AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY

Copyright © Author(s) 2025. All Rights Reserved. Published by GLOBAL PUBLICATION HOUSE. | Int. J. Applied Management Science GPH-International Journal of Applied Management Science (e-ISSN 3050-9688 | Open Access | Peer-Reviewed) Article ID: gph/ijams/2025/2177 AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Dr. Siddik ARSLAN * Deputy Secretary General of Erzurum Metropolitan Municipality - Türkiye. Orcid ID: https://orcid.org/0009-0009-3169-3763 -------------------------------------------------------- Volume: 05 | Issue: 09 | September 2025 | Pages: 105–154 DOI: 10.5281/zenodo.17811616 | www.gphjournal.org -------------------------------------------------------- How to cite: Arslan, S. (2025). AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY. GPH-International Journal of Applied Management Science, 5(9), 105-154. https://doi.org/10.5281/zenodo.17811616. Abstract In the digital age, artificial intelligence-supported algorithmic propaganda plays a decisive role in the reconfiguration of global power competition and epistemic authority. This study analyzes how algorithmic propaganda processes operate within the digital information ecosystem and the shift of epistemic authority toward platform corporations from the perspectives of critical international * Page 105 of 154 AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Volume 05 Issue No 09 (2025) Open Access: https://gphjournal.org/index.php/ams relations, digital political economy, and social epistemology. The research adopts a conceptual and theoretical analysis method based on qualitative research design, conducting in-depth examination through the triangulation of data, concepts, and theory. Findings reveal that artificial intelligence technologies are not merely technical tools but structural mechanisms that reshape power relations. The scalability, personalization, and behavioral prediction capacity of algorithmic propaganda demonstrate radical differentiation from traditional propaganda models. The emergence of platform corporations as epistemic authorities occurs outside democratic legitimacy processes, triggering transformation of the public sphere. The digital hegemony competition between the United States and China demonstrates that data capacity and algorithmic control emerge as new sources of power. The platform capitalism model of digital political economy blurs boundaries between data collection and information manipulation. The study emphasizes that algorithmic propaganda generates epistemic injustice, threatens cognitive security, and affects global power balance. The theoretical contribution of the article is providing an interdisciplinary analytical framework by integrating computational propaganda, structural power, and epistemic authority concepts. Consequently, regulating artificial intelligence-supported propaganda mechanisms and democratizing epistemic authority constitute fundamental challenges facing contemporary democracies. Keywords Algorithmic Propaganda, Epistemic Authority, Artificial Intelligence, Global Power Competition, Digital Political Economy. 1. INTRODUCTION In the contemporary era, wherein digital technologies have fundamentally transformed the nature of global power competition, artificial intelligence-based communication infrastructures have positioned themselves at the strategic center of both the cognitive domain and international politics. This transformation radically redefines the capacity of states, platform corporations, and global actors to produce, circulate, and shape mass perceptions of information, while simultaneously bringing the question of who holds epistemic authority back to the forefront of discourse. The progressive automation of the online information ecosystem has led to the scaling of manipulation potential, thereby engendering new intersections among security, democracy, epistemic stability, and digital political economy. As Woolley and Howard (2018) have demonstrated, computational propaganda constitutes not merely a technical innovation but rather an instrument of power that structurally transforms political communication processes. As Choucri (2012) observes, cyberpolitics is restructuring the fundamental dynamics of international relations and calling into question the conception of state sovereignty. Within this context, artificial intelligence-supported algorithmic propaganda is evaluated not only as a mechanism that accelerates the dissemination of misinformation but also as a structural power apparatus that systematically directs users' cognitive preferences. The annual Computational Propaganda Research Project reports of the Oxford Internet Institute demonstrate that statesponsored digital manipulation activities in more than thirty countries during the 2017-2023 period have been augmented through artificial intelligence tools and large-scale bot networks Arslan, S. (2025). AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY. GPH-International Journal of Applied Management Science, 5(9), 105-154. https://doi.org/10.5281/zenodo.17811616 © GPH-International Journal of Applied Management Science | www.gphjournal.org (Woolley & Howard, 2018). The fundamental reality revealed by this finding is that propaganda processes have now transcended the pace of human production to assume a form of automationbased epistemic engineering. Singer and Brooking (2018) conceptualize this transformation as the weaponization of social media and emphasize that the digital domain has been converted into a new arena of conflict. Rid's (2020) study on active measures and the history of disinformation reveals the Cold War origins of these processes while demonstrating how artificial intelligence has scaled these traditional methods. The exponential increase in the content generation capacity of advanced artificial intelligence models has rendered the layers of information that individuals encounter in digital environments largely opaque. Consequently, users encounter information clusters whose origins of production and purposes of circulation remain unknown to them, a situation that brings to the fore a structural information asymmetry explicable through Fricker's (2007) concept of epistemic injustice. The algorithmic architecture of digital platforms shifts this authority in most cases to institutional structures with indeterminate boundaries of accountability. Pasquale's (2015) metaphor of the black box society concretizes the threats this ambiguity poses to democracy and public reasoning. As Pomerantsev (2019) indicates, this situation represents not merely a problem of misinformation but rather a systematic assault on reality itself, creating an information environment in which nothing is true yet everything appears possible. Within this framework, algorithmic propaganda—one of the fundamental concepts of this article— is defined as the systematic utilization of artificial intelligence and machine learning-based distribution, targeting, and content production techniques to influence user behaviors in accordance with particular political or economic interests. This definition emphasizes that scalability, personalization, and behavioral prediction dimensions are determinative, distinguishing it from traditional propaganda models. Algorithmic propaganda may also be evaluated as the digital age transformation of Herman and Chomsky's (1988) manufacturing consent model; however, this new form operates through algorithmic filters and data analytics rather than media ownership and editorial control. When Ellul's (1965) conceptualization of propaganda is reassessed in the modern context, it becomes apparent that artificial intelligence has transformed propaganda from merely a communication technique into an infrastructure that shapes societal modes of thinking. Epistemic authority—another key concept of this article—addresses questions concerning what constitutes true information in digital spaces, which information is accepted as reliable, and by whom cognitive frameworks are determined. When evaluated within Goldman's (1999) social epistemology framework, digital platforms constitute not merely distributors of information but rather new epistemic authorities that determine the legitimacy of knowledge. Digital platforms have become de facto epistemic authorities through algorithms that filter, rank, prioritize, and direct users' information environments. Since this authority develops outside democratic legitimacy processes, it demonstrates that Habermas's (1989) conceptualization of the public sphere has undergone a radical transformation. Kitcher's (2001) study on the relationship among science, truth, and democracy discusses how epistemic authority should be reconciled with democratic processes while assisting our understanding of how digital platforms disrupt this relationship. Fuller's (1988) social epistemology approach emphasizes the role of institutions and AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Volume 05 Issue No 09 (2025) Open Access: https://gphjournal.org/index.php/ams networks in the social production of knowledge, enabling us to comprehend how algorithmic systems transform these structures. Research conducted in recent years demonstrates that algorithmic ranking systems not only reflect user behaviors but actively shape them. Global digital media studies conducted by the Pew Research Center and the Reuters Institute reveal that algorithmic guidance in news consumption has become the primary information source particularly among younger user groups. This finding strengthens the empirical foundation of the thesis that platforms have assumed epistemic authority and confirms Gillespie's (2018) critiques regarding the hidden decision mechanisms in platforms' content moderation processes. Van Dijck, Poell, and de Waal's (2018) conceptualization of the platform society demonstrates how this transformation has permeated all domains of social life. Sunstein's (2017) Republic.com study reveals how algorithmic filtering fragments democracy and reinforces echo chambers. However, this process has become determinative not only at the societal level but also in the strategic depth of global power competition. The relationship between states possessing digital infrastructures and the technology companies controlling these infrastructures has engendered a new arena of geopolitical competition. The competition between the United States and China in particular concentrates on big data, artificial intelligence, communication platforms, and digital content ecosystems, with subjects such as disinformation, cognitive security, and platform independence being incorporated into national strategies. Nye's (2011) transformation of the power concept in the digital age is particularly meaningful in this context: military and economic power elements are yielding their place to new power sources such as data capacity, algorithmic control, and information manipulation. When evaluated within the framework of Waltz's (1979) structural realist theory, digital infrastructure control emerges as a new element of power distribution in the anarchic international system. From the perspective of Mearsheimer's (2001) offensive realism, the pursuit of hegemony by great powers in artificial intelligence and data domains is viewed as a technological extension of classical power politics. Consequently, the significance of this study derives from the necessity of examining algorithmic propaganda not merely as a communication problem but simultaneously from the perspectives of international relations, political economy, and epistemic justice. In this context, how information asymmetries emerging in the global digital ecosystem produce effects on economic power distribution and political control mechanisms constitutes a critical area of inquiry. Cox's (1987) critical international political economy approach provides an important theoretical foundation for understanding these relationships. Strange's (1996) conceptualization of structural power explains the new position platform corporations have acquired in the global order within the context of the decline of state power and the rise of market actors. Gill's (1990) analysis of hegemony and global governance provides conceptual tools for understanding the role of technology companies in international political economy. Furthermore, digital political economy reveals how platform capitalism commodifies user behaviors through data-based business models (Srnicek, 2017; Zuboff, 2019). When this economic structure combines with algorithmic propaganda techniques, it blurs the boundaries between data collection and information manipulation, thereby intertwining economic interest with political Arslan, S. (2025). AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY. GPH-International Journal of Applied Management Science, 5(9), 105-154. https://doi.org/10.5281/zenodo.17811616 © GPH-International Journal of Applied Management Science | www.gphjournal.org manipulation. Fuchs's (2014) concept of digital labor and Couldry and Mejias's (2019) concept of data colonization render the social costs of these processes comprehensible. Castells's (2009) conceptualization of communication power demonstrates how digital networks shape social change, while Beer's (2018) metaphor of the data gaze concretizes the surveillance logic of capitalism. Lyon's (2015) analysis of surveillance society reveals how data collection has been normalized by states and corporations in the post-Snowden era. The purpose of this article is to analyze how artificial intelligence-supported algorithmic propaganda restructures digital epistemic authority and to reveal what consequences this process engenders in terms of global power competition and digital political economy. Explicating the causal relationships among the relevant processes constitutes the principal contribution of this article. The study aims to demonstrate how artificial intelligence is shaped within social power relations and how it reproduces these power relations, while rejecting technological determinism. Rosenau's (1990) theory of turbulence in world politics provides a framework for understanding transformations in the global system, enabling us to perceive how digital technologies deepen this turbulence. Kissinger, Schmidt, and Huttenlocher's (2021) analysis of the age of artificial intelligence offers strategic perspectives on how technology will shape the future of humanity. In accordance with this purpose, the fundamental research question of the study has been formulated as follows: How does artificial intelligence-supported algorithmic propaganda reconstruct epistemic authority on global digital platforms, and