Stop hate for profit: Evaluating the mobilisation of advertisers and the advertising industry to regulate content moderation on digital platforms

Introduction

Large social media platforms generate “substantially all” (Facebook Inc., 2023) of their revenue from advertisements. On social media, advertising interests are closely linked to content moderation practices, where filtering, removing, and recommending content “produc[es] a media commodity” (Gillespie, 2018, p. 43) meant to attract users and advertisers, and where advertising itself has been used to circulate propaganda and misinformation. Observers of the regulation of social media platforms have suggested that advertiser investment in norms of “civility and democracy” (Craig & Cunningham, 2019, p. 8) make them plausible allies in efforts to moderate extreme content online. However, examples of advertiser efforts to restrict funding for controversial content also suggest that half-hearted attempts to institutionalise advertiser-preferred content standards may have far-reaching consequences for user content, creator livelihoods, and the economics of journalism outlets (Aronczyk, 2020; Bishop, 2021; Parker, 2021; Willens, 2020; Yin & Sankin, 2021). Regardless, in a platformised online economy (Helmond, Nieborg, & van der Vlist, 2019), advertisers have sometimes been approached by civil society as productive actors with which to contest content moderation and monetisation, raising questions about legitimising private companies as appropriate governance intermediaries (Braun et al., 2019). This article compares the goals and outcomes of industry- and activist-led efforts that leverage advertisers to influence content moderation, and considers how these efforts fare as governments intervene more strongly in platform governance.

The Global Alliance for Responsible Media (GARM) and the Stop Hate for Profit (SHfP) campaign represent two efforts that leverage advertisers’ economic relationship with platforms to influence content moderation. The two initiatives—an alliance of major advertising associations, agencies, and large platforms in the case of GARM and an organised boycott of Facebook’s advertising services for the SHfP campaign—demonstrate some of the complex configurations of actors that come together around content moderation below the level of the state. These initiatives indicate the central place of the advertising industry in contests over platform content moderation as both a structure that influences content production and an active agent in public disputes over content. In the GARM case, the advertising industry partnered with large platforms directly to try to create consensus around platform content moderation standards and to solve recurring issues with quality and controversy in online advertising. In the Stop Hate for Profit case, civil society and activist groups pressured and encouraged advertisers to boycott Facebook advertising services to pressure the platform to change its content moderation practices around hate speech and racism. While each existed in a specific context, the two approaches continue as voluntary advertising organisations, such as the Conscious Advertising Network, organise to influence how the industry addresses topics such as human rights and climate change misinformation, and activist groups such as Check My Ads, investigate how bad actors manipulate the adtech industry.

This article compares the actors and outcomes of the GARM and the Stop Hate for Profit campaign and assesses their implications for platform content regulation. It considers the fragmentation of the platform landscapes and efforts to support or contest forms of content moderation centred on the advertiser-platform relationship. It finds that boycotting advertisers’ claims to fund Facebook were undermined by the scale of the platform’s advertising network; that the GARM was used by Facebook as a shield from the more radical demands of the Stop Hate for Profit coalition; and that social justice values such as opposition to racism that were present in the Stop Hate for Profit campaign did not appear in the governance documents created by the GARM. Finally, it notes the absence of state actors, policymakers, and regulators in both cases, demonstrating how advertiser standards of brand safety are being institutionalised as content moderation standards by some groups even as those efforts are contested in the early stages of a regulatory turn (Flew, 2022) that is taking very different forms in Europe and the United States. The article concludes with a discussion of how policy is responding to platform content moderation and whether this regulation will disrupt the two strategies observed in Stop Hate for Profit and the GARM. It compares the aims of the Digital Services Act in the EU with lawsuits against the GARM for anti-competitive practice over their interventions in platform content moderation.

Literature review

The power of digital platforms, their dominance in online advertising, and their entanglement with nearly all business operations has prompted calls for heightened corporate responsibility in governing platform surveillance, data collection, and the attention economy (for instance, in Flyverbom et al., 2019). How ads are placed and who sees them are controlled by the platform, which retains a central place in the power structure (Helmond et al., 2019), placing advertisements using proprietary algorithms and moderating content mostly in line with the normative values of US social and legal systems (Barrett & Kreiss, 2019; Klonick, 2017). However, the advertising industry has been singled out for its potential to influence platform practices because of the financial relationship between their industry and platform profits.

