The global political power and impact of social media are beyond dispute. Online content amplified by platforms can influence elections, mobilize protest and activist movements, imperil diplomatic relations, intensify societal and global polarization, empower disenfranchised communities, and inflame violent conflict. As International Relations (IR) scholarship increasingly comes to terms theoretically and empirically with the nature and implications of this power, we argue that this research agenda would be further enriched by examining the people and organizations at its heart: social media firms (SMFs) and their executives and employees across the globe.
Although algorithms and automation may, in many ways, undergird how social media content interacts with politics worldwide, these processes reflect the internal cultures, norms, sociologies, hierarchies, political economies, decisions, and external relationships of their different corporate owners and bodies. Moreover, some of the most politically consequential content moderation choices emanate from intra-organizational debate and contestation in specific temporal and spatial contexts or from the actions of a small number of wealthy and powerful tech officials who are rarely the object of IR research. Elon Musk’s 2022 acquisition of Twitter and subsequent changes to what is now “X”Footnote 1 represent perhaps the starkest example of how social media CEOs can reverse or unpick seemingly established decisions, policies, and norms, with significant implications for global politics. SMFs are more, however, than just their leaders. Different teams within each firm have diverse and sometimes competing agendas, priorities, and interlocutors, which can affect key decisions and influence how that firm’s agency and global impact are calibrated and experienced.
Recent agenda-setting IR scholarship on platforms/platform governance, Big Tech, and algorithmic governance has largely examined SMFs and their agency in the aggregate. This work provides critical insight into the collective and global power of Big Tech but tells us less about how and why different SMFs act in the ways they do in different spaces and at different times. We advocate an ultimately complementary lens through which to examine social media’s influence on global politics, drawing on recent work in international political economy and organization studies. Understanding SMFs as actors in global politics, we suggest, requires us to analyze them as complex, sometimes huge, organizations of teams and individuals with multiple cultures, institutional logics, and relationships. At the level of the organization itself, they merit study as distinctive political units existing in space and time. At the intra-organizational level, they should be appreciated as institutional environments that contain multiple power centers, perspectives, and expressions of agency, including ones where even the CEO’s authority can at times be disrupted.
Analyzing SMFs in this manner helps us better understand the contingent, organizational, and sociological drivers of social media’s entanglement with global politics and, indeed, the multiple ways in which SMFs’ influence is magnified or constrained in different contexts. Social media’s interaction with global politics, we underscore, is heavily informed by the norms, debates, and decisions that take place within SMFs, by their US-centric context and networks,Footnote 2 and by the top-down and often reactive character of SMF policy making, particularly on content moderation. Partly because of these factors, the character and enactment of SMF agency in much of the Majority World can vary quite considerably from that in North America and Western Europe. These differences often reinforce or leverage existing global inequalities, but they can also provide opportunities for some marginalized actors.
This reflection also sheds critical light on the distinctive and deeply ambiguous character of SMFs within the international system. As private companies, SMFs’ “product” is effectively the curation of the public sphere, and they ultimately straddle and mediate the relationships of and between states, politicians, parties, and societies (Srivastava Reference Srivastava2023). They complicate, in many ways, conventional social-science conceptual boundaries (Atal Reference Atal2021), at times arrogating for themselves the role of public goods guardian and at other times identifying as simply private enterprises providing infrastructure, akin to a telephone company (Wintour Reference Wintour2020). This blurring of the lines is also evident in the comingling of SMF corporate power and the semi-autonomous politicking of its CEOs. Our study, in this regard, also contributes to emerging research into the role and significance of billionaires in global politics (Hägel Reference Hägel2020; Krcmaric, Nelson, and Roberts Reference Krcmaric, Nelson and Roberts2023; Youde Reference Youde2019).
Drawing on four years of research engagement with Meta and interviews with current and former Meta and Twitter staff, our study proceeds as follows. After a discussion of methods, we examine existing research on social media’s entanglement with global politics, situating it within the broader scholarship on Big Tech and IR. In doing so, we contend that this literature could go further in specifying and delineating SMFs’ global influence by analyzing them in their specific organizational contexts. The second half of the study introduces and demonstrates the value of this approach for IR scholars, concentrating first at the organizational level and then at the intra-organizational level. Focusing on the politics and processes surrounding content moderation, and drawing on perspectives from a range of actors and polities, we underline not only how the cultures, sociologies, hierarchies, political economies, and external relationships of different SMFs can help determine the nature of their involvement in and impact on global politics but also how and why the nature of this engagement—and the space for agency for those involved—can vary so markedly across polities. We conclude by reflecting on the wider conceptual and empirical implications of this analysis for IR researchers.
Methods and Data
The empirical basis of this study is an ongoing engagement since 2019 with Meta, current owner of three of the five largest social media platformsFootnote 3 based on monthly active users (MAUs): Facebook (1), WhatsApp (3=), and Instagram (3=; Statista 2024). We conducted two Meta-funded research projects between 2018 and 2023, one on the use of WhatsApp during Nigeria’s 2019 elections and one on internet shutdowns in Africa. The research included eight meetings in the United States, the United Kingdom, and online with about 25 Meta staff. We do not draw explicitly on these encounters in this study; they, however, informed our thinking on Meta, and insights taken from them give us further confidence in the strength of our arguments.
Instead, the principal interview data we draw on falls into two categories. First, we conducted interviews in the United States, the United Kingdom, Kenya, Nigeria, and online in 2022 and 2023 with 18 current or former Meta and Twitter/X employees. They were current or former members of teams involved in public policy, content moderation,Footnote 4 global affairs, or product development and included those who have worked for Instagram, WhatsApp, and Facebook itself. We also conducted interviews with 15 current or former senior government officials in Nigeria (Abuja, 2022–2023) and Uganda (Kampala, 2023), all of whom were involved in their governments’ 2021 decision to ban Twitter (Nigeria) or Facebook (Uganda) nationally. To recruit the study subjects, participants were approached on the basis of their (current or former) official position and remit within government (in the cases of Nigeria and Uganda) or Meta/Twitter, with further recruitment undertaken through snowball sampling.
Both sets of respondents had been interviewed as part of the internet shutdown project described earlier. That project had explored the drivers and ecosystems of internet restrictions in Africa through assessing the role and influence of actors, including states and SMFs themselves, on the moderation, regulation, and criminalization of online speech and fora.Footnote 5
Interviews were semi-structured, with time built in to enable each respondent to speak to their own experiences, perspectives, and place within this ecosystem (Fujii Reference Fujii2018). Each respondent was asked to reflect on their views on—and experiences of—online content moderation norms and practices and the role, actions, and responsibilities of states in these spaces. These two themes—drawn deductively from an analysis of relevant literature—then structured the thematic coding analysis of interviews undertaken after we transcribed them.Footnote 6 This enabled us to identify the series of subthemes explored in the empirical analysis.