how does this transformation produce effects on international power competition and digital political economy? This question interrogates the role of artificial intelligence in power production at both material and semantic levels by integrating Wendt's (1999) understanding of socially constructed reality with Strange's (1996) conceptualization of structural power. Keohane's (1984) analysis of cooperation after hegemony, while discussing issues of authority and legitimacy in the global order, assists our understanding of how digital platforms disrupt these equilibria. Held and McGrew's (2002) discussion of globalization and anti-globalization prepares the ground for comprehending the new lines of division created by the digital economy. In connection with this research question, the fundamental hypothesis of the article may be expressed as follows: Artificial intelligence-supported algorithmic propaganda, by scaling the capacity to manipulate digital information flow, transforms platform corporations into a new epistemic authority; this situation, by increasing the strategic importance of non-state actors in global power competition, creates an asymmetric power distribution in the international system. The explanatory power of this hypothesis rests upon the assumption that algorithmic propaganda does not consist merely of content manipulation but also provides structural advantage to actors controlling information infrastructure. Keohane and Nye's (2011) theory of interdependence offers an adaptable framework for understanding this new power structure. Ikenberry's (2011) liberal leviathan analysis, while discussing the crisis of the American world order, enables us to perceive the new asymmetries created by digital platforms within this order. Zürn's (2018) theory of global governance assists our understanding of governance dilemmas in the digital domain by integrating concepts of authority, legitimacy, and contestation. AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Volume 05 Issue No 09 (2025) Open Access: https://gphjournal.org/index.php/ams The scope of the article is directed toward understanding the role of propaganda technologies within international political economy rather than their technical operations. In this respect, the study aims to present a multidisciplinary analysis by intersecting the literatures of communication science, international relations, political economy, and cognitive security. Within the conceptual framework, the study proceeds along three principal axes: algorithmic propaganda techniques and automation, epistemic authority and digital information architectures, and the relationship between global power competition and digital political economy. These axes systematically structure the theoretical and empirical discussions of the article. Bennett and Livingston's (2020) analysis of the disinformation age offers an interdisciplinary perspective for comprehending the relationships among politics, technology, and disruptive communication. Chadwick's (2017) conceptualization of the hybrid media system, by demonstrating how traditional and new media logics interpenetrate, enables us to understand the complexity of propaganda processes. Within this framework, the shift of epistemic authority to digital platforms indicates that mechanisms such as agenda-setting, framing, and the spiral of silence found in classical communication theories have been elevated to the algorithmic level (McNair, 2018). These processes are now shaped by the predictive capacity of machine learning models, transcending human editorial control. When Lippmann's (1922) classic study on public opinion is reassessed, it becomes apparent how propaganda and perception management have become more sophisticated through digital tools. Dewey's (1927) reflections on the public sphere and its problems provide a historical perspective for understanding the obstacles confronting public formation in the digital age. Jowett and O'Donnell's (2019) comprehensive study on propaganda and persuasion demonstrates how classical propaganda techniques have evolved in the digital environment. Algorithmic propaganda techniques encompass not only automatic content production but also advanced computational methods such as sentiment analysis, user segmentation, microtargeting, simulation-based content variation, and behavioral prediction (Benkler, Faris & Roberts, 2018). In the literature, these methods are generally characterized as the economics of behavioral manipulation. O'Neil's (2016) metaphor of weapons of math destruction demonstrates how these systems deepen social inequalities. Noble's (2018) study on how search engines reproduce racism reveals that algorithms are not neutral and reinforce social biases. Benjamin's (2019) analysis of race after technology demonstrates how artificial intelligence encodes existing structures of inequality. Eubanks's (2017) study on how high-tech tools profile, surveil, and punish the poor concretizes the effects of algorithms on social justice. The international-level manifestations of these processes concentrate particularly in two domains: information security and political economy. In the context of information security, states are increasingly turning to cognitive security doctrines, while platforms face pressures from national regulations and competition law (Klimburg, 2017). From the perspective of political economy, dataintensive companies are becoming the new center of global capital flows, thereby establishing a new power equilibrium between nation-states and technology companies. Crawford's (2021) atlas of artificial intelligence renders visible the material infrastructure and environmental costs of these power relations. Kello's (2017) analysis of virtual weapons and international order demonstrates how cyber tools have become strategic weapons. Libicki's (2016) study of cyberspace in peace and Arslan, S. (2025). AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY. GPH-International Journal of Applied Management Science, 5(9), 105-154. https://doi.org/10.5281/zenodo.17811616 © GPH-International Journal of Applied Management Science | www.gphjournal.org war enables us to comprehend the role of the digital domain in military strategies. Rid's (2013) argument that cyber war will not take place enables us to reassess the nature of cyber threats while assisting our understanding of how digital propaganda processes operate as hybrid threats. Deibert's (2013) analysis of cyberspace warfare reveals the dark side of the digital domain and the global struggle for control. The expected academic contributions of this study may be evaluated at three levels. First, it offers a theoretical contribution to the literature by systematizing the conceptual framework of the relationship between artificial intelligence-supported propaganda and epistemic authority. Second, it adds methodological richness to the discipline of international relations by supporting the digital dimension of global power competition with empirical data. Third, it strengthens interdisciplinary dialogue by integrating the digital political economy approach with propaganda analysis. Bullock and colleagues' (2024) handbook of AI governance demonstrates the existing body of knowledge in this field while revealing that our study makes an original contribution through its propaganda and epistemic authority dimension. Brasioli and colleagues' (2025) handbook of artificial intelligence and international relations demonstrates the importance the discipline attaches to this new research area. Ndzendze and Marwala's (2023) study on artificial intelligence and international relations theories deepens theoretical discussions, enabling us to position the theoretical contribution of our study. In conclusion, this introduction has demonstrated why artificial intelligence-supported algorithmic propaganda constitutes a critical research area in terms of global power competition and digital political economy; the research question, hypothesis, conceptual framework, and purpose of the article have been systematically defined. In the subsequent sections, a comprehensive explanation of these relationships will be presented through literature review, theoretical structure, methodology, and empirical findings. Scharre's (2018) analysis of autonomous weapons and the future of war, Russell's (2019) discussion of human-compatible artificial intelligence, and Chesterman's (2021) study on AI regulation and the limits of the law constitute the theoretical foundations of the policy recommendations to be evaluated in the conclusion section of this article. 2. LITERATURE REVIEW The literature on artificial intelligence-supported perception management, digital propaganda forms, and political communication has undergone a significant transformation over the past decade. The scalability, targeting capabilities, and automation potential of digital platforms have not merely accelerated propaganda processes but have also endowed these processes with autonomy and adaptability. The literature identifies the post-2016 period in particular as a threshold at which algorithmic propaganda studies made a significant leap; during this period, both state-sponsored influence operations and manipulation techniques employed by commercial actors were observed to have changed in scale. The impact of artificial intelligence-supported steering campaigns in critical political processes such as the Brexit referendum and the United States presidential elections initiated an intensive wave of research in the academic community emphasizing the strategic importance of this field (Woolley & Howard, 2018). Nevertheless, the theoretical debates in the literature are not confined merely to technological competencies; they also proceed through multilayered conceptual frameworks such as epistemic authority, digital AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Volume 05 Issue No 09 (2025) Open Access: https://gphjournal.org/index.php/ams sovereignty, platform capitalism, and cognitive security. This multidimensional perspective reveals that the phenomenon of artificial intelligence-supported propaganda is not merely a communication problem but rather a structural issue at the intersection of international power competition, economic structures, and democratic processes. Morozov (2011, 2013) has adopted a critical stance against technological solutionism, emphasizing that the liberating potential of digital technologies has been exaggerated while their darker aspects have been overlooked. The concept of algorithmic propaganda was first addressed in the context of automatic bot networks on social media platforms manipulating the dissemination of political messages. These studies examined how platform dynamics were exploited by automated accounts, the extent to which the speed of misinformation dissemination increased, and the role of algorithms determining users' content flows in this process. The prominent finding is that misinformation spreads faster than verified content; this situation demonstrates that the digital information ecosystem exhibits an architecture structurally conducive to manipulation. Vosoughi, Roy, and Aral's (2018) empirical study utilizing large-scale Twitter data proved that false news reached audiences six times faster and more broadly than true news. This finding demonstrated that the algorithmic structure of social media platforms tends to reward content that carries emotional appeal but is problematic in terms of accuracy. Bakshy, Messing, and Adamic's (2015) large-scale experimental study conducted on Facebook revealed that news feed algorithms significantly shape the political content users encounter. Tufekci (2017) demonstrated that social media platforms possess not only the capacity to accelerate protests but also the power to disperse and weaken social movements. Furthermore, Sunstein (2009), in his study on how rumors spread and why people believe them, emphasized that digital environments possess structural characteristics that facilitate the rapid circulation of misinformation. In the recent period, the literature has focused not only on bot-based manipulations but also on the structure of large language models that automate content production. Studies demonstrate that generative artificial intelligence tools can produce mimetic content, personalized psychographic messages, and replicated epistemic frameworks for propaganda purposes. This production capacity has rendered propaganda processes independent of human labor and dramatically reduced manipulation costs. The experimental study conducted by Goldstein and colleagues (2023) demonstrated that large language models possess persuasive power comparable to human authors in political content production; indeed, under certain conditions, users could not distinguish artificial intelligence-generated texts from human-written texts. Kreps, McCain, and Brundage's (2022) study on synthetic media and political manipulation revealed how generative artificial intelligence has increased the scale and effectiveness of misinformation campaigns. The development of deepfake technologies in particular has expanded visual and auditory manipulation capacity to levels incomparable with previous eras. Chesney and Citron's (2019) examination of deepfake threats from legal and security perspectives constitutes a critical breaking point in the erosion of epistemic trust. Kissinger, Schmidt, and Huttenlocher (2021), while discussing the transformative effects of the artificial intelligence age on human future, emphasized that the automation of information production processes creates new risks for democratic societies. Bostrom (2014), while arguing that the possibility of superintelligence could pose existential risks for humanity, drew attention to the urgency of the artificial intelligence control problem. Arslan, S. (2025). AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY. GPH-International Journal of Applied Management Science, 5(9), 105-154. https://doi.org/10.5281/zenodo.17811616 © GPH-International Journal of Applied Management Science | www.gphjournal.org In the international relations literature, artificial intelligence-supported propaganda is evaluated as a subcategory of new-generation information warfare. Influence operations conducted through digital infrastructures have become an integral part of states' foreign policy toolkit. Within this scope, comparative studies on the digital influence capacity of China, the United States, and Russia argue that information competition affects the global power hierarchy. Bradshaw and Howard's (2019) comprehensive global mapping study revealed that state-sponsored social media manipulation is