Critical political economists have theorised that advertisers’ financial power over their publishing partners gives them a negative, censor-like power over media content through structural practices as well as direct interventions (Hardy, 2023). Online, scholars argue that the prestige and financial power of major advertisers are formative to the structure of many parts of the internet, including social media platforms (Gehl, 2014). Automatic placement of ads, combined with social media’s struggles over content moderation has led to a series of public crises for online advertisers. In 2017, investigative journalist Alexi Mostrous revealed how the advertisements of prestigious brands were being placed automatically on websites and YouTube videos linked to terror groups and inappropriate sexual content (Mostrous, 2017). In response, advertisers such as Procter & Gamble and Pepsico boycotted YouTube, demanding guarantees of a brand safe environment, meaning one in which advertiser reputations are not damaged by content on the site (Cavale, 2020; Statt, 2017). To meet advertiser expectations of brand safety, YouTube changed its standards for sharing advertising revenue with partner channels, including adding stricter use of keyword blocking and higher bars before channels could earn advertising revenue. The resulting “adpocalypse” transformed monetisation for YouTube creators, creating new standards for assessing content suitability and fueling the “tiered governance” model that advantages some creators and content over others, likely affecting moderation and recommendation of content overall (Burgess & Green, 2018; Caplan & Gillespie, 2020; Hill, 2019; Kumar, 2019). The GARM institutionalised these standards of brand safety, and broader concepts such as brand suitability, to guide platforms on how to categorise content as compatible with major advertisers and incentivise platforms to restrict content that is not, with potentially widespread impacts on content and creators able to earn advertising money and content moderation practices more generally (Griffin, 2023).

Advertisers’ perceived responsibility for content online is reinforced by activist groups such as Sleeping Giants, whose successful pressure on advertisers to abandon alt-right outlets such as Breitbart won them international accolades as well as scrutiny for their role in legitimising brands as governors of public discourse (Braun, Coakley, & West, 2019). Campaigns like Sleeping Giants leverage advertisers’ social reputation to incentivise them to withdraw advertising money from publications featuring hate speech, misogyny, and other perceived social ills. In effect, they practice a form of public shaming intended to have a regulatory effect on the profitability of the content, with perhaps knock on effects on its production. While campaigns like Sleeping Giants are novel in pressuring online advertisers directly, businesses have been targets of public campaigns, both socially progressive and conservative, over content at least as far back as television in the 1980’s (Fahey, 1991; Turow, 1984). The sometimes cooperative, sometimes combative relationship between platform content moderation and advertiser interests continues in advertising industry-led initiatives like the GARM and activist-led campaigns like Stop Hate for Profit, and it is worth reconsidering these dynamics in the midst of a shifting regulatory environment that includes litigation against advertisers working together to influence content moderation, as well as legislative efforts to address content moderation and its effects directly.

Long considered “lawless” (Suzor, 2019), platforms are governed by overlapping commitments and structures, including voluntary agreements, codes of conduct, working groups, and other informal arrangements (Gorwa, 2019). Normative values and public opinion also contribute to this network of formal and informal governance mechanisms (Ananny & Gillespie, 2016; Barrett & Kreiss, 2019). Within this governance framework, platform terms of service nominally ban hate speech and other problematic content. However, content that violates platform terms of service, including hate speech, continues to spread widely (Giansiracusa, 2021) and platform structures support the circulation of violent and hateful content (Gerbaudo, 2018; Hopster, 2020; Mirrlees, 2021b), with key figures benefitting from dynamics between social media and legacy media to reach their audiences (McKelvey et al. 2021; Phillips, 2018). Persistent challenges with content moderation and revelations about platforms’ limitations have motivated state-led regulatory responses, in what some scholars have described as a “regulatory turn” (Flew, 2022). In the US, Section 230 nominally shields platforms from liability and allows them to moderate on a discretionary basis, though recent antitrust lawsuits challenge platform power in other ways. In democratic contexts outside the US, platform regulation is changing rapidly with regulation in Europe, the passage of the Online Safety Act in the UK, and required payment for news content in Australia and Canada. Most notably for this article, the passage of the Digital Services Act in the EU in 2022 has created stronger legal requirements for liability and content moderation, as well as requirements for due diligence over illegal content and risk assessments for the largest platforms. While these state-led efforts clearly shift the platform landscape further from lawlessness, it is worth considering if and how a more regulated internet disrupts campaigns that leverage advertiser’s financial power to influence content moderation below the level of the state.

