons behind them. People, infrastructure, and remedies still encounter territorial law. Any design has to explain those relationships, especially when a national instruction conflicts with an institutional decision.
The proposed starting structure separates standards, technical assessment, adjudication, and review. Its jurisdiction is bounded by the defined risk, and its own allocation of power remains open to challenge throughout this inquiry.