We complemented the interview data with analysis of SMFs’ public statements and documents; publicly available media or other interviews with SMF staff; and autobiographies or exposés published by former SMF staff, close associates, or investigative journalists (Bergen Reference Bergen2022; Haugen Reference Haugen2023; McNamee Reference McNamee2019; Stokel-Walker Reference Stokel-Walker2023), and internal SMF documents made publicly available by whistleblowers. We also drew on media sources and relevant scholarship. Doing so enabled us to consider a wider range of major SMFs, although we did not examine SMFs headquartered in China because of the distinctively restrictive digital environment in which they operate (Bradford Reference Bradford2023, 69–104; Hern Reference Hern2019). We believe that these combined data offer an important basis for analysis given Meta’s dominance of the market outside China and the popular use of X/Twitter (the twelfth-largest social media platform by monthly active users (MAU) globally and ninth outside China) by political actors to share content and shape debate (Barbera and Zeitzoff Reference Barbera and Zeitzoff2018; Bestvater and Loyle Reference Bestvater and Cyanne E2023; Robertson Reference Robertson2023; Statista 2024).
Our examination of the perspectives of Africa-based and focused actors allows us to consider how SMFs’ political influence is felt, negotiated, and calibrated in the Majority World, filling an important empirical gap in the literature.Footnote 7 In doing so, we do not, of course, claim that the Nigerian and Ugandan examples are representative of a single African or Majority World experience. Equally, we are not suggesting that this analysis should be of interest only to those focused on the politics of social media beyond North America and Europe. If anything, our findings bolster our broader conceptual representation of SMFs as organizations of people anchored in and conditioned by specific spatial, temporal, political, and normative contexts (cf. Gagliardone et al. Reference Gagliardone, Diepeveen, Findlay, Olaniran, Matti and Tallam2021, 2).
All respondents were informed in advance that the research formed part of a Meta-funded project but were reassured that the project’s research design, findings, and future publications were independent of Meta or any other SMF. Moreover, as was also explained, the funding was provided to the first author’s institution with the explicit commitment that “Facebook (/WhatsApp) will not have any influence in the independent conduct [of the] research, or in the dissemination of your findings.”Footnote 8 The research team members were not required to nor did they provide any research data to Meta. Relatedly, Meta/WhatsApp did not make any nonpublicly available data available to the researchers; nor did they seek to influence the findings. Indeed, Meta imposed virtually no reporting requirements on the funding.Footnote 9
The Meta funding call sought research applications from “global social science researchers” for projects focused on “challenges related to misinformation, polarization, information quality, and social conflict on social media” with the aim of “contribut[ing] to a shared understanding…on how social technology companies can better address social issues on their platforms” (Meta 2021). The call included a wide range of priority research areas (from climate change to conflict), placed limited restrictions on methods and research design beyond “inclusion…of non-Western regions, researchers, or collaborations,” and clarified that “research is not restricted to focusing on Facebook Inc apps and technologies” (Meta 2021). Given all the above, we are confident that this source of funding did not substantively shape the project’s research questions, methodology, implementation, or the interpretation of its findings.
We accept, however, that the funding source may have affected our access to respondents. This influence was not, however, felt in a unidirectional manner. On the one hand, the fact that the project was funded by Meta provided us with a degree of initial contact with the firm’s research and public policy teams that we would likely not have had otherwise. Some of these officials later agreed to be interviewed, and some may have felt a degree of corporate responsibility to engage with the researchers. Yet, on the other hand, around 20 current or former SMF staff or associates approached for an interview declined to respond or refused to participate. They included current and former executives at both Meta and Twitter. Ultimately, the reasons why some of those approached refused to participate or to respond are a matter of conjecture; those who did explain their decision did so in brief, legalistic terms.Footnote 10 It is important to acknowledge in this regard, however, that as private companies SMFs do not face the same cultural, normative, and professional pressures to engage with researchers as those in the public sector. Current and former SMF staff also potentially face legal consequences from disclosing corporate information to outside parties, and researchers must sign a form of confidentiality agreement before being allowed into company buildings.Footnote 11
Indeed, SMF CEOs have been unambiguous, sometimes menacingly so, about the negative consequences for staff of sharing “internal information” with external actors (Woodward Reference Woodward2022; Zuckerberg Reference Zuckerberg2010). This may also have contributed to serving SMF executives’ reluctance to engage with the project. It also suggests that current SMF staff we did interview were more likely to present their organization in positive or “approved” terms. The need to navigate such dynamics is not, however, unique to scholars of SMFs or even of corporations more broadly but will be familiar to scholars of authoritarianism and, indeed, of closed contexts more generally (Allam Reference Allam2024; Koch et al. Reference Koch2013; Stroup and Goode Reference Stroup and Goode2023). Our approach to these challenges, as with researchers across these areas, has been to engage as many potential respondents as possible and to triangulate as many diverse data sources as possible with these limitations and potential biases in mind. Given the risks for some respondents, we refer to all SMF interviewees as “current/former Twitter/Facebook/WhatsApp/Meta staffers.” To indicate interview location, we state either the country name or “virtual.”
In summary, we took a clear and consistent approach to managing our research interactions, given the sensitivities of the research and funding source. This entailed (1) being clear throughout about the topic under study (SMFs and questions of content moderation or platform shutdown); (2) agreeing that in future publications we would use citation details that do not identify interviewees where requested; (3) deciding that we would not refer in future publications to off-the-record engagement with SMF personnel; (4) and, as noted, clarifying at the start of each interview that our research was independent of Meta or any other SMF. The last point was particularly important in our engagement with non-SMF interviewees because our perceived proximity to Meta had the potential to interact with other power dynamics, including some of the global disparities we address later.
With regard to our positionality, the first author is a British scholar based in a UK university with extensive research experience in the African countries concerned, whereas the second author is a Nigerian researcher based, at the time of the work, in a leading Nigerian policy institute with a focus on digital rights and integrity. For logistical reasons, the first author conducted most of the interviews. Planned interviews and core questions were agreed on between the authors in advance, with interview transcripts shared securely online and discussed either remotely or, where possible, in person. Doing so helped ensure that potential assumptions and biases of each author concerning interpretations of the findings were identified and reflected upon on an ongoing basis. This included, for example, the highlighting of silences, omissions, “discomfort,” or ambiguities in interviews that reflected political, cultural, organizational, or interpersonal norms (Fujii Reference Fujii2015). It also allowed the authors to continuously reflect on possible blind spots derived from dominant narratives and orthodoxies within our respective fields and relevant scholarship. This included, for example, the nature and calibration of Majority World agency/agencies within SMF ecosystems.Footnote 12
More generally, we follow a range of scholars in acknowledging the shifting and hybrid character of researcher identity and power dynamics in data collection processes such as those outlined earlier (Stroup and Goode Reference Stroup and Goode2023; Zhao Reference Zhao2017). During interviews with Nigerian and Ugandan officials, for example, author one’s positionality often balanced elements of (semi-/)insider and outsider status, given his long-standing research work on East African policy elites and author two’s wider role in the project and professional networks in Nigeria. Respondents also represented a spectrum of positions on social media regulation and media freedoms. We emphasized in every interview that our source of funding and general positionalities did not tie us to a particular posture on Facebook/SMFs and state regulation. Nor did they provide us with direct mechanisms to influence Facebook/SMF policy making beyond the ability to share our independent findings with interested SMF staff who had engaged with the project. Engagement and interviews with current or former Meta staff were also characterized by a complex negotiation of trust-building because respondents were aware that, even though we had received funding from the company, our published findings would reflect our independent analysis.
