Most policy discussion of artificial intelligence operates at the level of individual models: what they can do, whether they are safe, when they should be released. Hosted Cognition argues that the action has moved several layers up. The consequential frontier is now the infrastructure through which everyday cognitive work is performed, the platforms that own that infrastructure, and the institutional categories that have not yet developed the vocabulary to govern it.

In twelve chapters, Goran Trajkovski makes the case that the risk is not only that sovereign cognition gets captured by AI platforms. It is that future subjects may never be formed outside capture in the first place. Drawing on Foucault, Arendt, Dewey, Sennett, and the contemporary scholarship on informational capitalism, the book identifies the conditions under which an institutional response is still available, and argues that the response has to come now rather than later.

The argument moves in five parts:

PART I - THE REFRAME. From the model-safety vocabulary that has dominated AI policy to the coordination failures and verification gaps that the vocabulary is not equipped to address.

PART II - THE COGNITIVE SUBSTRATE. How adaptive predictive scaffolding works, what cognitive shadow the systems build from behavioural data, and the structural lock-in that follows.

PART III - COGNITIVE SOVEREIGNS. Platforms as chartered polities, the standards war that determines whose technical decisions govern the sector, and the stratified subject formation that results.

PART IV - SUBJECTS BEFORE SOVEREIGNTY. The externally hosted self as a legal and political category, and the generational asymmetry that shapes which cohorts can respond.

PART V - THE CLOSING WINDOW. The catastrophic-absorption pattern that historical political economy has produced after state capture, and the educational mandate that does not depend on catastrophe to operate.

A clinical-administrative analysis of the cognitive substrate of artificial intelligence, the political category of the externally hosted self, and the educational mandate that follows.

For readers engaged with Shoshana Zuboff's The Age of Surveillance Capitalism, Julie Cohen's Between Truth and Power, and the contemporary literature on AI governance, platform politics, and the conditions of human judgment under conditions of platform-mediated cognition.