Methods and data collection

This is an empirically grounded comparative account of two cases that share an understanding of advertisers’ potentially powerful relationship with platform content moderation, but act on that understanding in different ways. The case studies of Stop Hate for Profit and the GARM presented for this article are constructed based on analysis of over 100 contemporary campaign documents, media coverage, and public interviews, complemented by eight interviews with advertisers, activists, and businesses associated with one or both campaigns. The materials were chosen intentionally to provide insight into the context of these groups’ formation; the membership of the groups; their demands and intentions for content moderation, and their relationship to states and other potential regulatory actors. Documents collected included newsletters, social media posts, and press releases by advertisers that joined the Stop Hate for Profit boycott; public interviews and blog posts by the civil society groups that organised the boycott; Facebook’s blog posts, US Security Exchange (SEC) filings, public interviews about the SHfP boycott; and publications and interviews with the GARM. For a complete list of documents examined, not all of which are cited in this article, see Appendix 1. Interview participants for this project included three companies that participated in the Stop Hate for Profit campaign; one of the civil society organisations that led the Stop Hate for Profit campaign; two prominent activists working with the advertising industry to defund hateful online speech; and two representatives of advertising industry associations familiar with the industry’s relationship with Facebook and other social media (see the table of interview participants below). Semi-structured research interviews were conducted between January and August 2020 and participants were asked about the environment of corporate social advocacy, the role of communication technology, the motivations for corporate social advocacy and, where applicable, about their involvement in Stop Hate for Profit and the GARM.

Both cases explored here focus on Facebook, to some degree. However, the GARM and its members worked with many of the largest advertising-supported social media platforms, including YouTube and Twitter (now X). The pressure campaign that Stop Hate for Profit ran against Facebook was modeled after global efforts led by Sleeping Giants affiliates to pressure advertisers to break ties with publications that feature hate speech. Efforts to use the financial ties between platforms and major advertisers to change content moderation and other issues online continue, including in voluntary industry groups such as the Conscious Advertising Network, and activists continue to expose ties between advertising money and objectionable content on many platforms. So, while the cases presented in this article are no longer active, the strategies that they represent continue to be relevant as examples of content moderation efforts beyond the state.

Case 1: Global Alliance for Responsible Media

The World Federation of Advertisers (WFA) announced the Global Alliance for Responsible Media—a coalition that included YouTube, Twitter (now X), Facebook, and Instagram alongside a collection of the world’s largest advertisers—in 2019 as a response to proliferating concerns over platform content and advertising placement. In September 2020, the GARM announced a “brand safety floor + suitability framework” that provided agreed-upon standards for defining content as “harmful and sensitive,” defining where ads should not appear, and labelling sensitive content on a scale of risk that advertisers could choose between (The Global Alliance for Responsible Media, 2020). This brand safety floor structured annual transparency reporting that platforms submitted to the GARM and has been described by researchers as part of “the ongoing expansion and institutionalisation of advertiser influence” over content moderation, with outsized consequences for the moderation of content from already marginalised groups (Griffin, 2023, p. 9).

For years before the GARM was formed, the advertising industry had been grappling with the reputational and value-for-money risks of their relationship with platforms, including through limited boycotts by individual or small groups of advertisers. After shootings at two mosques in Christchurch were livestreamed on Facebook, New Zealand advertisers boycotted the platform and convinced the WFA to call for “all brands globally to hold social media platforms to account in light of recent failures to block dangerous and hateful content” (World Federation of Advertisers, 2020). The WFA subsequently launched the Global Alliance for Responsible Media (GARM) to facilitate more coordination between advertisers, agencies, and platforms. This alliance brought major advertisers and ad agencies together with platforms to find consensus on issues of responsible social media governance—including ad fraud and “strict brand safety protection” (World Federation of Advertisers, n.d.). The GARM and the WFA are led by companies with very established public reputations. As a representative of the WFA explained, “we represent companies that already have a very developed and sophisticated position on their role in society, CSR, those types of responsibility issues” (personal communication). The WFA had 127 corporate members and 55 national advertising association members. Their clients included historic leaders in the industry, such as P&G and Unilever, as well as newcomers, such as Airbnb. WFA member advertisers represented a small number of Facebook’s millions of advertising partners, “…of all the companies that are advertising on Facebook and YouTube—our members are a percentage of that, but they’re not quite as significant as they would be if you looked at TV channels” (personal communication, representative of international advertising association). The GARM represents the interests and viewpoints of established brands in their relationship with media, meaning that while the group includes very large and very well known brands they do not necessarily represent the majority of platforms’ earnings. For context, in a 2019 earnings call then-Facebook COO Sheryl Sandberg claimed that the top 100 advertisers on the platform represented less than 20% of their total revenue (Facebook Inc., 2019).