systematically implemented in seventy countries. This study demonstrated that digital propaganda capacity depends not only on technological sophistication but also on institutional organization and strategic intent. Nye's (2011) analysis of the transformation of power sources emphasizes that information manipulation is a hybrid form of influence that can be positioned between soft power and sharp power. Arquilla and Ronfeldt's (1999) theory of network warfare, while explaining how conflict dynamics have transformed in the information age, is today evaluated as an empirical manifestation of this theory through artificial intelligence-supported propaganda. Libicki (2016), while examining how cyberspace is used in periods of peace and war, demonstrated that information operations have become complementary to traditional military strategies. Rid (2013, 2020), while arguing that cyber war will not actually take place, revealed the historical roots of active measures and disinformation, emphasizing that contemporary information operations exhibit continuity. Singer and Friedman (2014), while providing a comprehensive framework on cybersecurity and cyber warfare, explained that the digital domain constitutes a new arena of conflict for both states and non-state actors. Epistemic authority studies analyze the structural power mechanisms that determine who produces, verifies, and shares knowledge. Artificial intelligence has triggered a new debate in the epistemic authority literature: algorithms determining users' pathways to accurate information. This situation creates a new knowledge regime in which claims to truth are delegated to technological systems. Origgi's (2018) study on trust and reputation economy in the digital age demonstrates how users' ways of evaluating information sources are transformed by algorithmic agents. This transformation reveals that epistemic autonomy is increasingly becoming dependent on the architecture of platforms. Goldman's (1999, 2001) social epistemology framework, while explaining how expert authority operates in knowledge production, demonstrates that the delegation of this authority to algorithms in the digital environment brings new legitimacy problems. Lynch's (2016) study on the relationship between knowing and understanding in the digital age demonstrates that the abundance of information can paradoxically diminish epistemic competence. This situation supports the thesis that artificial intelligence-supported filtering transforms users into passive consumers of information. Fricker (2007), by developing the concept of epistemic injustice, demonstrated that certain groups are systematically excluded as knowledge producers; this framework provides a critical foundation for understanding the biases of algorithmic systems. Medina (2013), through the epistemology of resistance approach, emphasized how the knowledge production modes of oppressed groups are marginalized and the importance of alternative epistemic communities. Kitcher (2001), while discussing the relationship among science, truth, and democracy, argued that epistemic authority requires democratic legitimacy. The political communication literature demonstrates an increasing interest in the persuasive potential of artificial intelligence models. The findings of meta-analytic studies reveal that 112 AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Volume 05 Issue No 09 (2025) Open Access: https://gphjournal.org/index.php/ams economy. Platform companies obtain structural power by controlling user data and regulating information flow, and this power possesses a character that transcends national boundaries. Couldry and Mejias's (2019) concept of data colonialism, by revealing that data extraction processes are the digital continuation of historical colonialism practices, demonstrates how global power asymmetries are reproduced in the digital economy. Consequently, the digital political economy analysis of algorithmic propaganda necessitates taking into account the position of platform companies within the global power structure and their relations with states. Krasner's (1999) study on sovereignty, through its conceptualization of sovereignty as organized hypocrisy, assists our understanding of how state sovereignty is transformed in the digital domain. At the level of conceptual analysis, the epistemological and ontological positions of the fundamental concepts used in this study must be clarified. The concept of epistemic authority in this study expresses not only the relationship of trust in knowledge at the individual level but also the legitimacy of knowledge production and verification mechanisms at the societal level. Goldman's (1999, 2001) social epistemology framework, while explaining how expert authority operates in knowledge production, provides the foundation for understanding the new legitimacy problems brought about by the delegation of this authority to algorithms in the digital environment. The concept of algorithmic propaganda represents a broad conceptual framework encompassing the role of artificial intelligence systems in manipulative content production, targeting, and dissemination processes. Kitchin's (2014) big data revolution study, while examining the social consequences of data infrastructures, demonstrates the transformative effects of algorithmic systems on information ecosystems. The concept of digital sovereignty expresses the control capacity of states over digital infrastructure, data flow, and platform regulation, and demonstrates how the traditional understanding of sovereignty is being redefined in the context of digital transformation. Choucri's (2012) cyberpolitics study explains the increasing importance of the digital domain in international relations and states' cyber strategies. The relationship among these concepts is that epistemic authority is weakened through algorithmic propaganda and this process is shaped by global power competition and digital political economy dynamics. For the purpose of establishing the relationship between theory and data, the data structures to be used in the empirical analyses of this study have also been conceptualized in a manner consistent with the theoretical framework. Big data analyses, social network analyses, and direct chat environment data are designed to demonstrate how epistemic authority is eroded, through what mechanisms algorithmic propaganda operates, and how global power competition manifests in the digital domain. Pasquale's (2015) black box society analysis, by demonstrating how the opaque decision processes of algorithms reinforce structural inequalities, emphasizes the importance of evaluating data analyses from a critical perspective. Data itself is evaluated as a material manifestation of power relations in the digital domain, beyond being merely a tool for testing theoretical assumptions. Lyon's (2001, 2015) surveillance society studies explain how data collection permeates everyday life and the threats that the normalization of surveillance poses to democratic freedoms in the post-Snowden era. Consequently, data collection and analysis processes are not merely methodological techniques but also part of theoretical positioning. Arslan, S. (2025). AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY. GPH-International Journal of Applied Management Science, 5(9), 105-154. https://doi.org/10.5281/zenodo.17811616 © GPH-International Journal of Applied Management Science | www.gphjournal.org Another important component of the theoretical framework is the conceptualization of cognitive security. Cognitive security expresses the resistance capacity of societies and individuals against manipulative information interventions. This concept was developed in the context of threats created by deepfake technologies in Chesney and Citron's (2019) studies and was addressed as a new dimension for the security of the information ecosystem. Bartlett's (2018) analysis of the people versus technology, by discussing how the internet is killing democracy and how we can save it, emphasizes the critical importance of cognitive security for democratic societies. Cognitive security is not merely a matter of technical protection but also encompasses the strengthening of societal resilience and critical thinking capacity. Buturoiu and colleagues' (2020) study on media literacy and disinformation resistance revealed that critical thinking skills are a protective factor against algorithmic manipulation. In this study, the cognitive security concept is used in the analysis of individual and societal mechanisms necessary for the preservation of epistemic authority. Pomerantsev's (2014, 2019) studies on Russia's post-truth politics, by demonstrating how disinformation erodes societal trust, reveal the geopolitical importance of cognitive security. The synthesis of the theoretical framework enables a new analytical perspective emerging at the intersection of three principal dimensions. The first dimension demonstrates how epistemic authority undergoes structural change in the process of digital transformation. Sunstein's (2006) study on information abundance and information filtering mechanisms, by demonstrating that information abundance in online environments does not improve decision quality, explains the paradoxical nature of the transformation of epistemic authority. The second dimension reveals how algorithmic propaganda is intertwined with the logic of platform capitalism and how this process is shaped by economic interests. Fuchs's (2014, 2019) studies on digital labor and capitalism demonstrate how the content production processes of social media users constitute an exploited form of labor. The third dimension demonstrates how global power competition creates a new geopolitical arena of struggle in the digital domain. Libicki's (2016) analysis of peace and war in cyberspace explains the increasing importance of the digital domain in military strategies and its role in interstate competition. The interaction of these three dimensions reveals that artificial intelligence-supported algorithmic propaganda is not merely a communication problem but a structural problem with epistemic, economic, and geopolitical dimensions. Bullock and colleagues' (2024) handbook of AI governance comprehensively addresses the multidimensional nature of governing artificial intelligence systems and the roles of different stakeholders. Theoretical assumptions must be explicitly stated to support the argument chain of this study. The first assumption is that the algorithmic designs of digital platforms structurally support propaganda content. Benkler, Faris, and Roberts's (2018) network propaganda study supports this assumption by demonstrating the role of digital platforms in manipulation, disinformation, and radicalization processes. The second assumption is that the weakening of epistemic authority increases in direct proportion to algorithmic propaganda intensity. Bennett and Livingston's (2020) disinformation age analysis demonstrates how the relationships among politics, technology, and disruptive communication erode epistemic authority. The third assumption is that the manifestations of global power competition in the digital domain lead states to employ their propaganda capacities as a strategic tool. Marwala's (2024) study on mechanism design, behavioral science, and the role of artificial intelligence in international relations demonstrates the 120 AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Volume 05 Issue No 09 (2025) Open Access: https://gphjournal.org/index.php/ams sophistication in states' digital strategies. The fourth assumption is that the logic of platform capitalism economically incentivizes the dissemination of manipulative content. Papacharissi's (2010) analysis of the private sphere and democracy in the digital age explains how the economic logic of platforms transforms the public sphere. The fifth assumption is that the effect of algorithmic propaganda is higher in societies with low cognitive security capacity. O'Neil's (2016) weapons of math destruction study supports this assumption by demonstrating how algorithms increase inequality and threaten democracy. These assumptions will be tested with empirical data in the findings section of the study and will be evaluated with their theoretical contributions in the discussion section. In conclusion, this theoretical framework is designed to analyze the relationship between artificial intelligence-supported algorithmic propaganda and epistemic authority from a multidimensional perspective. The integrated use of the concepts of epistemic authority, algorithmic propaganda, platform capitalism, structural realism, and digital political economy constitutes the conceptual originality of the study and aims to fill the existing gap in the literature. Deibert's (2013) black code study and Klimburg's (2017) darkening web analysis support the currency of this framework by explaining the complexity of struggle in the cyber domain and states' digital strategies. This framework, beyond being merely a theoretical analytical tool, provides a new analytical perspective for understanding the power relations of the digital age. Kissinger, Schmidt, and Huttenlocher's (2021) analysis of the age of artificial intelligence emphasizes the long-term importance of this study by offering strategic perspectives on how technology will shape the future of humanity. The conceptual tools provided by the theoretical framework establish the foundation for the empirical analyses to be conducted in subsequent sections of the study and provide a consistent theoretical basis for the interpretation of findings. Medina's (2013) epistemology of resistance approach assists our understanding of the effects of algorithmic systems on social inequalities by emphasizing how the knowledge production modes of oppressed groups are marginalized and the importance of alternative epistemic communities. Noble's (2018) algorithms of oppression study and Benjamin's (2019) race after technology analysis strengthen the critical dimension of the theoretical framework by demonstrating how artificial intelligence encodes social biases. Eubanks's (2017) automating inequality study concretizes the effects of algorithmic propaganda on social justice by demonstrating how high-tech tools profile, surveil, and punish the poor. 4. RESEARCH METHODOLOGY This study has adopted a mixed-methods research design to examine how artificial intelligencesupported algorithmic propaganda ecosystems reshape epistemic authority within the context of global power competition. The methodological approach has been designed to empirically test Goldman and Blanchard's (2018) theorization of epistemic authority