Combatting disinformation with crisis communication: An analysis of Meta’s newsroom stories

Social media and disinformation

Based on recent events in which fake news has received a lot of media attention, such as the Covid-19 pandemic and the accompanying vaccination campaigns (e. g., Rocha et al., 2021), and Russia’s war in Ukraine (e. g., Kreft et al., 2023), one might get the impression that fake news is a recent phenomenon, but this is not the case. It has been around for much longer and has likely existed as long as mankind (see, e. g., Cortada and Aspray, 2019 for an overview). But it does not seem wrong to say that, since 2016, the concept of fake news has been undergoing a kind of revival, and has attracted a great deal of interest from academia, politics, and the news media. One could argue that an article published by the online-only news medium Buzzfeed on 20 October 2016 was the start of the renewed attention to the concept of fake news. That day, the news site published an article about its own research showing that, in the context of the 2016 US presidential campaign, hyperpartisan Facebook pages were posting false and misleading information on a large scale (Silverman et al., 2016). Certainly, when Trump became elected president a few weeks later, the debate about the role that Facebook – and by extension social media – play in the spread and success of fake news gained traction more or less all over the world. For example, 2016 and 2017 saw a spike in mainstream media coverage on fake news, and the volume of articles was incomparably higher than it was in the 35 years before (Al-Rawi, 2019). Google has also recorded an immense increase in online posts containing the term “fake news” since November 2016 (Kapantai et al., 2021). Since then, there has been a great deal of attention from academia to what has been labeled with an umbrella term “fake news,” both for deliberately false information with the aim of misleading people, better known as disinformation, and for false misinformation that has little or no intention of misleading people (see, e. g., Kapantai et al., 2021; Tandoc, Lim, and Ling, 2018 for a classification of the multitude of concepts). Moreover, the focus was no longer exclusively on political-ideological misreporting and began to shift to other topics such as health (e. g., Grimes, 2021) and climate change (e. g., Lutzke et al., 2019), but almost always in the context of social media as a breeding ground for this type of misinformation.

To explain the popularity and rapid spread of disinformation on social media platforms, the literature often refers to the algorithms that give certain posts extra visibility by injecting them into users’ timelines as, for example, a trending topic. In doing so, the algorithms prefer topics that show a sudden, steep spike in volume over topics that may have been discussed on social media for some time but without a spike in volume (Poell and van Dijck, 2014). This preference plays into the hands of spectacular, breaking, and even toxic news (e. g., Massanari, 2017) that is suddenly fiercely discussed within social media, and therefore elicits a rapid emergence and wide reach of bizarre messages containing disinformation. In addition, Stöcker (2019) refers to the monetization goals of social media platforms to capitalize on those types of metrics that favor watch time and engagement over relevance, and therefore provide a perfect ecosystem for targeted disinformation.

The above shows the role which the users of social media play, since they are primarily responsible for the engagement with certain topics. That it is not merely the algorithms but rather the users themselves whose behavior is reinforced by the algorithms is evident from a great deal of research on disinformation and the role that echo chambers play in this regard (see, e. g., Zimmer et al., 2019). There is a significant number of users who forward dubious news to others within their networks, even when they know the information is incorrect (Chadwick et al., 2018; Metzger et al., 2021). In short, social media provide a good platform for the rapid rise and spread of disinformation, and this has led to studies that zoomed in on disinformation on specific platforms such as Facebook (e. g., Buchanan and Benson, 2019), Twitter (Keller et al., 2020), Instagram (Mena et al., 2020), and TikTok (Basch et al., 2021).

Disinformation as a crisis with multiple stakeholders

Based on the above, we can state that social media platforms have been in a disinformation crisis since 2016. Coombs (2007, pp. 2–3) defines a crisis as “the perception of an unpredictable event that threatens important expectancies of stakeholders and can seriously impact an organization’s performance and generate negative outcomes.” And these characteristics of a crisis also apply to social media platforms in the context of disinformation. After all, we know that public perception is not good, as evidenced by the many negative media reports about social media as the place where disinformation is created and shared, as well as the fact that the CEOs of Facebook and Twitter had to defend themselves before the US Congress in 2021 (Fung, 2021).