Social Media (Firms) and International Relations
Social media’s entanglement with global politics—defined here as political interactions between major actors at all levels of the global system—has only recently begun to be considered explicitly within IR (Atal Reference Atal2021, 337; Srivastava Reference Srivastava2023, 997). This work has come to theorize social media’s significance for IR scholars across two main areas of concern. In both cases, these literatures generally examine the organizations behind social media platforms themselves in collective or abstracted terms. Their agency, in this regard, is either unspecified or is enacted by something with, in Arun’s words (Reference Arun2021), a single “face”—one belonging to the firm itself or, more often, to Big Tech more broadly conceived.
The first literature considers SMFs within the broader category of Big Tech companies and platforms and principally concerns the relationship(s) between Big Tech and the state. Big Tech, it is argued, has played an increasingly central role in variously enhancing, buttressing, reconstituting, and undercutting the power of states and (some) governments (Diamond Reference Diamond2010; Shirky Reference Shirky2011). Social media’s wider significance as a mobilizer of activist movements opposing dominant and oppressive political, economic, and societal systems has also been highlighted by scholars in relation to phenomena as diverse as the 2019 anti-government protests in Iran, the Occupy Wall Street and #BlackLives Matter movements, and the 2011 anti-austerity protests in Spain (Schirch Reference Schirch and Schirch2021; Zeitzoff Reference Zeitzoff2017). This optimistic take on Big Tech and global politics has nonetheless been increasingly balanced by analysis of the sector’s role in shoring up state machineries of surveillance and repression (Duncombe Reference Duncombe2018; Kreps Reference Kreps2020; Morozov Reference Morozov2009; Srivastava Reference Srivastava2023; Tan Reference Tan2020; Vaidhyanathan Reference Vaidhyanathan2022; Zuboff Reference Zuboff2018).
Srivastava has deepened and expanded this set of debates through conceptualizing Big Tech’s emergence as a—largely unaccountable—“private authority” within the international system. Big Tech’s collection of vast amounts of data from those using their services and, critically, the role played by these corporations’ algorithms in structuring, organizing, and presenting this data to different actors (“algorithmic governance”), Srivastava (Reference Srivastava2023, 989) argues, place Big Tech in a uniquely powerful position within the global system vis-à-vis both users, states, and private companies. This position raises, she notes, foundational questions about Big Tech’s accountability to users, states, and populations across the world (993). In the same vein, other scholars have exposed how Big Tech has often managed to elude statutes and mechanisms that would otherwise regulate actors with such power (Atal Reference Atal2021; Bradford Reference Bradford2023; Collier, Dubal, and Carter Reference Collier, Dubal and Carter2018; Thelen Reference Thelen2018). States and societies in the Majority World have often been particularly challenged in this regard, reflecting wider global inequalities, as we note later (Nothias Reference Nothias2020; Prasad Reference Prasad2018).
A second literature emphasizes the major role that social media networks have come to play in shaping public debate and global politics, sometimes with major consequences, especially in electoral contests and conflict theaters. This research remains divided on how direct a link can be drawn between activity or content on social media and, for example, an election result or intensification of violent conflict (Aral and Eckles Reference Aral and Eckles2019; Barbera and Zeitzoff Reference Barbera and Zeitzoff2018; Gilardi et al Reference Gilardi, Gessler, Kubil and Müller2022; Zeitzoff Reference Zeitzoff2017). There is more consensus, however, around the role that social media plays in agenda setting and shaping narratives around particular political events or contexts, often through the algorithmic amplification of divisive, sensationalist, misleading, or polarizing content. This has been most impactful, it is suggested, in conflict theaters such as 2020–22 northern Ethiopia or late 2010s Myanmar, where it is argued that Facebook’s algorithms “super-charged” hate speech (Crystal Reference Crystal2023; Sablosky Reference Sablosky2021).
The spread of rumor, disinformation, and other harmful material through the media is not, of course, an innovation of the digital age (Gilboa Reference Gilboa2005; Robinson Reference Robinson1999). As Kreps (Reference Kreps2020, 9) notes, however, social media permit the spread of information at a speed and scale almost unimaginable until recent history, meaning that the impact of some content can spread across the planet and provoke diplomatic and public outrage, counterclaims, and responses almost instantaneously. Moreover, social media content is generated by users, who see a unique curation of material organized by algorithms that are almost wholly focused on capturing and maintaining engagement. This renders the influence of SMFs on global politics much more varied and unpredictable than that of traditional media houses and means that these firms cannot shape debate in a linear fashion, relying instead on broad-spectrum algorithm tweaks or blunt instruments such as profile or “shadow” bans.
This literature also underlines how political actors are themselves active participants in these processes. Here in particular is where social media’s influence extends beyond the national level to that of interstate relationships. Scholars underline, for example, the ways in which political parties and governments across the globe have increasingly come to engage with social media platforms and, in some cases, “weaponize” them to undermine opponents and, in the case of so-called bad actors, to destabilize elections and other contentious processes in “enemy” states (Cheeseman et al. Reference Cheeseman, Fisher, Hassan and Hitchen2020; Dowling Reference Dowling2021; Duncombe Reference Duncombe2018; Gunitsky Reference Gunitsky2015; Kofi Annan Foundation 2020). The use of social media by terrorist and insurgent movements has been tied by some researchers to recruitment and fundraising efforts (Borelli Reference Borelli2023). Social media are also increasingly the primary means through which many political leaders and organizations engage with supporters, citizens, governments, and the wider global system (Duncombe Reference Duncombe2018; Gilardi et al. Reference Gilardi, Gessler, Kubil and Müller2022; Loyle and Bestvater Reference Loyle and Bestvater2019; Malefakis Reference Malefakis and Shirch2021; Zeitzoff Reference Zeitzoff2017).
This range of contributions make a clear and convincing case for the impact and influence of social media on global politics and the need for IR scholars to focus their analyses on social media and on Big Tech more widely. First, they establish how social media provide a vehicle of unprecedented reach and popularity through which political actors can engage directly with one another and with audiences at every level. Second, they demonstrate that sometimes as a consequence of the former, social media can inspire, stimulate, undergird, or facilitate mass behavior of all kinds from activism and informed debate to voting trends, protests, insurgency, and conflict. Existing scholarship also underscores how central content moderation—the processes whereby SMFs both set and police community standards around acceptable user-generated content on their platforms—can be to the calibration of social media’s influence.Footnote 13 That is to say, the influence of SMFs on global politics comes not so much from their hosting platforms but from their delineation and implementation of the rules, processes, and systems that ensure that some content and profiles enjoy greater or lesser prominence among different audiences.