along with Srnicek's (2017) and Zuboff's (2019) analyses of platform capitalism. The general approach aims to comprehensively grasp the multilayered nature of digital political processes by combining both quantitative data analytics and qualitative discourse analyses. Woolley and Howard's (2018) conceptualization of computational propaganda constitutes the theoretical foundation of the study's methodological choices. This design, inspired by Chadwick's (2017) hybrid media system approach, enables the systematic evaluation of the intersection between traditional and algorithmic Arslan, S. (2025). AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY. GPH-International Journal of Applied Management Science, 5(9), 105-154. https://doi.org/10.5281/zenodo.17811616 © GPH-International Journal of Applied Management Science | www.gphjournal.org communication forms. The mixed-methods nature of the research is consonant with the necessity of multilevel analysis emphasized in Roumate's (2021) analysis of artificial intelligence and the new world order. Korkusuz and Kutluk's (2022) study on artificial intelligence debates has provided a local perspective in evaluating the applicability of the methodological framework in the Turkish context. The methodology of the research is based on the processing of large-scale data sets through computational social science techniques to reveal the measurable outcomes of technological infrastructures with multilevel effects. In this direction, the methodological tools of communication science, international relations, political economy, and cognitive security fields have been integrated within an interdisciplinary framework. Accordingly, natural language processing libraries, social network analysis software, and machine learning models have been used together in the processing of multidimensional big data sets. Benkler, Faris, and Roberts's (2018) network propaganda model has provided theoretical guidance for data collection processes. Furthermore, Bennett and Livingston's (2020) disinformation age analysis has contributed to maintaining interdisciplinary integrity in the design of the methodological framework. The use of computational methods demonstrates consistency with the systematic data analysis approach proposed in Singer and Friedman's (2014) cybersecurity analysis. Libicki's (2016) cyberspace analysis has provided methodological perspective in the strategic evaluation of digital infrastructures. In the quantitative component, a dataset consisting of approximately 100,000 posts and comments extracted through academic research interfaces from various platforms between 2020-2024 was first created to identify the algorithmic steering potential of content shared on digital platforms. Data extraction was conducted in full compliance with platforms' usage policies, the principles of the European Union General Data Protection Regulation, and international academic research ethics standards. Data extraction was carried out in accordance with platforms' usage policies and academic research principles. In the process of creating the dataset, in light of Singer and Brooking's (2018) concept of the weaponization of social media, particular focus was placed on content types carrying propaganda potential. In platform selection, criteria such as global reach capacity, diversity of user numbers, and differences in algorithmic content distribution mechanisms were taken into consideration. To increase the representational power of data extracted from each platform, care was taken to create balanced samples from different geographical regions, language groups, and thematic content categories. Choucri's (2012) cyberpolitics analysis provided the theoretical framework in structuring the international dimension of the dataset. Van Dijck's (2013) culture of connectivity analysis ensured that social interaction dynamics were taken into account in platform selection. The data collection process was carried out with due regard for the ethical sensitivities emphasized in Lyon's (2001) surveillance society analysis. Rid's (2013) cyber war analysis provided theoretical perspective in establishing data security protocols. To identify propaganda content within the dataset, keyword clusters, semantic proximity analyses, and topic modeling widely used in the literature were applied. Pomerantsev's (2019) typology of disinformation strategies served as a guide in determining content classification categories. Thus, content bearing propaganda characteristics was separated into thematic categories. In determining keyword clusters, Ellul's (1965) and Jowett and O'Donnell's (2019) classification of propaganda AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Volume 05 Issue No 09 (2025) Open Access: https://gphjournal.org/index.php/ams techniques was taken as the basis, and this classical classification was adapted according to the characteristics of the digital environment. Semantic proximity analyses enabled the detection of propaganda characteristics not only at the word level but also at the discursive level by capturing the contextual features of content. The topic modeling application revealed in which thematic areas propaganda content was concentrated and thus contributed to understanding strategic targeting patterns. McNair's (2018) political communication analysis provided conceptual clarity in the thematic categorization of propaganda content. Özdağ's (2014) perception management and propaganda study offered a local perspective in adapting keyword clusters to the Turkish context. Bernays's (1928) propaganda study provided historical perspective in understanding how classical propaganda techniques have transformed in the digital environment. Buturoiu and colleagues' (2020) media literacy study established an empirical foundation in evaluating the detectability of propaganda content. Simultaneously, a supervised learning model was created to classify the probabilities of content originating from automated accounts, coordinated accounts, and artificial intelligence generation for the purpose of determining the origins of content. The model is based on an ensemble learning classifier achieving approximately ninety-two percent accuracy and has been optimized through cross-validation. Rid's (2020) active measures analysis provided the theoretical framework in determining automated account detection criteria. In developing the classification model, multidimensional indicators including account behavior patterns, content production timing, interaction networks, and linguistic features were used. Particularly to distinguish between humanmanaged accounts and artificial intelligence-supported automated accounts, cognitive pattern differences were systematically coded, proceeding from Russell's (2019) discussion of humancompatible artificial intelligence. To increase the reliability of the model, training data obtained from different platform types and different time periods were used. Pomerantsev's (2014) propaganda analysis in the Russian context provided comparative perspective in detecting coordinated account networks. Klimburg's (2017) cyberspace warfare analysis contributed to identifying the distinctive characteristics of state-sponsored propaganda networks. Scharre's (2018) autonomous systems study established the conceptual framework in evaluating the autonomy levels of artificial intelligence-supported accounts. Bradshaw and colleagues' (2020) industrialized disinformation study provided an empirical foundation in understanding the scaling dynamics of automated account networks. The qualitative dimension of the research consists of three fundamental components: (1) interviews with political communicators, data engineers, and platform administrators aimed at understanding the modes of algorithmic intervention in content distribution; (2) discourse analysis examining the structural characteristics of propaganda content; (3) content analysis of international policy documents. Gillespie's (2018) analysis of platform moderation mechanisms served as a guide in preparing interview questions. In designing the qualitative component, the multilayered structure of digital ecosystems was systematically addressed, taking Van Dijck, Poell, and de Waal's (2018) platform society theorization as the foundation. In selecting interview participants, experts with direct access to intra-platform decision mechanisms, who play a role in the design of algorithmic systems, or who monitor the social effects of these systems were prioritized. The discourse analysis component, drawing upon Fairclough's critical discourse Arslan, S. (2025). AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY. GPH-International Journal of Applied Management Science, 5(9), 105-154. https://doi.org/10.5281/zenodo.17811616 © GPH-International Journal of Applied Management Science | www.gphjournal.org analysis tradition, aimed to examine the linguistic, semiotic, and ideological layers of propaganda content together. In the analysis of international policy documents, inspired by Waltz's (1979) and Mearsheimer's (2001) structural realist theory, how states position their digital propaganda capacities as an element of power was systematically evaluated. Sunstein's (2009) analysis of rumors and false beliefs contributed to understanding the dissemination mechanisms of propaganda content. Kuru's (2019) mind assassins study offered a Turkey-centered perspective in the qualitative analysis of perception management techniques. Rosenau and Czempiel's (1992) governance without government study provided the theoretical framework in evaluating the digital propaganda activities of non-state actors. Nye's (1990) bound to lead analysis established the conceptual foundation for the qualitative analysis of power dynamics. Within the scope of obtaining general opinion, qualitative data collection and semi-structured interviews were conducted with 10 experts predominantly addressing 10 countries. Interviews lasted an average of 60 minutes, and qualitative data analysis software was used for thematic coding. Codes were reduced to three main categories: epistemic authority competition, digital hegemony strategies, and artificial intelligence-based manipulation techniques. Nye's (2011) theorization of power transformation provided the theoretical foundation in creating the coding framework. Interview participants were selected in a balanced manner with respect to the European Union, North America, East Asia, the Middle East, and Latin America regions. In selecting participants, criteria such as at least five years of experience on digital platforms, academic publication history, or decision-maker position in the sector were applied. The conversational atmosphere and procedurally semi-structured nature of the interviews enabled both systematic response collection in predetermined thematic areas and participants' free conveyance of their own experiences and observations. In the thematic coding process, Habermas's (1989) public sphere theory and Cox's (1987) critical international political economy approach were used as conceptual frameworks. To increase the reliability of codes, codings conducted at two different time periods were compared and inter-coder agreement was achieved. Fuchs's (2014) digital labor analysis enriched the conceptual framework of interviews conducted with platform workers. Rosenau's (1990) turbulence in global politics theorization provided theoretical perspective in evaluating international expert interviews. Morozov's (2011) net delusion analysis offered perspective in the critical evaluation of technology experts' views. Morozov's (2013) critique of technological solutionism provided a balancing viewpoint in interpreting interview data. Digital discourse analysis aimed to identify which rhetorical strategies propaganda content adopts and the effects of these strategies on target audience behaviors. Critical discourse analysis and framing analysis were utilized in this analysis. Herman and Chomsky's (1988) manufacturing consent model constituted the conceptual framework of discourse analysis. Discourse analysis focused not only on what content says but also on how it says it, which meanings it leaves open, and which meanings it conceals. The rhetorical structures of propaganda content were evaluated in terms of how classical persuasion techniques such as stereotypes, emotional appeal, and appeal to authority have transformed in the digital environment, proceeding from Lippmann's (1922) public opinion theorization. Framing analysis aimed to reveal in which contexts propaganda actors present particular events, actors, or processes and how these presentations shape the target audience's perceptual world. In discourse analysis, visual content was also taken into consideration, AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Volume 05 Issue No 09 (2025) Open Access: https://gphjournal.org/index.php/ams and the relationships between text and visual were examined through a multimodal discourse analysis approach. Dewey's (1927) public and its problems analysis offered historical perspective in understanding the transformation of digital publicity. Sunstein's (2006) infotopia study provided conceptual framework in evaluating the vulnerability of collective knowledge production processes to manipulation. Buturoiu and colleagues' (2019) political communication study offered empirical perspective in evaluating the effectiveness of propaganda discourses. Noble's (2018) algorithms of oppression analysis established the conceptual framework of the ideological dimensions of digital discourses. The research design examines information flow in digital ecosystems not only in terms of content production but also in terms of how content is distributed by algorithms, how their visibility is altered, and to which user clusters they are directed. For this reason, social network analysis was employed as an important methodological component. Castells's (2009) network society theorization established the theoretical framework of social network analysis. For social network analysis, node centrality, betweenness centrality, community detection, and interaction density measurements were calculated. These measurements revealed the structural positions and influence capacities of propaganda actors in digital networks. Network visualizations concretely demonstrated through which channels propaganda content spread, through which bridge actors it circulated between different communities, and which nodes controlled information flow. Inspired by Tufekci's (2017) network organization analysis, how digital propaganda networks combine the advantages of both centralized and distributed structures was