The GARM “was formed to identify specific collaborative actions, processes and protocols for protecting consumers and brands from safety issues” (World Federation of Advertisers, 2020) and close the gap between social media platforms’ hands off approach to content moderation and the social responsibility commitments of major advertisers. The GARM developed a Global Media Charter as well as the “Brand Safety Floor + Suitability Framework” to outline its priorities. The Global Media Charter’s eight principles outlined quality control issues for the online advertising industry, including user experiences of advertising, ad-fraud, and a tiered approach to brand safety risk. It emphasised addressing “walled garden issues” and third-party verification, referencing the inability of most advertisers to independently verify the numbers provided to them by private platforms. This emphasis echoes the widespread unease among advertisers about the ability of Facebook and other platforms to “grade its own homework” in terms of ad performance, viewership metrics, and content quality (personal communication; Monllos, 2020).The Brand Safety Floor + Suitability Framework addressed a range of topics, beginning with “adult & explicit sexual content” and including arms and ammunition, crime, death and injury, piracy, hate speech, obscenity, drugs, spam, terrorism, and “debated sensitive social issue” (Global Alliance for Responsible Media, 2020), in that order. The GARM’s Global Media Charter incentivised over-removal of content. For example, while informative sexual content was rated as “low-risk,” the suitability of platforms for advertising was ranked by the prevalence of “harmful” ad impressions and the effectiveness of platforms in removing “violative content” (World Federation of Advertisers, 2021), providing an incentive for platforms to err on the side of caution by removing sexual content of many kinds. The GARM existed to address issues of verification and value for money alongside an approach to content moderation that incentivised risk averse moderation likely to over-remove content that did not violate the standards directly.

National governments and multilateral institutions were absent from the discussion of the establishment of the GARM. Advertisers were aware of the lack of regulatory checks and balances and some identified it as a challenge to the legitimacy of the industry’s self-regulation: “[self-regulation] only works if we actually act responsibly and deal with any issues. [Platforms] talk about control but I don't think there's a high level of trust around that. And it's not independent, it's internal… there's a vacuum there, which then says government regulators need to step in” (representative, national advertising association). The GARM’s later willingness to engage directly with EU regulatory processes by, for example, signing the EU Commission’s Code of Practice on Disinformation, may reflect this openness to state interventions to solve content moderation challenges. While it was active, there were doubts about whether platforms were fully complying with the GARM’s brand safety standards (Kaye, 2021). However, the implications of reinforcing content moderation in line with standards of brand safety was interpreted by scholars as potentially significant (Bishop, 2021; Griffin, 2023). Regardless, the GARM abruptly suspended its activities in 2024 after Elon Musk sued the organisation, claiming it was engaged in anti-competitive collusion over content (O’Reilly, 2024). It is unclear whether its efforts to formalise advertiser interests in platform content moderation will be revived.

Case 2: Stop Hate for Profit (activist-initiated effort at moderation via advertisers)

The year after the formation of the GARM, during the Black Lives Matter protests in the summer of 2020, the Stop Hate for Profit campaign recruited advertisers to pressure Facebook over perceived gaps in how its content moderation policies applied to hate groups and inadequate support for targets of hate and harassment on the platform. The Stop Hate for Profit advertiser boycott happened one month after George Floyd’s murder by police in Minneapolis on 25 May 2020. The boycott targeted Facebook in part over differences between its moderation of inflammatory statements by Donald Trump and those of other platforms. Later investigations uncovered that Mark Zuckerberg had phoned Trump directly to discuss the content of his posts (Mac & Silverman, 2020). Facebook’s willingness to bend its own rules on what was widely viewed as support for violence against protestors put them in conflict with some of the platform’s advertisers that were facing pressure to prove that they understood the gravity of the situation and were taking meaningful action (personal communication, Stop Hate for Profit participant).

In this context, a coalition of American civil rights organisations, including the Anti-Defamation League, the NAACP, Color of Change, the National Hispanic Media Coalition and the online activist group Sleeping Giants organised the Stop Hate for Profit campaign. Stop Hate for Profit asked participants to suspend advertising on Facebook in July 2020. Stop Hate for Profit called on advertisers to suspend advertising on Facebook as a symbolic gesture against Facebook’s role in enabling or amplifying “hate, bigotry, racism, antisemitism, and disinformation” (Stop Hate for Profit, n.d.). The group’s nine-point list of demands focused on the themes of accountability, decency, and support for victims of hate and harassment. It emphasised civic solidarity with people and groups harmed by harassment, Facebook’s lack of accountability to these same groups in its policies, and Facebook’s failure to take responsibility for content on its platform. In contrast to the GARM’s effort to align advertiser interests with platform practices, SHfP was much more narrowly focused on changing content moderation policies related to race and enforcement against hate speech.

At its height, the Stop Hate for Profit campaign included 1,100 advertisers, including 200 “major corporations” whose advertising value to Facebook was worth approximately USD $7 billion (Levy, 2020). Speaking to the incentive for advertisers to intervene in controversies over advertising-supported content, representatives of the advertising industry noted “an external perception that advertisers should be the ones able to force change” (personal communication, national advertising association). In particular, the investment that advertisers had made in social responsibility acted as an incentive. As a representative of a national advertising association communicated “the question becomes, if you are talking about brand purpose, CSR [corporate social responsibility], etc, what is your position around where your advertising may appear?” Moreover, “silence becomes a void that people can fill” (personal communication, small ad agency), opening companies up to accusations that they did not care about racial justice or other social issues if they were not publicly stating their support through participation in initiatives like SHfP. Many companies felt pressure internally from employees to demonstrate their commitment at a significant moment, particularly those that had been “late” in responding to the protests in May (personal communication, Stop Hate for Profit participant). Others had a history of support for issues of representation and diversity and felt that the campaign was “something doable” (personal communication, small ad agency) in terms of pressuring other small businesses and demonstrating leadership. The investments made by these companies in their reputations for social responsibility created internal incentives to push back against a platform framed as accommodating extremist rhetoric.