We suggest, in this regard, that contemporary IR has thus far underspecified or underexamined SMFs’ place in global politics in two interlinked ways. First, although considerations around accountability, algorithmic governance, and the mediation of state–society relations arguably apply across all platforms, those such as Amazon, Uber, or PayPal play a limited role in shaping or influencing politics outside their sectors. Similarly, although most major corporations seek to lobby and influence policy makers and publics to secure better terms for their business, as many scholars have demonstrated SMFs’ business is to host, curate, and oversee connectivity, speech, and debate itself, at least within the digital realm. SMFs merit examination, therefore, as, at the very least, a subset of Big Tech, one with a quite distinctive set of relationships with global politics by virtue of their product and profit model.
Second, although recent IR and legal scholarship has wrestled with the question of the kind (s) of actor SMFs are in the global system, SMFs have generally been explored in the aggregate or at the highest level of abstraction. They are presented, implicitly or otherwise, as the name behind a product, the corporate owner of a particular set of algorithms, or the fulcrum of a network of “infrastructure, services, devices, and knowledge production” (Kreps Reference Kreps2020; Monsees et al. Reference Monsees, Liebetrau, Austin, Leander and Srivastava2023, 2; Srivastava Reference Srivastava2023). We contend that this important work would be enriched by taking a more organization-centered approach to analyzing SMFs and their influence. SMFs outside China share many characteristics, ideas, and even personnel. At the same time, they rarely act as a collective. Their engagement and entanglement with global politics—ultimately, their influence—are partly determined by internal debates and dynamics and by the relationships of CEOs, executives, and teams across the firm. The remainder of this reflection outlines what such an approach might look like and reveals how it can help us further understand the nature of SMFs’ influence in global politics.
Social Media Firms as Actors in Global Politics: Inside the “Black Box”
The place and agency of corporations within the international system have received significant scholarly interest (Cutler Reference Cutler2003; Green Reference Green2013; Korten Reference Korten2015; Kustermans and Horemans Reference Kustermans and Horemans2022; Mikler Reference Mikler2018; Nye Reference Nye1974; Strange Reference Strange1988; Reference Strange1996). Particular attention has been paid to the nature of corporate power, influence, and authority, especially in relation to that of the state. Recent work on Big Tech falls within this tradition, focusing on strategies deployed by companies such as Google and Uber to sidestep or renegotiate state regulatory regimes (Atal Reference Atal2021; Bradford Reference Bradford2023; Srivastava Reference Srivastava2023; Thelen Reference Thelen2018). What this set of literatures have paid less attention to are the organizational dynamics of firms themselves and how they come to calibrate the influence of SMFs in global politics. SMFs “act,” we contend, not only as a corporate sector or as companies with a single voice but as organizations of people and teams with differing interests, objectives, and relationships, both internally and externally.
In adopting this lens, we suggest an analysis at two levels. As we demonstrate later, these levels constantly interact and intersect; they can both reinforce and rub up against one another. First, at the organizational level, SMFs merit examination as political actors in space and time. Analysis of the cultural and environmental influences on corporations and other large bureaucracies has a lengthy pedigree in management and organization studies, often centering around questions of organizational effectiveness and performance (O’Doherty and De Cock Reference O’Doherty and De Cock2024; Pettigrew Reference Pettigrew1979; Pfeffer and Salancik Reference Pfeffer and Salancik1978). Such considerations have been introduced into IR more recently through the work of international political economy scholars such as Mikler (Reference Mikler2018), who argue that global corporations’ political influence is refracted through their spatial and temporal characteristics. Few corporations, it is argued, are truly “transnational” as opposed to multinational in the sense of where they operate; where their employees, executives, and decision makers are drawn from and work; and the extent and significance of their external networks. Consequently, their agency is enacted differently in particular settings and polities. “Re-territorializing” and “re-embodying” SMFs—to use Mikler’s terminology—requires a closer examination of their different sociologies and demographics, as well as their territorial presence, character, and political embeddedness.
Second, at the intra-organizational level, SMFs should be analyzed as places of internal debate, contestation, and power plays. Different teams have different objectives, and policy decisions—and their impact on global politics—can reflect these dynamics. SMFs contain within them multiple and sometimes competing institutional logics, defined by Battilana, Leca, and Boxenbaum (Reference Battilana, Leca and Boxenbaum2009, 69) as “shared understanding[s] of [the organization’s] goals…and how they are to be pursued”—and institutional cultures. The extent to which these logics and cultures are reconciled, accommodated, hybridized, or suppressed mirrors the internal politics of the organization, as well as the coalitions built around specific debates and issues. These moments of negotiation can also lead to the emergence of new agential constellations that can come to condition SMF influence in different global contexts. Within organization studies, this level of analysis has been applied across a range of private and public sector bodies but not yet, to our knowledge, to SMFs (Battilana and Dorado Reference Battilana and Dorado2010; Czarniawska Reference Czarniawska2009; de Aquino and Batley Reference De Aquino and Batley2022; Lepori and Montauti Reference Lepori and Montauti2020; Pache and Santos Reference Pache and Santos2013; Ratinho and Bruneel Reference Ratinho and Bruneel2024; Thornton, Ocasio, and Lounsbury Reference Thornton, Ocasio and Lounsbury2012). In this section, we apply this two-level framework to analyzing SMFs, shedding new light on the character and negotiation of their influence in global politics. We focus principally on policy and practice around content moderation, having identified this as the foremost mechanism through which SMFs influence global politics.
SMFs as Organizations
The sometimes highly consequential decisions that SMFs make about the amplification, removal, or minimizing of online content are made within often large, complex infrastructures of automated technologies and human teams based across multiple jurisdictions and time zones. Understanding the organizational drivers of these dynamics is key to understanding the nature of SMF influence in global politics. In this regard, although the larger SMFs may have offices, hubs, service centers, and staff across the globe—Meta has 28 offices outside North America and Europe, at the time of this writingFootnote 14—they remain, in essence, US organizations with fairly homogeneous staff profiles, at least at the headquarter and leadership level.
Each SMF has its own content moderation guidelines (“community standards”): These are both “living documents” and reflections of the organization’s culture and its evolving assessments of political and financial imperatives (Gillespie Reference Gillespie2018). Facebook, YouTube, and Twitter moved at different speeds during the 2000s and early 2010s with regard to content moderation, with Twitter’s leadership and workforce “most defensive of free speech norms” (Klonick Reference Klonick2018, 1620–21). More recently, Elon Musk and his lieutenants have presented X’s commitment to “rigorously adher[ing] to … free speech” as integral to the platform’s identity.Footnote 15 This is not to say, however, that SMF content moderation norms differ dramatically across firms; Klonick emphasizes that emerging content moderation cadres across 2010s Facebook, Twitter, and YouTube were composed largely of “American lawyers trained and acculturated in American free speech norms and First Amendment law” (1621–22). Shifts in content moderation policies have generally taken place in an ad hoc, reactive fashion in response to pressure from governments, the media, civil society, or users themselves around platforms’ perceived failure to adequately tackle the sharing of misleading, harmful, hateful, or illegal material. Increasingly, these criticisms have developed into legislation in many polities, with which SMF compliance has varied considerably across jurisdictions (Bradford Reference Bradford2023).