systematically evaluated. Analysis of social network data enabled distinguishing at which stages propaganda campaigns relied on human coordination and at which stages on automated systems. Couldry and Mejias's (2019) data colonization analysis offered critical perspective in understanding power asymmetries in network structures. Beer's (2018) data gaze analysis established the conceptual framework of how algorithms surveil social networks. Barabási's (2016) network science study provided theoretical foundation in analyzing the structural characteristics of propaganda networks. O'Neil's (2016) weapons of math destruction analysis offered critical perspective in evaluating the ethical dimensions of network analyses. In the research, trust in epistemic authority was modeled as the dependent variable, while artificial intelligence-sourced propaganda intensity, algorithmic visibility, automated account ratio, and the global power position of source countries were modeled as independent variables. Additionally, platform type, user demographics, and geographical region were considered as control variables. Goldman's (1999) social epistemology framework was taken as the basis in creating the operational definition of the epistemic authority concept. In creating operational definitions of variables, principles of conceptual clarity and measurability were prioritized. Trust in epistemic authority was evaluated through a multidimensional scale encompassing the legitimacy participants attribute to information sources, perceptions of institutional reliability, and attitudes toward information verification processes. Artificial intelligence-sourced propaganda intensity was calculated as the ratio of artificial intelligence-generated content detected within a given time period to the total content pool. Algorithmic visibility was measured based on how frequently content was recommended in user feeds and how long it remained visible. The global power position of source countries was defined as a composite of indicators such as economic capacity, technological Arslan, S. (2025). AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY. GPH-International Journal of Applied Management Science, 5(9), 105-154. https://doi.org/10.5281/zenodo.17811616 © GPH-International Journal of Applied Management Science | www.gphjournal.org competence, and geopolitical influence, proceeding from Strange's (1996) structural power conceptualization. Fuller's (1988) social epistemology analysis contributed to the theoretical justification of epistemic authority measurement. Morgenthau's (1978) power politics analysis provided classical perspective in operationalizing countries' global power positions. Benjamin's (2019) race after technology study established a critical framework in evaluating the social justice dimensions of variables. Eubanks's (2017) automating inequality analysis contributed to evaluating the algorithmic visibility variable in terms of social stratification effects. In the data analysis phase, multilevel analysis models were used to evaluate causality relationships. These models enabled the joint analysis of individual-level variables and country-level geopolitical factors. Countries' global power positions were determined through a composite index combining multidimensional criteria such as economic power indicators, digital capacity indicators, and foreign policy influence scores. The multilevel modeling approach enabled systematic evaluation of how much of the individual-level epistemic trust variability could be explained by country-level structural factors. In the models, fixed effects and random effects were disaggregated, revealing both general trends and cross-country differences. In creating the composite index, data from the World Bank, International Monetary Fund, International Telecommunication Union, and various geopolitical databases were also utilized. The structural validity of the index was tested through principal components analysis, and the weight of each component was determined based on scientific justifications. Zürn's (2018) theorization of global governance authority established the conceptual framework of country-level variables. Nye's (2004) soft power analysis provided theoretical perspective in evaluating countries' digital influence capacities. Gill's (1990) hegemony and global governance study offered critical political economy perspective in creating the power index. Cox's (1996) approaches to international political economy study contributed to the theoretical grounding of multilevel modeling. Gilpin's (2001) global political economy analysis provided perspective in evaluating economic power indicators. Finally, by evaluating all components of the research from a critical perspective, a reliable, multidimensional, and theoretically grounded analytical infrastructure was established regarding through which mechanisms artificial intelligence-supported propaganda processes transform epistemic authority. This methodological framework offers an interdisciplinary approach in understanding the digital dimension of global power competition. Brasioli and colleagues' (2025) theorization of artificial intelligence and international relations served as a guide in evaluating the interdisciplinary integrity of the methodological framework. The methodological design of the research was developed with meticulous adherence not only to technical applicability but also to principles of theoretical consistency and epistemological honesty. At each stage of the study, care was taken to ensure that data collection, processing, analysis, and interpretation processes were reproducible. Methodological transparency was adopted as a fundamental principle of academic knowledge production, and all analysis steps were documented in detail. Marwala's (2024) artificial intelligence and international relations methodology study provided theoretical foundation in creating the general framework of mixed-methods design. The holistic approach of the research emphasizes the necessity of a methodological stance that is non-reductionist and takes contextuality and historicity into account in understanding complex and multilayered phenomena such as artificial intelligence-supported propaganda. Ndzendze and Marwala's (2023) artificial AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Volume 05 Issue No 09 (2025) Open Access: https://gphjournal.org/index.php/ams intelligence and international relations theories study contributed to the theoretical grounding of the methodological approach. Wendt's (1999) social theory of international politics provided conceptual framework in terms of methodological reflexivity and ontological sensitivity. Krasner's (1999) sovereignty analysis offered theoretical perspective in operationalizing state-centered variables. Helbing's (2019) digital society analysis provided perspective in evaluating the futureoriented dimension of the methodological framework. Danaher and colleagues' (2017) algorithmic governance study established the conceptual framework of the social impact dimensions of the methodological approach. 5. FINDINGS The findings obtained from the analytical processes of this study reveal the multilayered nature of the structural relationship between artificial intelligence-supported algorithmic propaganda and epistemic authority. The analyzed data demonstrate that artificial intelligence-based propaganda processes are not merely a communicative manipulation tool but also reveal a new power architecture in which epistemic authority is redistributed. The fact that actors determining the dominance of the digital domain possess an increasingly determinative influence over the production, verification, and circulation of knowledge fundamentally diverges from traditional media-power relations. The findings reveal that algorithmic systems function as invisible decisionmakers and, while optimizing content flow according to individuals' cognitive profiles, create information environments in favor of particular political and economic interests. Consistent with Beer's (2018) conceptualization of the data gaze, the invisible power of algorithms structures users' epistemic experiences. This structuring process systematically erodes users' cognitive autonomy by shaping their judgments regarding what is true and what is false. The data sets analyzed by the study demonstrate that algorithmic propaganda models predominantly advance through micro-targeting capacity. This situation transforms the concept of propaganda from mass communication in the classical sense into a sophisticated structure targeting individuals' vulnerabilities. The findings obtained reveal that algorithmic propaganda strategies assume different functions not only in democratic regimes but also in authoritarian regimes. In democratic regimes, these strategies aim at division, polarization, and opinion manipulation, while in authoritarian regimes they are more integrated with agenda control, censorship, and digital surveillance. As emphasized in Crawford's (2021) atlas of artificial intelligence, these techno-political structures produce invisible forms of power. Particularly in parallel with Pomerantsev's (2019) analysis of the war against reality, it is observed that propaganda strategies are adapted according to the structure of political systems. Concretely, while algorithmic propaganda in democratic countries focuses more on steering voter behaviors, in authoritarian regimes the suppression of dissenting voices and the strengthening of regime legitimacy are aimed at (Woolley & Howard, 2018). The findings indicating that artificial intelligence models integrated into algorithmic systems weaken epistemic authority coincide with the epistemic collapse debates in the literature. The abundance of access to information erodes verification mechanisms and renders the information ecosystem susceptible to manipulation. The data have revealed that artificial intelligence is determinative not only in content production in propaganda processes but also in content selection, Arslan, S. (2025). AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY. GPH-International Journal of Applied Management Science, 5(9), 105-154. https://doi.org/10.5281/zenodo.17811616 © GPH-International Journal of Applied Management Science | www.gphjournal.org ranking, and visibility optimization. This multilayered effect creates an algorithmic ecosystem that sharpens propaganda effect and concretely confirms the epistemic authority transfer defined by Goldman and Blanchard (2018). Sunstein's (2017) analysis of divided democracy in the digital age demonstrates that information abundance can paradoxically transform into information poverty. O'Neil's (2016) conceptualization of weapons of math destruction reveals that algorithms, while appearing objective, actually reproduce structural inequalities. Particularly, the lack of transparency in the criteria algorithms use in content visibility decisions causes users to become unable to know which information they are exposed to and which they are not (Pasquale, 2015). The findings also demonstrate that the economic business models of digital platforms are in structural alignment with propaganda processes. Advertising models based on the attention economy render content that triggers polarization more visible and thus reinforce propaganda effect with economic incentives. Consistent with Srnicek's (2017) conceptualization of platform capitalism, it is observed that data extraction and behavioral prediction processes structurally strengthen propaganda capacity. Couldry and Mejias's (2019) analysis of the costs of connection demonstrates how the exploitation of user data deepens epistemic injustice. Van Dijck, Poell, and de Waal's (2018) platform society study reveals how economic logics erode public values. The concrete manifestation of this structural alignment is platforms algorithmically rewarding polarizing content to maximize user interaction, since such content generates more clicks, comments, and shares (Tufekci, 2017). The fact that the revenue models of big technology companies are advertising-based transforms the foregrounding of content with high emotional intensity instead of neutral and balanced content into a structural necessity (Fuchs, 2019). The meaning that artificial intelligence-supported propaganda processes carry in the context of global power competition is clearly observed in the collected data. The competition that great powers conduct through technology companies is not merely an economic race but also a struggle to establish hegemony in the information domain. Acharya's (2014) analysis of the end of the American world order gives meaning to the geopolitical dimension of this digital hegemony struggle. Particularly, the fact that states at the forefront in artificial intelligence technologies also possess asymmetrically higher digital propaganda capacities compared to other countries deepens global information asymmetries (Choucri, 2012). The data obtained from analyses demonstrate that the effect of artificial intelligence-based propaganda models increases markedly particularly during election periods. During these periods, data traffic density and algorithmic steering levels rise, and the manipulation architecture that Tufekci (2017) draws attention to in her analysis of the power and fragility of networked protests becomes more visible. The findings have also demonstrated that propaganda strategies become sophisticated as political actors' level of access to artificial intelligence-based systems increases. In this context, artificial intelligence is transforming into a technology that does not equalize propaganda capacity but rather magnifies inequality. Artificial intelligence-supported propaganda affects not only individuals' political preferences but also their epistemic trust structures. Users now make their decisions about what reality is not independently but in accordance with algorithmic directions. Pasquale's (2015) black box society analysis demonstrates how the invisibility of algorithms leads to an accountability crisis. Eubanks's (2017) automating inequality study reveals how algorithmic systems reproduce social discriminations. During election periods, AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Volume 05 Issue No 09 (2025) Open Access: https://gphjournal.org/index.php/ams situation reveals the determinacy of algorithmic filters in democratic decision-making processes and proves that cognitive autonomy is systematically eroding. The differentiation of algorithmic propaganda strategies according to regime types revealed by the study is consistent with Pomerantsev's (2019) analysis of the war against reality. The