Just as most crises have an impact on a variety of stakeholders (Heath and Coombs, 2006), the crisis surrounding disinformation on social media is also one that affects various stakeholders of social media platforms. If we look at Meta’s internal stakeholders, we see mainly its own employees and shareholders. Employees could become less proud of the company and possibly want to change jobs due to the disinformation crisis and criticism of Meta’s handling of the problem. Shareholders in turn could lose faith in the company’s value. In terms of external stakeholders, users are the most important. Several studies show that social media users are concerned about disinformation they encounter on those platforms. For example, in the context of disinformation on Covid-19, users of Facebook state that they encounter more disinformation on the platform than on news sites or through search engines (Newman et al., 2021), and the same seems true for users of Instagram (Jurkowitz and Mitchell, 2020). Then there are the advertisers who use the platforms to spread their messages, sell products through sponsored posts and ads, and are the main source of revenue for platforms like Facebook. The presence of disinformation on the platforms and the negative discourse around it could be a reason for companies to advertise less or not at all on these platforms. For example, in June 2020, the CEO of Levi Strauss and Co stated via a blog post that she was not satisfied with the way Facebook and Instagram addressed misinformation and announced that she would pause ad spending on both platforms for a while (Sey, 2020). Governments and politicians are also an important external stakeholder. On the one hand, they use the platforms to spread their messages (Peeters et al., 2022), but, on the other hand, they are the driving force behind initiatives to regulate the platforms through national and supra-national regulations and laws. Examples of this type of law include the 2018 Code of Practice on Disinformation and its 2022 enhancement (European Commission, 2022) and the DSA approved by EU member states in the European Parliament in 2022, which aims to better protect the users of digital platforms by holding the platforms accountable for illegal and harmful content (European Commission, 2023). This can lead to a dual stance where political actors need the platforms to spread their ideology and get advice from Facebook employees on how best to do so (Kreiss and McGregor, 2018), but at the same time are the ones to criticize and regulate the platforms through guidelines and laws. Next, the news media are important stakeholders since negative coverage on disinformation on social media platforms often reaches the public through those news media, like the Buzzfeed article mentioned in the introduction, and the numerous news articles that followed (see, e. g., Al-Rawi, 2019). Finally, even competitors are relevant stakeholders (Freeman, 2015), in the sense that other social media platforms such as Twitter and TikTok are also facing a disinformation crisis so that the communication Meta sends out to the world can turn out positively or negatively for them as well.

Crisis communication as impression management

It may be clear from the above that social media platforms are engaged in a crisis of disinformation, and that this is of concern to a variety of stakeholders dealing with those platforms. Crises can be seen as violations of stakeholder expectations, “meaning an organization does something stakeholders do not think it should have done” (Coombs, 2018, p. 52). To respond to this crisis, it makes sense for Meta to communicate about it, which we therefore logically consider in this paper as crisis communication. Crisis communication can be defined as the practice of minimizing the negative effects or damage of a crisis on the company’s reputation or loss of revenue (Williams and Treadeway, 1992) and as the strategic use of words and actions to manage information and meaning during a crisis process (Coombs, 2010). It is about managing information (collecting and disseminating crisis information) and managing meaning (influencing how the stakeholders perceive the crisis and/or the organization in crisis) (Coombs, 2018). Crisis communication can even be used to use a crisis in such a way that it enhances a company’s reputation, as the company can appear decisive in its response to the crisis to the various stakeholders (Sellnow et al., 2022). We therefore follow the premise that research on crises and associated crisis communication should focus on the different stakeholders where, according to stakeholder theory (Freeman et al., 2010), one must strike a good balance between all these stakeholders with sometimes conflicting interests.

A great deal of research has been done on crisis communication, from which a number of important theories and insights emerge, even though we know that practitioners do not always translate them perfectly into practice (Claeys and Opgenhaffen, 2016). While it is true that every crisis is unique and requires its own crisis communication (Coombs, 2015), we can draw insights from the numerous studies and the existing models and theories such as Image Repair Theory (IRT; Benoit, 2005) and Situational Crisis Communication Theory (SCCT; Coombs, 2007), which can be seen as the two most influential theories in the field of crisis communication (Claeys and Opgenhaffen, 2016). Both theories propose several crisis-response strategies that can be used by crisis communicators to best restore the reputation of the company in crisis. IRT argues that the central goal of crisis communication should be positive corporate reputation. The theory suggests a number of concrete advices to achieve this, such as strategies to deny self-responsibility (such as “shift the blame,” i. e., put the blame on something outside the organization), avoid responsibility (by emphasizing good intentions, for example), reduce vulnerability (by using the strategy of “bolstering,” i. e., focusing on the company’s positive attributes or actions), or simply by admitting guilt and expressing regret) (see, e. g., Benoit, 2014 for an overview). The SCCT also recommends different crisis-response strategies, taking strongly into account the nature of the crisis and the perception of stakeholders, focusing on the acceptance of responsiveness, and offering apologies (see, e. g., Coombs, 2007), even if this is not always easy for the legal departments of companies (Claeys and Opgenhaffen, 2021). In this context, Hall’s (2020) study around the 2018 apology campaign launched by Facebook in the wake of the Cambridge Analytica scandal found that the company used the strategy of apology to create a divided perception of the company that strengthened the brand identity. In addition, there are numerous other studies that provide useful insights about the benefit of self-disclosing (elements of) a crisis, better known as stealing thunder, in order to maintain control over the communication (Claeys et al., 2016) or the importance of consistency in the message in order to come across as credible (Coombs, 2020; Seeger, 2006), which ties in with the strategic advice to focus on a few key messages in crisis communication (De Waele et al., 2020). These insights align nicely with what is known as impression management strategies (Leary and Kowalski, 1990), in which the communication efforts of an organization are aimed at restoring the legitimacy and thus the way different stakeholders view the company. Impression management attempts to respond to pressure from the different stakeholders through communication by emphasizing good news and minimizing bad news, known as “thematic manipulation” (Brennan and Merkl-Davies, 2013; Hellmann et al., 2020).