Across SMFs, the more impactful and sensitive decisions around content moderation globally are generally made by largely US-based executives, with staff employed outside the United States or Western Europe given limited influence, despite their being embedded in relevant contexts. Linked to this pattern is sustained underinvestment by most SMFs in content moderation staff and technology with the ability to understand and contextualize non-European languages and contexts (Kofi Annan Foundation 2020, 49–54; Siegel Reference Siegel, Persily and Tucker2020, 72–74). Internal company documents leaked by whistleblower and former Facebook product manager Frances Haugen in 2021 show, for example, that at that time 87% of Facebook spending on combating misinformation was devoted to English-language content (Milmo Reference Milmo2021). This is an issue even within Europe, with only 11% of YouTube content moderators reviewing content not in English and only 8% of X content moderators being proficient in an EU language other than English in recent years (Global Witness 2023).
Indeed, in the case of X, there appears to have been an active disinvestment in content moderation outside Western markets after Musk’s 2022 takeover. According to one former Twitter staffer, for example, the platform’s Accra office was originally meant to focus only on Nigeria and Ghana but rapidly had its remit stretched to include Kenya and other African states, forcing its around 20 employees to “have to try and work out how to understand different languages, slang terms, and cultural contexts” across an entire continent.Footnote 16 This office—then Twitter’s sole site in Africa—was largely and unceremoniously closed shortly after Musk’s acquisition of the platform, with some employees made aware of their dismissal only after being unable to log into their Twitter-issued laptop.Footnote 17
Former staff of both X/Twitter and Meta argue that this apportionment of resources derives, in part, from a general lack of familiarity with, interest in, or understanding of non-Western societies by SMF executives, who are mostly US, Canadian, or European citizens, the majority of whom are white and male, and, indeed, by their wider workforces, which also reflect these national, ethnic, and gender imbalances (Statista 2022).Footnote 18 “They don’t prioritize the [African] market,” one former Twitter/X employee noted. “It is almost like an afterthought … they are not culturally aware of the African social media landscape. Why would they prioritize something they don’t know?”Footnote 19 Another argued that Musk’s takeover had seen a further shift away from the continent: “Pre-Elon, Twitter was trying to better understand Africa … but now it’s all about the money, not development, and Africa is seen as too unpredictable.”Footnote 20 Indeed, several former Twitter staff highlighted the company’s color-coded “internal classification” of countries and markets based on a range of factors, “including human rights, hostility to Twitter, and so on”—noting that countries such as Nigeria were “closer to North Korea, red” than to the UK or US “green.”Footnote 21
For many of our African state respondents, this culture has also manifested in the development, application, and adjudication of SMF content moderation guidelines perceived to be ill-suited to the contexts of many postcolonial states. Nigerian and Ugandan officials, for example, argued that X and Facebook platform guidelines are not “sufficiently attuned to the local context” and reflect the views and experiences of their “American leaders” rather than “the realities that African countries have.”Footnote 22 They provided evidence for these claims with a range of examples in which government requests for notionally harmful content and profiles to be removed were turned down by the platforms, usually with generic text explanations and unclear pathways of appeal.Footnote 23 Clearly state justifications for any closure of space for public debate must be interpreted with caution, particularly those from states with long histories of autocratic and abusive governance. At the same time, SMFs are certainly not consistent in responding to legislation and government requests based on that government’s democratic credentials (Amnesty International 2020; Whistleblower Aid 2022, 40).
These organizational characteristics suggest several implications for understanding SMFs’ influence on global politics. In particular, the spread and augmentation of inflammatory or inciteful online content in the Majority World hold much less attention and interest for SMFs than that in the Global North. Content moderation policy decisions related to the former, therefore, tend to be particularly reactive and to reflect lobbying by actors and coalitions that can readily command the attention of Silicon Valley executives. Such actors include, as Arun (Reference Arun2021) has noted, Western media houses, analysts, and NGOs—but also other SMFs.Footnote 24 Facebook’s engagement algorithms, for example, were argued to have amplified online hate speech and significantly contributed to human rights violations committed against Myanmar’s minority Rohingya community during the 2010s. Despite reportedly being aware of such allegations and being lobbied by local activist groups since around 2012, Facebook’s comprehensive efforts to tackle the issue were not initiated until 2018, on the back of concerns raised by the UN, New York Times, and Reuters (De Guzman Reference De Guzman2022; Warofka Reference Warofka2018). Reflecting on the firm’s response, a former Meta employee told Amnesty International that “different countries are treated differently [by Meta].… If 1,000 people died in Myanmar tomorrow, it is less important than if 10 people in Britain die” (Amnesty International 2022, 37).
Within SMFs themselves, the structural enfeebling of what some Silicon Valley interviewees described as “local [SMF] staff” can further weaken relationships between states and an SMF, placing citizen access to online spaces and, for some, even their livelihoods in jeopardy. After Facebook banned hundreds of Ugandan government accounts in 2021, the platform itself was ultimately banned in the country, with state officials arguing that Facebook executives were deliberately placing distance between themselves and Ugandan officials.Footnote 25 Ugandan state actors “negotiating” with the company on the breakdown in relations recalled that the “[Nairobi-based] regional team was open and wanted to find a solution. They [Kenya-based Facebook staff] understand the structure and culture of the region but we did not feel that they had sufficient influence with the team in the US.”Footnote 26 Another claimed that “we had some virtual meetings with Facebook’s regional team in Nairobi but nothing was addressed. They promised to investigate for a while, but no action was taken.”Footnote 27 For many of these respondents, interaction with Facebook further reinforced global political and economic asymmetries governing many African states’ relationships with Western states and the wider global system (Guardian 2023). In the words of one senior Ugandan civil servant,
Social media companies are very strong, very powerful, and in the African context, there is no [Facebook recognition of state] regulation [of platform content].… 3.5 million [Ugandan] citizens using the platform but they have no say, whereas the US Congress can summon [Meta CEO Mark] Zuckerberg before it. These African countries do not have any remedy, so countries have little choice to make decisions that may appear to interfere in freedom of expression.Footnote 28
Yet, this political and spatial distance between SMF decision makers and the Majority World, creates some space for agency among local actors, if not for SMF staff themselves, around the issues and content prioritized in content moderation work. SMFs rely on a range of mechanisms and actors to inform and place in context their content moderation work across the globe, particularly during contentious processes such as elections. Meta’s Trusted Partners Program, which provides a “designated channel” to 465 civil society and human rights organizations enabling them to provide input into the company’s policies, is one such mechanism (Meta 2023; however, see Internews 2023). This can place some NGOs, activists, and other groups in a relatively influential position vis-à-vis parsing local dynamics to Silicon Valley executives and amplifying a particular perspective or issue. Indeed, some of our Ugandan state interlocutors accused Facebook (the platform) of being unduly influenced by “opposition [groups] and NGOs” in arriving at their 2021 decision to suspend certain state-affiliated profiles, with one senior communications official contending that the firm had acted on information that was “fictitious, bogus, misconstrued.… It was a bloody [political] campaign!”Footnote 29
In this context, some African governments have claimed that banning a platform is, in part, a mechanism, in the words of one Nigerian official, “to get the attention of [the SMF’s executives] and to get them to sit down and talk.”Footnote 30 In the case of Nigeria, which banned Twitter in 2021 after the platform took down a post by then-president Muhammadu Buhari, adopting this “nuclear option” opened up a dialogue with the firm. According to our Nigerian state respondents, this dialogue touched on a range of issues from Twitter registering as a company in Nigeria and agreeing to a local code of conduct (both reportedly rejected by the platform’s representatives) to the opening of a Twitter office in the country.Footnote 31 These different enactments of African agency in influencing and shaping SMF engagement in certain polities nonetheless took place within and reflected the broader political and economic asymmetries that characterize many African states and constituencies’ interaction with the international system (Fisher Reference Fisher and Cheeseman2018; Mwambari, Munyi, and Ylönen Reference Mwambari, Munyi and Ylönen2020).