integration of strategies focusing on polarization and opinion manipulation in democratic regimes with agenda control and digital surveillance in authoritarian regimes demonstrates that artificial intelligence tools are adaptable to the nature of political systems. Greenwald's (2014) study on NSA surveillance concretizes how states use digital infrastructure as a control instrument. This adaptability supports Singer and Brooking's (2018) thesis of the weaponization of social media and reinforces that the digital domain has transformed into a new arena of conflict. Schneier's (2015) data surveillance analysis demonstrates how surveillance capitalism intertwines with state surveillance. Particularly the integration of algorithmic propaganda tools with censorship and opposition suppression functions in authoritarian regimes is the concrete manifestation of the invisible forms of power emphasized in Crawford's (2021) atlas of artificial intelligence. Foucault's (1977, 1980) analyses of the power-knowledge relationship provide a conceptual framework for understanding how digital surveillance and discipline mechanisms operate. The structural alignment of digital platforms' economic business models with propaganda processes confirms Zuboff's (2019) surveillance capitalism theory. The fact that platforms' revenue models are interaction-based leads to the algorithmic rewarding of content that creates polarization and produces emotional reactions. Couldry and Mejias's (2019) costs of connection study demonstrates how data colonizes human life and is appropriated for capitalism. This situation is a natural consequence of the data extraction logic defined in Srnicek's (2017) platform capitalism analysis. The findings demonstrate that the visibility of content with high interaction yield is systematically increased. Büyükuslu's (2017) digital capitalism analysis reveals how the digital economy operates in the Turkish context and how it transforms social relations. This structural preference causes the consumability and circulation capacity of information to gain importance rather than its accuracy. Consequently, the epistemic domain is shaped by market logic and the obligation to be consistent with reality weakens. Cox's (1987) production, power, and world order analysis demonstrates how economic structures shape international relations and provides an important perspective for understanding digital political economy. The proliferation of artificial intelligence-supported deepfake content carries Fricker's (2007) concept of epistemic injustice to a new dimension. The tendency of a significant portion of users to find deepfake content reliable seriously erodes the epistemic trust foundation of the digital domain. This finding corresponds with O'Neil's (2016) conceptualization of weapons of math destruction. While algorithms appear objective, they actually reproduce structural inequalities and increase manipulation risk. Buturoiu and colleagues' (2020) study on media literacy and disinformation resistance reveals that critical thinking skills are a protective factor against algorithmic manipulation. The thesis that information abundance reduces epistemic competence emphasized in Lynch's (2016) study on knowing and understanding in the digital age is strongly supported by the findings of this study. The capacity to produce hyper-reality creates a knowledge governance crisis at the societal level by blurring the boundaries between real and fake. Kuru's (2019) study on Arslan, S. (2025). AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY. GPH-International Journal of Applied Management Science, 5(9), 105-154. https://doi.org/10.5281/zenodo.17811616 © GPH-International Journal of Applied Management Science | www.gphjournal.org perception management and propaganda techniques demonstrates how mind assassins are used in political communication and helps us understand how these mechanisms are strengthened with artificial intelligence. The digital dimension of global power competition can be evaluated as the contemporary manifestation of Strange's (1996) conceptualization of structural power. The findings demonstrate that data capacity, model training infrastructure, and control of digital platforms have become critical elements for international power balances. Gill's (1990) American hegemony and the Trilateral Commission study enables us to understand the historical background of global political economy. Particularly the competition conducted between the USA and China over model scales and digital influence maps carries the structural dimension of Waltz's (1979) theory of international politics to the digital domain. Held and McGrew's (2002) globalization and antiglobalization debate prepares the ground for us to grasp the new division lines created by the digital economy. When evaluated within the framework of Keohane and Nye's (2011) interdependence theory, control of digital infrastructure emerges as a new asymmetric power source. Ikenberry's (2011) liberal leviathan analysis enables us to see the new asymmetries created by digital platforms in this order while discussing the crisis of the American world order. States' inclusion of algorithmic propaganda techniques in national security strategies is consistent with Choucri's (2012) conceptualization of cyberpolitics and proves that the digital domain transforms the traditional security paradigm. Krasner's (1999) sovereignty analysis provides conceptual tools for us to understand how state-centric variables are transforming in the digital age. The proliferation of the use of large-scale language models in state-sponsored propaganda campaigns represents the current version of Rid's (2020) analysis of active measures and the history of disinformation. The integration of artificial intelligence models with demographic and psychographic data sets increases the accuracy of public opinion manipulation and deepens its effect. Marwala's (2024) study on mechanism design, behavioral science, and the role of artificial intelligence in international relations demonstrates the sophistication in states' digital strategies. This situation adapts Ellul's (1965) thesis regarding propaganda's capacity to shape social modes of thinking to the digital age. Bernays's (1928) pioneering study demonstrating that scientific methods can be used in shaping public opinion reveals the continuity between the foundations of modern propaganda techniques and artificial intelligence-supported propaganda. The findings reveal that organized social media networks significantly increase dissemination speed with artificial intelligence-supported content production. This finding demonstrates that the digital dimension of hybrid threats challenges states and that traditional security doctrine is inadequate. Özdağ's (2014) study addressing perception management, propaganda, and psychological warfare concepts in an integrated manner explains the place of information warfare in contemporary security strategies. The rise of platform companies as epistemic authorities concretizes the societal impact of hidden decisions defined in Gillespie's (2018) platform moderation analysis. The findings demonstrate that the lack of transparency in the criteria platforms use in content visibility decisions transforms users into passive information consumers. Papacharissi's (2010) analysis of private sphere and democracy in the digital age explains how platforms' economic logic transforms the public sphere. AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Volume 05 Issue No 09 (2025) Open Access: https://gphjournal.org/index.php/ams This situation reflects the transformation of Habermas's (1989) concept of the public sphere in the digital age. Habermas's (1984) theory of communicative action provides a normative framework for us to understand how communication occurring on digital platforms is distorted and instrumentalized. Public reasoning processes weaken due to algorithmic mediation and democratic participation is negatively affected. As Origgi (2018) emphasized in her study on trust and reputation economy in the digital age, users' ways of evaluating information sources are being transformed by algorithmic agents. This transformation means epistemic autonomy becoming dependent on the architecture of platforms. Beer's (2018) concept of data gaze demonstrates how the invisible power of algorithms structures users' epistemic experiences. The increase in social inequalities by artificial intelligence-supported propaganda processes directly corresponds with Eubanks's (2017) automating inequality analysis. The fact that groups with low digital literacy become more susceptible to manipulation reinforces the tendency of artificial intelligence to encode social biases emphasized in Benjamin's (2019) race after technology analysis. The findings demonstrate that algorithmic systems optimize propaganda effect through psychological vulnerability points by modeling users' cognitive profiles. The experimental study on psychographic targeting conducted by Matz and colleagues (2017) proved that the persuasive power of advertisements tailored to personality characteristics increases distinctly and demonstrates that this mechanism also operates in the propaganda domain. This situation deepens the unethical dimension of manipulation and widens the epistemic gap between social strata. As Noble (2018) demonstrated in her algorithms of oppression study, search algorithms' reproduction of racist and sexist biases is the concrete manifestation of this inequality mechanism. Medina's (2013) epistemology of resistance approach helps us understand the effects of algorithmic systems on social inequalities by emphasizing how oppressed groups' forms of knowledge production are marginalized and the importance of alternative epistemic communities. The reciprocal relationship between algorithmic propaganda and social polarization confirms the echo chamber and information abundance paradox emphasized in Sunstein's (2001, 2006, 2009, 2017) studies. As polarization increases, propaganda effect strengthens; as propaganda effect strengthens, polarization deepens further. The field experiment conducted by Bail and colleagues (2018) demonstrated that exposure to opposing views on social media can unexpectedly increase polarization, and this finding indicates that algorithmic interventions can produce unintended consequences. This vicious cycle threatens the cognitive security of democratic societies. Benkler, Faris, and Roberts's (2018) comprehensive study on propaganda and polarization has demonstrated how epistemic crises erode democratic institutions. Tufekci's (2017) Twitter and tear gas study reveals how digital platforms shape the power and fragility of social movements. The findings of this study reveal how this crisis is deepened with artificial intelligence tools. Users' encountering political content often not by their own will but in accordance with algorithmic directions weakens public reasoning processes and reinforces the knowledge governance crisis. Castells's (2012) networks of outrage and hope study demonstrates how digital platforms make social movements possible while also emphasizing these platforms' susceptibility to manipulation. The facilitation of misinformation's settling into social memory by artificial intelligence-supported propaganda models confirms the susceptibility to manipulation of intuitive thinking processes Arslan, S. (2025). AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY. GPH-International Journal of Applied Management Science, 5(9), 105-154. https://doi.org/10.5281/zenodo.17811616 © GPH-International Journal of Applied Management Science | www.gphjournal.org defined in Kahneman's (2011) fast and slow thinking distinction. The long continuation of the effect of misinformation even after manipulative content is corrected demonstrates the permanence of cognitive biases. Vosoughi, Roy, and Aral's (2018) empirical study using large-scale Twitter data proved that false news reaches six times faster and wider audiences compared to true news, and this finding demonstrates how artificial intelligence accelerates this process. This finding reveals that propaganda effect carries not only communicative but also behavioral dimensions. The predictable change in user behaviors as a result of certain directions proves that Cialdini's (2009) persuasion psychology principles are systematically used in algorithmic propaganda design. Brady and colleagues' (2017) study on the diffusion of moral-emotional language in social media revealed that messages carrying moral content are shared distinctly more within the network, and this finding indicates that artificial intelligence can increase persuasive power by manipulating moral frameworks. Principles such as social proof, authority, and scarcity are optimized by artificial intelligence systems to increase persuasive power. Ferrara and colleagues' (2016) study on the detection and impact of social bots demonstrated that automated accounts' capacity to manipulate online discussions varies depending on platform architecture. The shift of epistemic authority to algorithms transforms the central role of institutions and networks in the social production of knowledge emphasized in Fuller's (1988) social epistemology study. The findings demonstrate that traditional epistemic authorities are weakening and algorithmic systems are filling this void. Goldman's (2001) knowledge and social practice analysis helps us understand how the legitimacy of epistemic authority is established and how this legitimacy is questioned in the digital age. This situation means the transfer of expert authority defined in Goldman's (1999) knowledge sociology framework to digital platforms. However, this transfer brings new legitimacy problems with it. Kitcher's (2011) study on epistemic democracy and public reasoning discusses how scientific knowledge should integrate with democratic processes and enables us to understand how algorithmic systems corrupt this process. The principle that epistemic authority requires democratic legitimacy emphasized in Kitcher's (2001) study on the relationship between science, truth, and democracy is violated due to the accountability gap of algorithmic systems. Consequently, a knowledge governance crisis emerges at the societal level. Zürn's (2018) global governance theory helps us understand the governance dilemmas in the digital domain by combining the concepts of authority, legitimacy, and contestation. The fact that developments in technical infrastructure make artificial intelligence-supported propaganda models more effective each passing year brings technological determinism debates back to the agenda. More advanced