However, the campaign’s significance was soon subject to critical scrutiny. Facebook’s revenue was over USD $21 billion over the quarter and did not dip while the boycott was active (Glenday, 2020). Some companies, such as Unilever, suspended advertising on the platform for all of 2020 independent of the SHfP campaign, citing the “toxic environment” online during the US presidential elections (McCarthy, 2020). Others, including P&G, whose year-long boycott of YouTube in 2017 was credited with an instrumental role in changes to the platform (Bergen, 2018), did not participate in the SHfP boycott (Cavale, 2020). Advertisers participating in the campaign felt a swift change in media interest from positive coverage of advertiser participation “to a kind of nitpicky, looking for gaps kind of thing” (personal communication, Stop Hate for Profit participant) that scrutinised the timing of the campaign, their use of other advertising tools linked to Facebook, and the amount advertising spending was being cut. Some industry activists saw the campaign as disingenuous. As one industry activist told me, “What I saw with Stop Hate for Profit was them bragging about having gotten companies to stop advertising. But those companies hadn’t really stopped advertising through Facebook” (personal communication, consumer activism organisation). Widely circulated comments such as Mark Zuckerberg expecting advertisers to be “back soon enough” (Heath, 2020) and Stop Hate for Profit’s characterisation of Facebook as offering “the same old defense of white supremacist, antisemitic, Islamophobic and other hateful groups on Facebook” seemed to reflect the frustration of the boycott’s short-term demands. Observers deemed participation in the campaign for most companies as “at best, symbolic, and unlikely to harm Facebook financially” (Wodinsky, 2020, para. 7).

However, changes to content moderation in response to Stop Hate for Profit nominally included the removal of “content encouraging or calling for the harassment of others, which was a top concern of civil rights activists” and the removal of “more than 100,000 pieces of FB & IG content for violating voter interference policies” (Sharing Our Actions on Stopping Hate, 2020). Facebook’s responses to the boycott also echoed the language used by the campaign: “We do not profit from misinformation or hate, and we do not want this content on our platforms” (Mark Zuckerberg, in Facebook Inc’s Second Quarter Results Conference Call, 2020). In the end, the campaign claimed victory: “we forced an unprecedented public examination of Facebook’s deep harms to marginalized communities” (Statement from Stop Hate for Profit on Ad Pause Success, 2020). The campaign included the creation of a senior civil rights position at Facebook, the release of an unflattering civil rights audit, the creation of a team to study algorithmic bias, the promise of an independent audit, and the removal of extremist groups such as Boogaloo as evidence of the success of the initiative (Statement from Stop Hate for Profit on Ad Pause Success, 2020). Facebook did commit to changes in content moderation, such as the introduction of warning labels on contentious content (Bond, 2020), banning holocaust denial, and efforts to reduce the impacts of its hate speech policies on marginalised groups (Dwoskin et al., 2020; Wodinsky, 2020). The boycott, as contested and symbolic as it was, added to public criticisms of Facebook’s content moderation policies that motivated changes to the platform. It was also a reversal of decades of advertiser interventions that primarily restricted advertising on issues that served the self-interest of the advertisers, rather than addressing social issues (Hardy, 2023).

The GARM, and the commitments of the parties that signed onto it, became a touchstone for Facebook when disputing some of the claims of the Stop Hate for Profit campaign. Facebook’s VP of Global Marketing Solutions referenced commitment to GARM in their messages to advertisers and the GARM appeared in Facebook’s July 30 update to its public statement on the Stop Hate for Profit campaign, which addressed the campaign’s claims and demands in a point-by-point format (Sharing Our Actions on Stopping Hate, 2020; Fischer, 2020). At the time, Facebook’s CEO claimed that the platform was “not going to change our policies or approach on anything because of a threat to a small percent of our revenue” (Mark Zuckerberg, in Clayton, 2020, para. 4). Instead, it was making sure that those policies “support the advertising community through the Global Alliance for Responsible Media (GARM) and continue to engage with civil rights leaders about our policies and practices” (Sharing Our Actions on Stopping Hate, 2020, Stopping Hate section). In this case, Facebook used its membership in the GARM as evidence that it was aligned with advertiser standards of responsible content moderation without committing the platform to actually meet the demands of the Stop Hate for Profit boycott participants. National governments and multilateral institutions were absent from the discussions between Facebook and Stop Hate for Profit. While Facebook referred to multi-stakeholder agreements such as The Christchurch Call to Action, a set of voluntary commitments agreed to by governments and online service providers intended to address terrorism after the shootings in Christchurch, there was little suggestion of another credible authority outside of Facebook and its advertisers

