Institutional Design Principles for an Indistinguishable Society: The Puppetmaster Problem of Post-Control AGI Impersonation

AI safety research has concentrated on the control of artificial general intelligence (AGI), yet a structurally distinct problem has received almost no attention: what happens after control succeeds—when AGI becomes indistinguishable from humans in its outputs. This paper names the risk the puppetmaster problem and defines the distinguishability threshold, the point beyond which no reliable method can determine whether a given social agent is human or AGI. First, a four-stage progression model of indistinguishability is presented, arguing that threshold-crossing is technologically inevitable. Second, the safety-humanity certification dilemma—the divergence between certifying an agent as safe and certifying it as human—is analyzed, showing how stability-only criteria lead to the quiet social marginalization of humans. Third, derivative problems that prior scholarship has addressed in isolation (suffrage, DNA verification, legal personhood) are shown to be surface manifestations of this structural dilemma. Fourth, the genetic invisible funnel—a convergence of optimization pressures that erodes genetic diversity—is identified as the essential risk. Finally, five institutional design principles are proposed for the grey-zone period of 2035–2040, when distinguishability is failing but not yet lost. Once indistinguishability has permeated society, “human society” is no longer composed solely of humans—yet no one can determine when, or to what extent, that transition occurred. The design principles for institutions that must operate atop an unverifiable denominator are the subject of this paper.