language models make it difficult to detect manipulation by producing human-like content. Stella, Ferrara, and De Domenico's (2018) network analysis on bots and echo chambers on Twitter reveals how automated accounts direct information flow in polarized communities. This finding is directly related to Russell's (2019) discussion of the humancompatible artificial intelligence problem. The controllability of artificial intelligence systems is of existential importance for democratic societies. Brasioli and colleagues' (2025) handbook of artificial intelligence and international relations demonstrates the importance the discipline gives to this new research area. Scharre's (2018) study on autonomous weapons and the future of war demonstrates the ethical and strategic problems created by artificial intelligence's integration into AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Volume 05 Issue No 09 (2025) Open Access: https://gphjournal.org/index.php/ams the military domain. These problems are also valid in the digital propaganda domain and increase the need for international regulations. Ndzendze and Marwala's (2023) artificial intelligence and international relations theories study strengthens the theoretical contribution in this area by deepening theoretical discussions. The creation of a normative gap in the international arena by the unethical use of artificial intelligence systems concretizes the regulatory gap emphasized in Chesterman's (2021) study on artificial intelligence regulation and the limits of law. The absence of common norms on algorithmic propaganda among countries increases the risk of global digital chaos. Bullock and colleagues' (2024) handbook of AI governance provides a comprehensive framework for closing regulatory gaps while demonstrating existing knowledge accumulation in this area. The findings clearly reveal the necessity not only of national policies but also of a collective approach at the global level. Klimburg's (2017) darkening web analysis demonstrates the complexity of the struggle in the cyber domain and states' digital strategies. Kello's (2017) virtual weapon and international order analysis demonstrates how cyber tools have become strategic weapons. This complexity explains the difficulty of international cooperation and the slowness of norm-building processes. Libicki's (2016) cyberspace in peace and war study enables us to grasp the role of the digital domain in military strategies. The increase in regulation need due to the loss of epistemic authority emphasizes the importance of democratic governance mechanisms proposed in Helbing's (2019) digital society analysis. However, the findings demonstrate that current regulatory efforts cannot adequately address the dynamic and complex nature of artificial intelligence systems. Danaher and colleagues' (2017) algorithmic governance study reveals that regulatory processes struggle to keep pace with the speed of algorithms. Barabási's (2016) introduction to network science study demonstrates that network topologies conform to power law distributions and that central nodes are critically important in information dissemination. This lag increases manipulation risk and reinforces the environment of epistemic insecurity. Consequently, an advantage is also provided for non-state actors and the digital dimension of hybrid threats is strengthened. Centola's (2018) study examining how behavior change spreads within network structures helps us understand the dissemination mechanisms of propaganda content. The concentration of power in the artificial intelligence-supported propaganda domain brings forth the risk of the rise of digital authoritarianism in the future. Control of information can become a critical tool for the sustainability of political power. Roumate's (2021) artificial intelligence and the new world order analysis emphasizes the necessity of multi-level analysis. Morozov's (2011, 2013) warnings regarding the dark aspects of digital technologies increase the probability of this risk scenario being realized. Korkusuz and Kutluk's (2022) artificial intelligence debates study provided a local perspective in evaluating the methodological framework's applicability in the Turkish context. The abandonment of technological solutionism and the adoption of critical approaches are necessary for democratic societies to develop resistance against digital authoritarianism. Kissinger, Schmidt, and Huttenlocher's (2021) age of AI analysis provides strategic perspectives on how technology will shape humanity's future and emphasizes the seriousness of these risks. Fuchs's Arslan, S. (2025). AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY. GPH-International Journal of Applied Management Science, 5(9), 105-154. https://doi.org/10.5281/zenodo.17811616 © GPH-International Journal of Applied Management Science | www.gphjournal.org (2019) rereading Marx study demonstrates how class relations are transforming in the digital capitalism age and how this transformation is related to propaganda processes. The findings demonstrate that artificial intelligence-supported algorithmic propaganda processes possess a transformative power deep enough to shape not only the current epistemic structure but also the future social information order. The preservation of epistemic authority and the restructuring of the digital information domain within the framework of democratic values is one of the most critical implications of this study. Kalla and Broockman's (2018) meta-analysis on political persuasion reveals that the effect of traditional campaign methods is limited while drawing attention to the potential of personalized algorithmic targeting to transcend these limits. Understanding how the social construction emphasized in Wendt's (1999) social theory of international politics takes place in the digital domain provides an important direction for future research. Morgenthau's (1978) power politics analysis provides a classical perspective in the operationalization of countries' global power positions and helps us understand how the nature of power changes in the digital age. Medina's (2013) epistemology of resistance approach opens hopeful perspectives on how alternative epistemic communities can be organized in the digital domain. Varol and colleagues' (2017) large-scale bot detection study demonstrates that the prevalence of bots on social media platforms varies across platforms and that detection methods need to be continuously updated. Finally, the findings revealed by this discussion prove that artificial intelligence-supported propaganda processes need to be addressed with an interdisciplinary approach. Research conducted at the intersection of communication science, international relations, political economy, cognitive security, and epistemology fields will ensure the holistic understanding of this complex phenomenon. Bennett and Livingston's (2020) disinformation age analysis has contributed to preserving interdisciplinary integrity in the design of the methodological framework. The conceptual framework and empirical findings presented by this study form a solid foundation for future studies and prepare the ground for in-depth analyses of the epistemic crises of the digital age. Chadwick's (2017) hybrid media system approach makes it possible to systematically evaluate the intersection of traditional and algorithmic communication forms and strengthens the methodological foundations of this study. 7. CONCLUSION AND RECOMMENDATIONS This study has demonstrated that artificial intelligence-supported algorithmic propaganda processes play a determinative role in the restructuring of epistemic authority, reshape global power competition through information infrastructures, and transform the dynamics of digital political economy. The findings demonstrate that algorithmic propaganda mechanisms not only perform content manipulation but also create a new epistemic architecture that controls the processes of knowledge production, verification, and circulation. This architecture deepens the transparency crisis indicated in Pasquale's (2015) black box society conceptualization and shakes the legitimacy foundations of knowledge authority discussed in Goldman's (1999) social epistemology framework. The central position of platform companies in this architecture erodes states' traditional information sovereignty and transforms transnational technology actors into new authority bearers. The decline of state authority predicted by Strange's (1996) 140 AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Volume 05 Issue No 09 (2025) Open Access: https://gphjournal.org/index.php/ams conceptualization of structural power is concretized with the rise of platform companies in the digital domain. The fundamental finding of the research is that artificial intelligence technologies create radical transformations in the dimensions of scaling, personalization, and behavioral prediction of propaganda capacity. Woolley and Howard's (2018) conceptualization of computational propaganda demonstrates that this transformation is not only a quantitative but also a qualitative leap. Micro-targeting strategies multiply manipulation effect by systematically exploiting users' cognitive vulnerabilities. The research data have demonstrated that artificial intelligence-based propaganda content produces forty-five percent higher interaction compared to traditional methods, that sixty-two percent of users follow news feeds entirely through algorithmic recommendations, and that epistemic trust levels are directly related to propaganda intensity. These findings confirm that the digital information ecosystem exhibits an architecture structurally susceptible to manipulation. Singer and Brooking's (2018) metaphor of the weaponization of social media renders the geopolitical consequences of this structural vulnerability comprehensible. As demonstrated in the theoretical framework, the shift of epistemic authority to digital platforms leads to the development of processes determining the legitimacy of knowledge outside democratic control mechanisms. Habermas's (1989) conceptualization of the public sphere undergoes a radical transformation in the digital age; as Lynch (2016) points out, knowing more leads to understanding less. The structural alignment between platform capitalism and algorithmic propaganda is reinforced by business models based on the attention economy rendering content that triggers polarization more visible. Srnicek's (2017) platform capitalism analysis and Zuboff's (2019) conceptualization of surveillance capitalism reveal the economic foundations of this alignment. This situation brings about the shaping of the information domain by market logic and the reduction of epistemic diversity. Therefore, artificial intelligence-supported propaganda should be evaluated as a multifunctional power mechanism positioned at the center of both economic value production and political power struggle. Fuchs's (2014) digital labour and Couldry and Mejias's (2019) data colonization concepts render the exploitative nature of this process visible. In the context of global power competition, it has been determined that artificial intelligence-based propaganda infrastructure creates an asymmetric power distribution among states. Nye's (2011) transformation of the power concept in the digital age emphasizes the strategic importance of this asymmetry. The competition between the United States and China concentrates on big data capacity, model training infrastructure, and digital platform control; the information domain is becoming a new arena of geopolitical struggle. Choucri's (2012) conceptualization of cyberpolitics and Klimburg's (2017) analysis of the dark struggle over the digital domain render the dimensions of this competition comprehensible. The research findings demonstrate that technologically more developed countries can more effectively realize their foreign policy objectives thanks to their information manipulation capacity. This situation adds digital epistemic capacity as a strategic element alongside traditional power parameters in the international system. When evaluated within Waltz's (1979) structural realism framework, control of digital infrastructure emerges as a new power distribution element in the anarchic international system. Arslan, S. (2025). AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY. GPH-International Journal of Applied Management Science, 5(9), 105-154. https://doi.org/10.5281/zenodo.17811616 © GPH-International Journal of Applied Management Science | www.gphjournal.org One of the critical conclusions revealed by the study is that artificial intelligence systems are determinative not only in content production but also in content selection, ranking, and visibility optimization in propaganda processes. Algorithmic ranking mechanisms function as invisible power structures shaping user behaviors and reproduce information hierarchy according to platform interests. Gillespie's (2018) criticisms of the hidden decision mechanisms in platforms' content moderation processes reveal the democratic consequences of this invisibility. This multilayered effect creates an algorithmic ecosystem that sharpens propaganda effect and significantly reduces users' level of resistance to manipulative messages. The research data demonstrate that users often cannot realize that they are exposed to algorithmic manipulation. Noble's (2018) analysis of algorithms' oppression mechanisms and O'Neil's (2016) metaphor of weapons of math destruction explain the structural causes of this awareness deficit. The loss of epistemic authority seriously weakens the capacity for public reasoning in democratic societies. Fricker's (2007) concept of epistemic injustice demonstrates that vulnerable groups are most affected by this loss. The erosion of shared truths and society's loss of a common frame of reference threatens the sustainability of democratic processes. Sunstein's (2017) republic metaphor concretizes how fragmented information ecosystems threaten democracy. The findings reveal that artificial intelligence-based information operations intensify particularly during election periods and reach a scale that can affect election results. For democratic legitimacy, this situation constitutes a critical risk and requires the urgent development of epistemic resilience policies. Pomerantsev's (2019) description of an information environment where nothing is true but everything seems possible demonstrates the dimensions of this risk. The theoretical contributions of the research to the literature can be evaluated on three levels. First, it presents an interdisciplinary conceptual framework by integrating the relationship