Because this study examines Meta’s crisis communication, it aligns with the study by Iosifidis and Nicoli (2020), who scrutinized Facebook’s communication of disinformation over a period between 2016 and 2019 to identify the company’s self-regulation. To our knowledge, this is the only study conducted on this topic. The authors argued that Facebook did not see itself as an arbiter of the truth and put forward AI and machine learning as tools to combat various forms of fake news. Our study tries to complement that previous study by also analyzing the period between 2019 and 2022, since a lot happened in that period regarding disinformation, such as the Covid-19 pandemic and the war in Ukraine. Moreover, we try to make the link with the insights from crisis communication and impression management by mapping the variety of different stakeholders and study the different strategies such as the use of key messages and taking responsibility. We formulate the following research question:

Variety of stakeholders

Based on the analysis of newsroom messages, Meta targeted a wide variety of stakeholders. This targeting rarely happened explicitly, addressing a specific group of stakeholders in the introduction, for example, but rather implicitly by pointing out throughout the text the interests of interventions for certain groups. We distinguished three main groups of stakeholders. The first group is its own employees. Throughout the stories there are many references to the good work of employees. There are explicit references to concrete job profiles within the company such as the developers, the moderators, people from the legal and operations departments, but just as regularly, in more general terms, there are references to “our teams” working in-house to fight disinformation. A second group are the external stakeholders, the individual users of the platform being described in the stories as users or members, but also as, for example, admins or owners of groups or pages. Other “positive” external stakeholders such as NGOs and health organizations (especially in the context of Covid-19), national and local governments (especially in messages about elections), academic partners, news organizations, and external fact-checkers are often referred to. Meta often refers to the term “community” in its communications to describe this heterogeneous group of stakeholders while distinguishing it from its own employees, for example: “We’ll continue to share changes we make to keep both our community and the people who review content on our platforms safe” (story 41 update 41, April 21, 2020, “Update on content review work”), or “We’re doing everything we can to keep our global teams and the community that uses our apps safe” (story 41 update 19, March 19, 2020, “Keeping our platform safe with remote and reduced content review”). Stakeholders that are – perhaps surprisingly – less often mentioned are the commercial companies that, as advertisers, still play an important role in the revenue model of the platforms. The few times they are referenced, they are clearly linked to the benefits these businesses can get from the platform, as in this reference in a story defending Facebook’s revenue model: “Small businesses (…) get access to tools that help them thrive. There are more than 90 million small businesses on Facebook, and they make up a large part of our business. Most couldn’t afford to buy TV ads or billboards but now have access to tools that only big companies could use before” (story 12, January 24, 2019, “Understanding Facebook’s business model”). We can label the third group as external, hostile stakeholders who may pose a threat. These include the creators and disseminators of harmful content. It is interesting to note that these are more or less the same stakeholders as in the second category, but they are described in a negative way or addressed critically throughout the communication. They are about (foreign) governments and political advertisers with bad intentions, individual spreaders of fake news, bad administrators of pages and groups, or news organizations that have written too negatively (and wrongly, according to Meta) about the company’s fight against disinformation. In other words, different stakeholders are sometimes considered good partners and members of the community while at other times they are just considered threats to the community. It is also striking that there is hardly any explicit reference to the authorities that are trying to regulate Meta and other social media players. For example, there is no mention of the Code of Practice on Disinformation, nor of the DSA. Only one reference to governmental regulation is made in a 2019 news story by Mark Zuckerberg himself, in which he refers to the European General Data Protection Regulation (GDPR) as a good thing for everyone.

Key messages about interventions

Important in crisis communication is the message proclaimed by which the organization communicates to stakeholders how to address the problem. The analysis of the stories allowed us to identify some important key messages that dealt with the strategies Meta deployed during the period of the study. During the first years (2016–2020), we can summarize the key messages as removing-reducing-informing, which are labels that Meta itself uses throughout its communication.

