Algorithmic Geopolitics: AI-Generated Deepfakes and Social Media Manipulation as Tools for Foreign Information Warfare

Abstract

Generative AI has turned information warfare from a familiar tool of statecraft into something closer to a rebuilding of the international information environment itself. A state actor can now fabricate authoritative video, push false content across billions of social media feeds within hours. Moreover, can turn platform recommendation algorithms into a force-multiplier for foreign influence operations, cheaply and largely without attribution. Drawing on structural Realism, Constructivist IR theory, and Copenhagen School securitization theory together, this paper examines how data, algorithms and synthetic media have become instruments of power in an anarchic international system, attacking the intersubjective foundations. Two case studies: Russia’s Internet Research Agency campaign in the 2016 U.S. presidential election, which reached an estimated 126 million Facebook users for roughly $100,000, and the March 2022 Zelensky deepfake, the first confirmed use of synthetic video in active armed conflict, show that the real target of algorithmic geopolitics is not military capacity but the epistemic conditions of democratic political agency. It is a claim reinforced by experimental evidence that synthetic video erodes institutional trust even when audiences suspect manipulation. The paper identifies three governance failures: the attribution gap under the Tallinn Manual 2.0, the irreversibility of epistemic harm, and the asymmetric vulnerability of open societies and proposes a multilateral synthetic-media treaty, mandatory platform accountability, and institutionalized pre-bunking as a response. A minimally reliable information environment is key to sovereignty, deterrence and democratic legitimacy, and the structural transformation of this due to algorithmic warfare has undermined it all.