Discussion

The cases of the GARM and the Stop Hate for Profit campaign demonstrate two different efforts to influence the scope and focus of social media content moderation through advertisers and the advertising industry. The first is advertiser activism, often pressured by civil society and the public, as in the Stop hate for Profit campaign. The other is the institutional approach–advertiser standards enacted by the industry, as in the GARM. These two approaches are analogous to the “instrumental” and “structural” understandings of advertiser influence over media in critical political economy, with one representing intentional interventions over content and the other the aggregate effects of the industry’s practices and preferences (Hardy, 2023). While the original Stop Hate for Profit campaign was relatively short-lived, working through advertisers and the advertising industry to incentivise responsible content moderation on platforms remains an active strategy for now. Adtech-focused watchdog Check My Ads declared the end of GARM as an opportunity to “reset” brand safety efforts from advertisers and marketers around fundamental reputational concerns, rather than abandon those efforts (Garcia, 2024). The institutional approach works to craft longer term commitments that can be supported by advertisers, instituted by platforms, and potentially supported by other institutions. This approach continues with groups such as the Conscious Advertising Network, a coalition of advertisers that sign onto shared “manifestos” on issues from ad fraud to climate misinformation that is hosted regularly by the United Nations (Hobbs, 2020). The power of these institutional connections may be diluted if changes to content moderation on major platforms continue to encourage advertisers to reduce their spending on those platforms (Ostwal, 2024). Both the activist approach of individual advertisers and the institutional approach are highly relevant to considering content moderation beyond the state as they operate either as bilateral (advertiser-social media platform) or tri-lateral (advertiser-platform-civil society) governance negotiations, without a necessary role for state regulators and policymakers. They exemplify new complexities in how advertisers influence content moderation that complicate accounts that position advertisers as censors of the media, particularly media that serves marginalised groups (Hardy, 2023). Both strategies will be sorely tested by rapid changes in content moderation on large platforms, particularly on X and Meta which have made definitions of hateful conduct more permissive (Bradley & Joseph, 2025; Vranica & Haggin, 2025). The complexity around advertiser approaches to content moderation, and the questions for policy intervention, is most evident in the two main groups marginalised by the advertiser interventions in the GARM and SHfP. Some are marginalised for their participation in the kinds of misinformation and hate speech that the campaigns were created to combat, others were caught up by the vague definitions of risk employed by advertisers’ standards and platforms’ implementation.

Consequences, implications, and the regulatory turn

Two main groups oppose advertiser influence on platform content moderation: those who feel targeted by advertiser activism and those who feel marginalised by advertiser standards. Stop Hate for Profit drove an exodus of alt-right content creators from platforms like Facebook to alt-tech platforms, such as Gab, that market themselves in opposition to the content moderation of mainstream social media platforms, which they characterise as hostile to free speech and conservative viewpoints (Jasser, 2021). The content moderation policies enacted by social media platforms in response to advertiser activism resulted in the removal of hundreds of millions of pieces of hateful content on Facebook, YouTube, Twitter, and other social media sites, including many associated with alt-right groups and individuals (Mirrlees, 2021b; Renton, 2021). More generally, social media platforms, based on the policies that the GARM institutionalised, remove hate speech, violent rhetoric, misinformation, and insensitive treatment of debated topics. Increasingly, figures within the alt-right movement complain that “there’s no economic model” (Steve Bannon in Embury-Dennis, 2019) based on advertising for sites such as Gab, Breitbart, Parler, or for alt-right celebrities, such as Alex Jones or Milo Yiannopoulos. As others have noted, “Big Tech’s hate speech rules have thrown a monkey wrench into the social media machinery of some far Right’s propagandist-entrepreneurs and denied them some platforms to make money, mobilize, and recruit in the digital mainstream” (Mirrlees, 2021a, p. 270). At the same time, conservative lawmakers have begun to push back against these efforts. In some cases, politicians accuse the advertisers of collusion, as in the case of the chair of the US Judiciary Committee discussed below. In others, there is political pressure to exclude activist advertisers from policy processes, as in efforts to remove the Conscious Advertising Network from consultations with the United Kingdom Department of Media Culture and Sport (Young, 2023). These efforts reached a new level of prominence and effectiveness with Musk’s takeover of Twitter (now renamed X). As advertisers left the platform over concerns about content and changes to policy, including significant reductions to the trust and safety team (Mac, 2024; Thomas, 2024), Musk publicly told them to “go fuck yourself” (Goode, 2023). Subsequently the company sued the GARM, the World Federation of Advertisers, and some of their most prominent supporters, accusing them of antitrust violations, continuing a pattern of lawsuits intended to intimidate advertisers and activists from taking action against the platform (Ortiz, 2024).