SMFs’ territorial and sociological embeddedness in the United States means that they are also sometimes deeply entangled in US politics, albeit in a quite different manner.Footnote 32 Their influence, in this regard, is often much more direct and multilayered. SMFs have been almost entrenched within recent US presidential campaigns themselves—playing, in the words of Kreiss and McGregor (Reference Kreiss and McGregor2018, 173), “an extensive infrastructural role in [US] democratic processes” as providers of products and campaign support. American companies owned and led by largely American CEOs and executives, SMFs are also economically, politically, and legally beholden to US-based constituencies in a way that they are not to those in other countries and continents. In Quarter 4 of 2024, for example, 45% of Meta revenue derived from the US/Canada (Meta 2025, 3). In the same period, nearly half of Alphabet’s revenue (the parent company of YouTube) came from the United States alone (49.1%; Alphabet 2025, 8). In this context, SMFs have at times been under greater pressure to demonstrate to US legislators, media houses, and civil society organizations that they are managing content responsibly but also without political bias.
Following 2016 allegations that Facebook was suppressing conservative political content on its Newsfeed feature, for example, CEO Mark Zuckerberg felt the need to host a wide-ranging, trust-building discussion with conservative US commentators at the company’s California headquarters, in which Facebook executives “acknowledged that there was a problem with getting messages out to conservatives,” according to one attendee (Woolf Reference Woolf2016). Subsequently, the firm has oscillated between seeking to persuade US constituencies of its distance from or deference to Donald Trump’s political campaigns, depending on the prevailing political winds. After Trump’s 2016 US presidential election, for example, the platform outlined steps it was taking to avoid capture and subversion by pro-Donald Trump forces, when evidence of Russian use of the platform to interfere in the poll led to a series of high-profile investigations (Wong Reference Wong2018). In the lead-up to Trump’s 2025 return to the White House, however, Zuckerberg announced a range of policy changes to Meta hiring practices and to its use of fact checkers in the United States—both previously criticized by Trump and his supporters—to reflect “the shifting and legal policy landscape” (Rodriguez Reference Rodriguez2025; Szadowski Reference Szadowski2025).
Examining SMFs at the organizational level therefore underscores how central spatial, territorial, and cultural dynamics are to the calibration of SMF influence across the globe. SMF engagement with US policy, media, activist, and business communities is of a fundamentally different nature and depth from that with equivalent groups in many other polities. This is particularly so in the Majority World, where SMF influence is often experienced as a potent but faceless external force, refracted through envoys and civil society groups whose own agency shifts frequently and unexpectedly. This is not, however, to say that SMF content moderation in the United States is necessarily more considered, appropriate, or politically sensitive. Indeed, SMF interventions in US politics have been some of the sector’s most controversial and contested actions. This reflects the organizational characteristics we already examined: SMFs are more deeply implicated in US political and economic dynamics than in those of any other state. It also, however, reflects intra-organizational factors, a level of analysis we turn to now.
Intra-Organizational Factors
Even at the headquarter level, SMFs have multiple teams working on and influencing content moderation principles and implementation. In many cases, separate teams engage with states than those who work with NGOs or researchers (Arun Reference Arun2021). These teams operate within different and sometimes competing institutional logics and cultures. A central dynamic here is tension around how far the platform’s content moderation policies and norms should be independent of its financial objectives and US-centricity, which, as noted earlier, are interlinked. Most social media platforms are funded by advertisers, rendering user engagement central to the company’s success. Inciting, offensive, or controversial content— including that posted by prominent political actors—often leads to increased user engagement, even if its character can be seen as contravening community standards. This creates a quandary for SMFs, which must balance reputational and governance concerns against the risk of diminished engagement and revenue. Critically, from the perspective of global politics, this debate plays out frequently within SMFs, with different teams and constituencies mobilizing around a particular policy option. The outcome of these negotiations can have significant consequences for the (de-/)amplification of particular political voices and agendas online and the application of consistent approaches worldwide.
Relevant, in this regard, is the pattern of acquisitions that has characterized the SMF sector since the mid- to late 2000s (Atal Reference Atal2021, 339–40).Footnote 33 This has led to the incorporation of teams and companies into the wider SMF that do not necessarily share the same perspectives, priorities, or culture. WhatsApp, for example, was granted a degree of autonomy after its 2014 acquisition by Facebook, and its staff continued to operate from a separate headquarters (WhatsApp 2014). Over time, however, according to one former WhatsApp employee, the company came to be gradually “colonized”Footnote 34 by Facebook executives and priorities, with a formal merger into the Meta brand in 2021. This has created opportunities for some staff but has bred resentment among others who perceive that WhatsApp’s historical focus on global accessibility and affordability has been steadily eroded by a parent company interested principally in the US market. As one former WhatsApp staffer recalls,
There were a lot of fights [with Facebook] over core values, particularly Facebook’s focus on the US.… there was an issue when trialling the “From Facebook” splash screen [which appears when first engaging] with WhatsApp because the internet connection in some countries, especially those in Africa and Asia, means it takes a longer time to actually open the app. The Facebook engineer types said, “Well, this is not an issue because there isn’t a lagtime in the US.” They just don’t have a global view, the US is seen as where the money is … Africa is just not considered important.Footnote 35
For this respondent, this example spoke to a wider disconnect between Facebook and WhatsApp staff regarding how far contexts beyond North America and Western Europe should be considered in product design and content moderation policy.