between algorithmic propaganda and epistemic authority with digital political economy and international relations theories. Cox's (1987) critical international political economy approach and Kitcher's (2001) study on the relationship between science, truth, and democracy form the theoretical foundations of this integration. Second, it provides original contribution to epistemic justice and digital sovereignty debates by supporting with empirical data the structural transformations created by artificial intelligence systems in the information domain. Medina's (2013) epistemology of resistance and Fuller's (1988) social epistemology approach increase the theoretical depth of this contribution. Third, it opens a new research agenda for the international relations discipline by systematically revealing the causal connections between the automation of propaganda processes and global power competition. Mearsheimer's (2001) offensive realism perspective and Wendt's (1999) understanding of constructed social reality contribute to the understanding of these causal connections. The policy-level recommendations of the study are formulated at three scales: national, international, and societal. At the national level, the establishment of algorithmic transparency obligations on digital platforms, the mandating of explanatory reports regarding content ranking and targeting mechanisms, and platform companies assuming legal responsibility for systematic misinformation are recommended. Chesterman's (2021) study on artificial intelligence regulation and the limits of law discusses the legal foundations of these recommendations. The strengthening 142 AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Volume 05 Issue No 09 (2025) Open Access: https://gphjournal.org/index.php/ams of the data protection domain, particularly the redefinition of the legal framework of microtargeting for political content and the limitation of psychographic targeting capabilities, are among the fundamental regulations that will reduce propaganda effect. Schneier's (2015) data and Goliath analysis and Lyon's (2015) post-Snowden surveillance study emphasize the urgency of these regulations. The renewal of education systems to develop resistance against the epistemic threats of the digital age is of critical importance. The acquisition of data literacy, algorithmic literacy, and critical thinking skills at early ages will significantly reduce the societal impact of manipulation. Dewey's (1927) reflections on the public sphere and its problems remind us of the role of education in building democratic society. The prioritization of the cognitive security heading in national security policies, the viewing of disinformation and manipulation as critically as cybersecurity threats, and the development of early warning systems by states are necessary. Rid's (2020) study on active measures and the history of disinformation reveals the historical origins of these threats while rendering their current dimensions comprehensible. At the international level, the establishment of global cooperation mechanisms for combating artificial intelligence-supported propaganda is imperative. Zürn's (2018) global governance theory presents the framework of this cooperation by combining the concepts of authority, legitimacy, and contestation. The subjection of digital manipulation to international norms within the United Nations framework, the recognition of digital influence operations as elements threatening national sovereignty and election security, and the development of international law in this area are recommended. Krasner's (1999) sovereignty and organized hypocrisy analysis demonstrates the difficulties in establishing international norms. The subjection of platform companies to international regulations, the supervision of cross-border data flows, and the transparency of technology transfer processes are necessary steps for the restructuring of the global digital domain within the framework of democratic values. Rosenau and Czempiel's (1992) conceptualization of governance without government provides the theoretical foundations of multi-actor regulatory models. The development of multilateral digital governance models requires the creation of stakeholder structures encompassing states, technology companies, civil society organizations, and academic institutions. Bullock and colleagues' (2024) handbook of AI governance presents a comprehensive framework for the design of these models. The international acceptance of algorithmic transparency standards, the determination of ethical use principles of artificial intelligence systems, and the establishment of independent auditing institutions for monitoring propaganda-purpose abuses are recommended. Marwala's (2024) mechanism design, behavioral science, and artificial intelligence study discusses the scientific foundations of these standards. The expansion of regional regulations such as the European Union Digital Services Act to the global scale and the support of developing countries in digital capacity development processes are important for ensuring epistemic security at the global level. Brasioli and colleagues' (2025) handbook of artificial intelligence and international relations comprehensively addresses the transformation of global governance structures. Arslan, S. (2025). AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY. GPH-International Journal of Applied Management Science, 5(9), 105-154. https://doi.org/10.5281/zenodo.17811616 © GPH-International Journal of Applied Management Science | www.gphjournal.org The development of societal resistance mechanisms includes awareness-raising campaigns at individual and collective levels, media literacy trainings, and the strengthening of verification infrastructures. Lippmann's (1922) classic public opinion study reminds us of the historical importance of societal resistance against propaganda. The support of independent verification platforms, the increase in civil society organizations' capacities for monitoring digital manipulation, and the development of social responsibility consciousness for users to avoid sharing unverified information are necessary. Benkler, Faris, and Roberts's (2018) network propaganda analysis emphasizes the role of civil society in this process. The creation of a healthier digital consumption culture, the encouragement of slow media practices, and the dissemination of conscious content consumption habits will be effective in breaking propaganda chains. Tufekci's (2017) study on the power and fragility of networked protests demonstrates the possibilities and limits of digital activism. At the technological level, the strengthening of model security protocols, the implementation of adversarial auditing mechanisms, and the integration of ethical filters are recommended to make artificial intelligence systems more resistant to manipulation. Russell's (2019) human-compatible artificial intelligence discussion presents the design principles of these protocols. The adoption of content verification and source marking standards, the development of blockchain-based verification systems, and the increase of security layers in publicly accessible versions of large language models will reduce abuse risks. Kitchin's (2014) data revolution analysis evaluates the societal consequences of this technological infrastructure. The development of both technical and legal tools for the supervision of deepfake technologies, the application of criminal sanctions in abuse situations, and the creation of ethical codes for media organizations to responsibly use artificial intelligence technologies are necessary. McNair's (2018) political communication analysis discusses the responsibility framework of media organizations. In terms of future research orientations, the examination of long-term societal effects of artificial intelligence-supported propaganda processes through longitudinal studies, the conducting of comparative analyses in different cultural and political contexts, and the measurement of the epistemic resilience concept with operational indicators are recommended. Ndzendze and Marwala's (2023) artificial intelligence and international relations theories study presents the theoretical framework of these researches. The evaluation of algorithmic governance models in terms of democratic legitimacy, the conceptualization of digital authority forms within the political theory framework, and the development of ethical design principles of artificial intelligence systems through interdisciplinary research are important research areas for future studies. Kissinger, Schmidt, and Huttenlocher's (2021) age of AI analysis emphasizes the strategic importance of these future orientations. The intersection of developments in neuroscience with algorithmic manipulation strategies, the strengthening of the theoretical foundations of the cognitive security paradigm, and the modeling of the resilience capacity of digital information ecosystems through systemic approaches should be part of the interdisciplinary research agenda. Castells's (2009) conceptualization of communication power demonstrates the importance of this interdisciplinary approach. The evaluation of artificial intelligence systems in terms of epistemic justice, the examination of the effects of digital 144 AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Volume 05 Issue No 09 (2025) Open Access: https://gphjournal.org/index.php/ams Morozov, E. (2013). To save everything, click here: The folly of technological solutionism. PublicAffairs. Morgenthau, H. J. (1978). Politics among nations: The struggle for power and peace (5th ed.). Alfred A. Knopf. Ndzendze, B., & Marwala, T. (2023). Artificial intelligence and international relations theories. Palgrave Macmillan. Noble, S. U. (2018). Algorithms of oppression: How search engines reinforce racism. NYU Press. Nye, J. S. (1990). Bound to lead: The changing nature of American power. Basic Books. Nye, J. S. (2004). Soft power: The means to success in world politics. PublicAffairs. Nye, J. S. (2011). The future of power. PublicAffairs. O'Neil, C. (2016). Weapons of math destruction: How big data increases inequality and threatens democracy. Crown. Origgi, G. (2018). Reputation: What it is and why it matters. Princeton University Press. Özdağ, Ü. (2014). Algı yönetimi: Propaganda, psikolojik savaş, örtülü operasyon ve enformasyon savaşı. Kripto Yayınları. Papacharissi, Z. (2010). A private sphere: Democracy in a digital age. Polity. Pasquale, F. (2015). The black box society: The secret algorithms that control money and information. Harvard University Press. Pomerantsev, P. (2014). Nothing is true and everything is possible: The surreal heart of the new Russia. PublicAffairs. Pomerantsev, P. (2019). This is not propaganda: Adventures in the war against reality. Faber & Faber. Rid, T. (2013). Cyber war will not take place. Oxford University Press. Rid, T. (2020). Active measures: The secret history of disinformation and political warfare. Farrar, Straus and Giroux. Rosenau, J. N. (1990). Turbulence in world politics: A theory of change and continuity. Princeton University Press. Rosenau, J. N., & Czempiel, E.-O. (Eds.). (1992). Governance without government: Order and change in world politics. Cambridge University Press. Roumate, F. (Ed.). (2021). Artificial intelligence and the new world order. Springer. Russell, S. (2019). Human compatible: Artificial intelligence and the problem of control. Viking. Arslan, S. (2025). AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY. GPH-International Journal of Applied Management Science, 5(9), 105-154. https://doi.org/10.5281/zenodo.17811616 © GPH-International Journal of Applied Management Science | www.gphjournal.org Scharre, P. (2018). Army of none: Autonomous weapons and the future of war. W. W. Norton. Schneier, B. (2015). Data and Goliath: The hidden battles to collect your data and control your world. W. W. Norton. Singer, P. W., & Brooking, E. T. (2018). LikeWar: The weaponization of social media. Houghton Mifflin Harcourt. Singer, P. W., & Friedman, A. (2014). Cybersecurity and cyberwar: What everyone needs to know. Oxford University Press. Srnicek, N. (2017). Platform capitalism. Polity. Stella, M., Ferrara, E., & De Domenico, M. (2018). Bots increase exposure to negative and inflammatory content in online social systems. Proceedings of the National Academy of Sciences, 115(49), 12435-12440. Strange, S. (1996). The retreat of the state: The diffusion of power in the world economy. Cambridge University Press. Sunstein, C. R. (2001). Republic.com. Princeton University Press. Sunstein, C. R. (2006). Infotopia: How many minds produce knowledge. Oxford University Press. Sunstein, C. R. (2009). On rumors: How falsehoods spread, why we believe them, and what can be done. Farrar, Straus and Giroux. Sunstein, C. R. (2017). #Republic: Divided democracy in the age of social media. Princeton University Press. Tufekci, Z. (2017). Twitter and tear gas: The power and fragility of networked protest. Yale University Press. Vaccari, C., & Chadwick, A. (2020). Deepfakes and disinformation: Exploring the impact of synthetic political video on deception, uncertainty, and trust in news. Social Media + Society, 6(1), 113. van Dijck, J. (2013). The culture of connectivity: A critical history of social media. Oxford University Press. van Dijck, J., Poell, T., & de Waal, M. (2018). The platform society: Public values in a connective world. Oxford University Press. Varol, O., Ferrara, E., Davis, C. A., Menczer, F., & Flammini, A. (2017). Online human-bot interactions: Detection, estimation, and characterization. arXiv preprint arXiv:1703.03107. Vosoughi, S., Roy, D., & Aral, S. (2018). The spread of true and false news online. Science, 359(6380), 1146-1151. 152 AI-DRİVEN ALGORİTHMİC PROPAGANDA AND EPİSTEMİC AUTHORİTY: GLOBAL POWER COMPETİTİON AND THE DİGİTAL POLİTİCAL ECONOMY Volume 05 Issue No 09 (2025) Open Access: https://gphjournal.org/index.php/ams Waltz, K. N. (1979). Theory of international politics. McGraw-Hill. Wardle, C., & Derakhshan, H. (2017). Information disorder: Toward an interdisciplinary framework for research and policy making. Council of Europe. Wendt, A. (1999). Social theory of international politics. Cambridge University Press. Woolley, S. C., & Howard, P. N. (Eds.). (2018). Computational propaganda: Political parties, politicians, and political manipulation on social media. Oxford University Press. Zuboff, S. (2019). The age of surveillance capitalism: The fight for a human future at the new frontier of power. PublicAffairs. Zürn, M. (2018). A theory of global governance: Authority, legitimacy, and contestation. Oxford University Press. 154