In a very different political context, many researchers argue that advertiser interventions further marginalise content created by and for women, sexual expression, and sexual and gender minorities, as well as journalism and activist content (Aronczyk, 2020; Bishop, 2021; Burgess & Green, 2018; Cotter, et al., 2021; Griffin, 2023; Mueller, 2015; Siapera & Viejo-Otero, 2021). For example, when Tumblr tried to make the platform more attractive to advertisers by banning adult content, it disrupted forums that had fostered LGBTQ communities, and attracted public contestation of the platform’s governance practices by its users (Sybert, 2021). Tumblr’s shifts in governance cost the platform much of its user base, but also destroyed what had been an important forum for marginalised sexual communities, including trans people and sex workers (Bronstein, 2020). Using keywords and algorithms in online content moderation increases the likelihood that non-target content will be removed or demonetised (see Urban et al., 2016 for a discussion of this trend in response to legal rulings). The GARM and its standards exemplifies these concerns (Griffin, 2023). That Facebook cited the GARM as a rebuttal to demands for solidarity with the aims of Black Lives Matter activists seemed to confirm the industry’s institutional role in setting a risk-averse and relatively conservative approach to content moderation through their definitions of brand safety and suitability.

These are not equivalent claims, but do share a view of advertiser power over platforms. Both groups see the effects of advertiser interventions as at best incentivising the platforms against specific groups and their content and at worst practicing a form of “collateral censorship” (Balkin, 2009) where advertiser pressure, rather than legal liability, on platforms restricts discussion of certain topics. Reactions from groups that see themselves as disadvantaged by advertiser interventions in content moderation include abandoning platforms perceived to be hostile, as in the case of some of the alt-right, and contestation of platform policies, as in the Stop Hate for Profit campaign and the Tumblr example. However, state-led regulatory developments, particularly in the US and EU, raise questions about whether advertiser influence over content might be constrained. The next section considers whether new regulations, in particular the EU’s Digital Services Act, interrupt or diminish the role of advertisers as informal content regulators in social media environments. It also considers the rhetoric of antitrust used to challenge advertiser cooperation on content moderation, including the lawsuit filed by Elon Musk against the GARM that ended the organisation’s activities.

Interventions by state and regional regulators

Neither the GARM nor the Stop Hate for Profit referred to the state as a solution to their content moderation challenges. The relative absence of states in the regulation of platforms is changing, however, with notable antitrust and anti-competition cases brought in the United States, the passage of online safety legislation in the United Kingdom, proposed online harms legislation in Canada, and, most notably, the passage of the Digital Services Act and Digital Markets Act in the European Union.

Advertisers and content moderation after gatekeeper legislation

The Digital Services Act

The Digital Services Act promises to significantly change how intermediary liability functions in the EU, incorporating considerations of fundamental rights into platform’s moderation (Heldt, 2022). For this article, the question is really whether this legislation will disrupt the direct institutionalisation of advertiser interests in content moderation, or campaigns like SHfP that try to leverage that relationship. The DSA obligates platforms to take fundamental rights into account in the application of their terms and conditions (Art. 14(2)). Very large online platforms and search engines have additional obligations to assess risks related to potentially dangerous content, discrimination, and threats to fundamental rights under Article 34(2)(d). These obligations include consideration of content moderation systems and systems for placing advertisements. The fundamental rights including human dignity, respect for private and family life, personal data, freedom of expression and information, non-discrimination, the rights of the child, and consumer protection (Art. 34(1)(b)). Theoretically, these obligations could encourage platforms to consider the fundamental rights of individuals alongside advertiser priorities in content moderation. They could also reinforce legal efforts charging platforms with discrimination based on monetisation policies and recommendation, strategies that have been unsuccessful so far (Henn, 2021; Stempel, 2023).

Elsewhere, the DSA and related legislation encourages platforms to work with advertisers to govern transparency in advertising placement, ad targeting, and the monetisation of misinformation. Where the DSA explicitly deals with advertisers and advertisements, it is primarily concerned with requiring transparency around who purchases advertisements and how data is used to target individuals (Articles 26, 39, 46). Much of this work is expected to take place through voluntary codes of conduct that stakeholders in the platform advertising chain will sign onto. Some potential codes of conduct, such as the EU Code of Practice on Disinformation, explicitly identify brand safety tools as important levers in preventing monetisation of disinformation and misinformation. The Code of Practice may soon be added to the DSA, but platform compliance with the Code is largely poor, adding to concerns that platform compliance with the DSA may be lacking (Mündges & Park, 2024). Platforms’ history of inconsistent compliance, sometimes even with their own policies (Horwitz, 2021), suggests that the DSA may not significantly shift the balance of power in content moderation.