SMF CEOs play a particularly critical role in setting the terms for and adjudicating such internal debates, making their actions significant for understanding SMF influence on global politics. Internally, they possess immense power over content moderation decisions, having the ability to accelerate, (de-/)prioritize, reverse, or veto virtually any development or process. This power is partly structural: They tend to be the majority shareholder (s) in their company (Lynley Reference Lynley2012; Srinivasan Reference Srinivasan2023). In the case of Meta, the creation of a dual-class share structure from early on has meant that Zuckerberg retains absolute control of the company and its decisions and, in the words of one former Facebook employee, is “basically untouchable” internally (Delouya Reference Delouya2022).Footnote 36
SMF CEOs also derive considerable personal authority over and legitimacy with staff from their reputations as tech or engineering geniuses or wunderkinds. In such contexts, many employees work to secure the CEO’s notice, approval, and patronage. Roger McNamee (Reference McNamee2019, 144), mentor and later a strong critic of Zuckerberg, has characterized the latter’s internal profile as “combin[ing] elements of rock star and cult leader”; one former Meta interviewee we spoke to noted how “as soon as Mark [Zuckerberg] prioritized something, it would pull everyone in the company into it … people were immediately jumping on planes and things like that.”Footnote 37 Indeed a range of respondents were particularly critical of the “Facebook bros” who compete for the CEO’s attention.Footnote 38 Others, however, offered a more sympathetic appraisal, with one employee arguing, “It’s easy to be critical of [Zuckerberg]— he’s such a massive public figure—but there’s a lot of respect and admiration there [from Facebook staff].… He’s someone who is super smart, has a vision, and is really changing the world. … That’s pretty amazing, right?”Footnote 39
This dimension of CEO power is amplified in many contexts by their unusual global status. Although some SMF CEOs, notably ByteDance founder Zhang Yiming and former YouTube CEO Susan Wojcicki, have deliberately sought to maintain a low profile on the global stage (Bergen Reference Bergen2022, 397; Stokel-Walker Reference Stokel-Walker2023), others have relished the limelight. In cases, such as with Zuckerberg/Meta and Musk/X, the economic and political influence of SMF executives and the platforms they own have become almost inseparable (cf. Hägel Reference Hägel2020). Musk’s Starlink satellite communication company and its (proposed, alleged, or actual) deployment in conflict theaters from Ukraine to the Middle East have, for example, been difficult to disentangle from the magnate’s sometimes incendiary or offensive tweets or his high-profile engagements with presidents and prime ministers (Burgess Reference Burgess2023; Klee Reference Klee2023; Marks and Newman Reference Marks, Newman and Bloomberg2023; Wall Reference Wall2022). This issue has been further complicated by the evolving alignment between Musk and Donald Trump since 2024, in which X was accused of amplifying conservative content and Republican campaigns during that year’s elections and Musk is now a part of Trump’s second administration (Hernandez Reference Hernandez2025; Murphy Reference Murphy2024; Ortutay Reference Ortutay2024).
With regard to Meta, speculation that Zuckerberg was himself going to run for US president was rife in 2017–18 on the back of the Facebook chief’s national “listening tour” (Newton Reference Newton2018; Quartz 2017), and he has been received in countries such as India by senior politicians as a pseudo-diplomatic figure.Footnote 40 SMF CEOs are therefore able to draw on a range of registers—global celebrity, tech magnate, free speech guardian, (in some cases) pseudo-humanitarian, business leader, political influencer, envoy of US capital, etc.—in part because of the ambiguous character of SMFs themselves (Atal Reference Atal2021). This spectrum of identities can further reinforce their authority internally.
Combined, these factors render CEOs—and their most trusted executives—central to some of the most consequential decisions vis-à-vis SMF influence on global politics. This is especially true of decisions perceived internally to be particularly “political,” including de-/amplification or removal of certain content and profiles or of content moderation processes (Levy 2020). According to whistleblower Frances Haugen, in 2019 Facebook’s Civic Integrity team was tasked with developing a rubric to regulate when to remove hate speech from politicians after concerns were expressed internally that the platform was applying different criteria to evaluating the posts of Indian and US politicians. Zuckerberg rejected the team’s draft policy as “insufficient,” instead producing his own version “over the weekend” (Haugen Reference Haugen2023, 182–83). Indeed, McNamee (Reference McNamee2019, 142–44) argued that “there is a core team of roughly ten people who manage the company, but two people—Zuck[erberg] and [2008–22 chief operating officer] Sheryl Sandberg are the final arbiters of everything.… Zuck is known for micro-managing projects and being decisive.… It is the most centralized decision-making structure I have ever encountered in a large company.”
In the case of Twitter (now X), soon after becoming CEO, Musk initiated a range of sweeping changes to content moderation, including a significant disinvestment in content moderator teams and staff themselves; this led to a subsequent significant increase in misogynistic content and targeted harassment campaigns, including foreign influence operations (Spring Reference Spring2023). Former Twitter staff focused on Africa report having seen the impact of these changes almost immediately. “I feel defeated when I look at Twitter these days,” one explained. “We made so many positive changes whether it was on spam, giving context, or 24-hour monitoring, but now it’s all gone.” Another noted, “It feels like we have gone three steps backward with [content moderation on] Twitter in Africa.”Footnote 41 Musk was also integral to restoring Donald Trump’s profile to Twitter in 2022, following the then-US president’s 2021 “permanent” ban from the platform (Meierhans Reference Meierhans2022; Veiga Reference Veiga2022).
An intra-organizational analysis of SMFs nonetheless includes examining SMF CEOs within the wider context of their company. Such an approach can reveal some of the internal checks and influences on their executive power and the role that different teams and staff can play in shaping key decisions. Twitter’s pre-Musk leadership, for example, was divided over banning Trump from the platform in 2021 but came under significant internal pressure from different teams to do so, ultimately conceding to their arguments (Conger and Isaac Reference Conger and Isaac2021). In the case of Facebook, internal discontent with the leadership’s approach to mitigating harmful content online emerged during the mid-2010s after a series of revelations exposed how the Cambridge Analytica consulting firm had harvested the Facebook data of millions of users to be deployed in political advertising (Hu Reference Hu2020). According to one former Facebook employee, “It was just an endless series of [reputational] body blows and Mark and the leadership just weren’t doing enough to respond.”Footnote 42 Partly in response to internal pressure, Zuckerberg pledged at the start of 2018 to “fix” Facebook, acknowledging, “We currently make too many errors enforcing our policies and preventing misuse of our tools.”Footnote 43 More recently, in mid-2020, Facebook staff held a virtual walkout in protest at the platform’s perceived half-measures in tackling then-US president Trump’s inflammatory online content (Arun Reference Arun2021, 252; Yurieff and O’Sullivan Reference Yurieff and O’Sullivan2020).
Current and former SMF staff also bring their influence to bear as whistleblowers, media interviewees, research participants, and authors of tell-all exposés. These actions often derive from a position of disaffection or disillusionment with the company and a sense that the leadership does not take some perspectives seriously.Footnote 44 At the same time, they can sometimes force SMF executives to publicly acknowledge, address, or mitigate issues that were previously neglected internally, with several SMF whistleblower allegations leading to US congressional hearings in recent years (Frenkel and Kang Reference Frenkel and Cecilia2021; O’Sullivan, Duffy, and Fong Reference O’Sullivan, Duffy and Fung2022). In other cases, public criticism by former senior figures can lend weight to internal campaigns that are already underway. Following the Cambridge Analytica revelations, for example, former Facebook vice president for user growth, Chanath Palihapitiya, told a Stanford audience that he felt “tremendous guilt” in his role in Facebook’s development, reflecting that “I think we have created tools that are ripping apart the social fabric of how society works” (quoted in Vincent Reference Vincent2017). This comment reflected, according to one former Facebook employee, “what quite a number of us were thinking and saying.”Footnote 45
In concluding this section, however, it is important to emphasize that these intra-organizational agency dynamics tend to mirror the same spatial, territorial, and normative asymmetries that we have shown to be at the heart of SMF influence in global politics. Some of the most prominent and influential whistleblowers have been embedded within the worlds of US policy, media, tech, and activism, and internal mobilization around banning Trump in many SMFs in 2021 intersected with wall-to-wall debate on the issue curated by media houses, politicians, and cable channels and consumed by SMF staff.