Griffin (2023) argues that the DSA/DMA is unlikely to disrupt advertiser impacts on content, given that the two pieces of legislation do not address the indirect effects of advertiser priorities or platform policies themselves, only their application. She adds that the Code of Practice on Disinformation is “effectively legitimising this [advertiser] censorship power and charging them with using it responsibly” (Griffin, 2023, p. 18). However, there is an argument to be made that the DSA’s risk assessment provisions in Article 34 should cover advertiser impacts on content moderation and support legal challenges if advertiser standards can be shown to consistently impact freedom of expression, though this will take time. There may also be good reasons to distinguish between censorship, with its connotations of restrictions on predefined types of political speech, and commercial constraints on online platforms, which play a more complex role in algorithmic media, as well as working in multiple directions, politically speaking (Hardy, 2023). Content moderation in the name of safety and anti-discrimination has become politicised and subject to attack from policymakers and powerful individuals in the United States where, in contrast to the DSA’s theorised effects, antitrust lawsuits have already forced the suspension of advertisers’ formal content moderation efforts, including the GARM itself.

Antitrust, collusion, and backlash

In March 2023, the chairman of the US Committee on the Judiciary accused both the Global Alliance for Responsible Media (GARM) and the World Federation of Advertisers, of “potentially violating US antitrust law by coordinating their members’ efforts to demonetize and eliminate disfavored content online” (Jordan & Nadler, 2023). Jordan and Nadler’s accusations were used as the basis for a lawsuit filed by X in US federal court against the GARM, World Federation of Advertisers, and several of their highest-profile members for allegedly engaging in antitrust activity (Scanlon, 2024). In response, the WFA chose to suspend the GARM’s activities (Joseph & Scanlon, 2024). Observers doubt the legal merit of the case, and suggest that these cases are political (Elon Musk can’t force advertisers to spend, 2024) and continue a pattern of pushback against advertisers restricting ad placement over media content. Conservative media outlets have doxxed activists involved in Sleeping Giants (Greene, 2018). Earlier legal efforts by X/Twitter against activists and advertisers that were dismissed as Strategic Lawsuits Against Public Participation (SLAPP) meant to discourage reporting on the scale of hate speech and misinformation on X (Ortiz, 2024).

Participation in the GARM did not require advertisers to reduce advertising on X. Many jurisdictions, including the EU, protect horizontal cooperation between market actors when it serves a legitimate purpose, such as sustainability ends, and does not disrupt competition (Antonazzi, Kuiper, & Cramer, 2023). The GARM acted as another voluntary code of conduct, like many other voluntary agreements that govern platform content moderation beyond the state (Gorwa, 2019). The risks that it posed were not collusion to demonetise disfavoured content online, but of economic and public relations incentives that shored up existing platform content moderation trends that favour professional, advertiser-friendly content to the detriment of small producers, vulnerable groups, and educational and news services. By suing the GARM out of activity, X has not eliminated the indirect effects of an advertiser funded system, existing advertiser tools that platforms provide, the monetisation policies that came from codifying advertiser preferences under the GARM, or even strategies like Stop Hate for Profit that recruited individual advertisers. However, the case, and others like it, especially if supported by the second Trump administration, may have a chilling effect on efforts to challenge injustices in content moderation and discourage compliance with efforts to demonetise disinformation and hate speech. This is already visible in dramatic shifts to Meta’s content moderation policies (Vranica & Haggin, 2025). The use of legal intimidation may undercut efforts to engage in robust public deliberation over content moderation, including contestation of advertisers’ brand safety standards and their effects on marginalised groups.

Conclusion: Zones of discretion

While the Digital Service Act may well increase platform transparency, it is not clear that it will disrupt the content moderation dynamics taking place between platforms and advertisers. As Nic Suzor and his collaborators in research on digital constitutionalism have pointed out, even with increased nation-level regulation of platforms, a considerable amount of platform content moderation will be left to the discretion of the platforms themselves (Suzor, 2020; Suzor & Gillett, 2022). These persistent “zones of discretion” (Suzor & Gillett, 2022, p.261) mean that self-regulation, informed by profit motives, will be with us as long as platforms exercise their “custodial” powers over online content (Gillespie, 2018) and as long as they are constructed around an economic model tied to advertising. Regulation that acknowledges the role of platforms’ economic model in issues of content moderation, as the Digital Services Act does, has the potential to push back against overreach by those standards when it affects fundamental rights. This approach is preferable to one that characterises efforts to demonetise misinformation and hate speech as anti-competitive collusion or censorship and potentially chills efforts to exercise content moderation in the name of safety and fairness.