Discussion and Conclusion
The role and influence of social media in global politics has become an increasingly central consideration for IR scholars. In this reflection, we argue that this important research agenda would be enriched by placing greater focus on the corporations behind platforms and their algorithms. Drawing on a range of primary sources, we advocate for the study of social media firms (SMFs) as organizations, highlighting the role, significance, and agency of the personnel and institutions behind their platforms and algorithms. In this concluding section, we summarize our findings and reflect on their implications for future research.
Our central critique of existing literature is its tendency to treat SMFs in an abstracted or aggregated form or to overlook them altogether. These approaches obscure important aspects of social media’s impact on global politics. Placing a focus on SMFs, and analyzing them as spatially and temporally located organizations, sheds critical new light on how and why social media have the influence they do, from electoral processes and conflict theaters to diplomatic relations and political trajectories. It also helps explain why this influence manifests so differently in various polities and contexts, as well as its intersection and interaction with other forces structuring global political and economic power.
Social media’s entanglement with global politics, we emphasize, is significantly refracted through the norms, worldviews, and hierarchies found within SMFs, dynamics that center and privilege American (and Western) politics, debates, standards, personnel, and networks. These are embedded within processes of content moderation, decision making, prioritization, and product development, reflecting SMFs’ provenance and key markets, the dominance of US citizens and US-trained personnel within their executive cadres, and the California and Western European locations of their headquarters and decision-making nerve centers. As we showed, this means that SMFs pay limited attention—and devote limited resources—to their platforms’ footprint, use, and impact in the Majority World. The consequences of this for global politics are significant. A lack of investment in—and disregarding of—staff, initiatives, and networks outside the Global North can lead to harmful online content being spread and amplified, ultimately underpinning the kinds of violence described in this study. At the other end of the spectrum, SMFs are consistently and deeply implicated in US electoral processes. As this reflection demonstrates, in recent US elections, major SMFs have been responsive, both by necessity and inclination, to pressure from legislators, media houses, candidates, advertisers, and their own staff on a raft of politically sensitive content issues, including profile suspensions, algorithm tweaks, content amplification, and “shadow bans.” For Meta, these efforts have been aimed principally at assuring US audiences of the company’s notional political evenhandedness. Under Musk, however, X has increasingly gone in the opposite direction (Connolly Reference Connolly2024; Davis Reference Davis2024; Schofield Reference Schofield2024).
SMFs’ entanglement in global politics underscores a key finding running throughout our analysis. Examining SMFs as organizations reveals how often their influence on global politics is the consequence of decisions made and pressure applied by human beings. This includes not only major one-off decisions but also more structural dynamics. We demonstrated, for example, how the de-prioritization or neglect of Majority World relationships, engagement, and personnel by major SMFs has often been the result of decisions taken and assessments made by executives and managers. This has sometimes had a knock-on effect for populations’ access to key social media platforms in Majority World states such as Nigeria and Uganda. In those cases, policy makers felt affronted by what they perceived to be aloof and disrespectful responses to their content moderation queries by US-based Twitter and Facebook staff, respectively, leading to a breakdown in state–SMF relations and a 2021 ban of the platform in their territory.
SMF founders, CEOs, and senior executives loom large in this reflection: Clearly, they exercise immense power over the direction and culture of their companies, as well as over politically sensitive processes and decisions. Examining SMFs as organizations nonetheless helps expose the wider range of actors and agential forces that calibrate and determine these firms’ influence. The limited SMF footprint outside the Western world, for example, provides space for some NGOs, analysts, and other “trusted” interlocutors in African and Asian states to play an outsized role in parsing SMF understandings of certain political dynamics. Even for these groups, however, their prominence for SMF teams in California, Washington, and London is often the result of curation by Western media outlets and advocacy networks. In this regard, looking within SMFs can be particularly instructive for scholars of social media and its influence. Contrary to what is implied in some analyses, SMFs rarely think and act as single entities. Interrogating these actors as organizations can help, as we have shown, identify the agency of different internal coalitions and teams in driving, revising, or reversing key policies—from profile suspensions to modulating content moderation in different national and political contexts.
Opening the organizational “black box” of SMFs and the roles and relationships of their key decision makers challenges us, ultimately, to further refine our understanding of agency and structure in global politics. As we have argued throughout this study, SMFs hold an almost unique role as governors and curators of both the mechanisms and content of political exchange and public diplomacy in, between, and beyond polities across the world. They blur the line—deliberately, in Atal’s analysis (Reference Atal2021)—between the public and the private in a manner that renders the extent, nature, and character of their power and authority particularly challenging to conceptualize and, as a range of governments have found, to discipline or regulate. Moreover, as scholars of philanthropy and billionaires in IR have noted, the relationships among private wealth, corporate power, celebrity, and global political influence are profoundly ambiguous and, in some cases, almost impossible to meaningfully disaggregate (Hägel Reference Hägel2020; Krcmaric, Nelson, and Roberts Reference Krcmaric, Nelson and Roberts2023). Elon Musk’s 2024–25 transformation into Donald Trump endorser, surrogate, intimate, and presidential adviser—in part through his stewardship of X—is particularly germane in this regard. Future research can help further specify and expose not only how and when spaces for agency open and close within SMF ecosystems but also the extent to which SMFs evolve and “learn” through their interaction with global politics, and with what theoretical and empirical implications for IR. This could include not only intra-organizational learning—for example, on content moderation or crisis management—but also how far SMFs learn from one another in and concerning the political and diplomatic arena. Finally, IR is well placed to critically reflect on how SMF teams and executives themselves increasingly approach some of these questions.
Acknowledgements
We would like to thank Niheer Dasandi, Marie-Eve Desrosiers, Julia Gallagher, Lipika Kamra, Tony Porter, Netina Tan, and the two anonymous reviewers for valuable and constructive feedback received on earlier drafts of this paper. Previous versions were presented at the Centre for International Policy Studies, University of Ottawa (October 2023), Research in Progress Seminar, Department of Political Science, McMaster University (October 2023), and Africa Talks/International Development Department Seminar Series, University of Birmingham (November 2023). We are grateful to the organisers of these events and to those in attendance for their thoughtful and insightful advice and contributions. This work was supported by funding received under Meta’s Foundational Integrity Research: 2021 Misinformation and Polarization call.
Open access