Meta declares its primary commitment to detecting and removing content that violates its policy. It stresses that for this it not only relies on manual efforts of its own moderators, but has also partnered with independent fact-checkers with whom it has worked since late 2016, and that this partnership has expanded in the years since. Meta very frequently uses the phrase “(outside) experts” (e. g., story 21, April 10, 2019, “Remove, reduce, inform: New steps to manage problematic content”) to refer to the fact-checkers, seemingly wanting to highlight the legitimacy and independence of that group. Specifically in the context of disinformation about Covid-19 and climate change, the company refers to scientists and academics it works with to uncover erroneous reports. It also refers several times to the fact that users can report malicious posts themselves and do not have to wait until the platform itself has spotted these. But it also sees clear added value in technology, such as artificial intelligence tools to automate the process of detection and removal. Meta often emphasizes that the fight against disinformation involves a combination of human and technical efforts. Formulations like “a mix of automated and human review” (e. g., story 15, March 4, 2019, “Working to safeguard elections in Indonesia”) or “a combination of artificial intelligence, human review, and user reports” (e. g., story 68, July 20, 2022, “How Meta is preparing for Kenya’s 2022 general elections”) are used dozens of times throughout the communication. This is probably done to reassure critics of excessive automation and to indicate the “best of both worlds,” namely the speed and efficiency of AI tools and the accuracy and interpretation of humans. As a basis for determining whether posts – once designated as misleading or erroneous – should be removed, its own internal policy or community standards are used. Meta refers several times in its communication to adjustments to these standards in order to better respond to current needs, with phrases such as “we made several updates to our Community Standards” (e. g., story 58, December 2, 2021, “An independent assessment of Meta’s human rights impact in the Philippines”), “strengthening our policy toward misleading manipulated videos” (e. g., story 22, January 6, 2020, “Enforcing against manipulated media”), or the fact that it announced it posted a new section on the website containing the Facebook Community Standards “where people can track the updates we make each month” (story 21, April 10, 2019, “Remove, reduce, inform: New steps to manage problematic content”). In other words, the company communicates that it keeps its finger on the pulse all the time and tries to be transparent about it.

Posts that are misleading but not in violation of community standards are not deleted, but their distribution is restricted. Users should therefore see these messages (and the pages and groups in which these types of messages are posted) less frequently in their timelines. This should minimize the reach and impact of those types of messages and could thwart the revenue model of those spreading them. According to Meta, this is happening not only on Facebook but also on Instagram, and, in a later phase, on WhatsApp. Informing includes the company’s efforts to provide the public with additional context for posts that are potentially misleading or erroneous but still allowed to remain, such as a label or link to a fact-check just below the post or a warning when forwarding a particular post.

In addition to these strategies as solutions, which throughout Meta’s six years of crisis communication in the newsroom are always repeated in more or less the same form and can thus be considered key messages it consistently formulates, there are also several other interventions that are regularly mentioned as solutions in the fight against disinformation. Whereas the previous strategies rather started from the negative, namely the presence of erroneous reporting that is removed or limited and can thus be considered interventions where the focus is on the negative, other strategies can be seen as interventions where a positive message is given. For example, Meta communicates about interventions aimed at strengthening the community, making it more self-reliant, and empowering it through a variety of information and tools to fight disinformation itself. It is striking, for example, that during the last few years of Meta’s communication it has been emphasized very regularly that its goal is not only to protect users from bad information by removing or restricting it but also to connect them with “trusted,” “credible,” and “authentic” information. This also came up very occasionally during the early years when referring to the intervention of adding a link to a fact-check in case of a questionable post, but this strategy was emphasized more often over the years. We saw this especially in the context of Covid-19 and stories about disinformation about climate change where, for example, stickers were launched to give extra attention to reliable posts. Separate information centers were started on the platforms, where users could always find verified news posts and statistics about those hot topics, and with which Meta could indicate that its goal was “to more clearly connect people with credible and accurate information about Covid-19” (Story 27 update 2, December 15, 2020, “An update on our work to keep people informed and limit misinformation about COVID-19”). By communicating about this kind of intervention, it is not only about the presence of bad information on the platforms anymore but also about the good information that can be found there, which in terms of crisis communication and impression management seems a defendable strategy. The same can be said about trainings in key digital skills that Meta provided to (local) governments, NGOs, and newsrooms, which zoom in on strengthening knowledge and improving skills rather than stressing the negative.

Taking responsibility

Throughout the coding and analysis of the newsroom messages, we also focused on whether and how Meta takes responsibility for the disinformation crisis on its platforms. Meta often refers in its communications to the fact that disinformation must be fought on its platforms, so it is not denying the crisis but does not seem to want to take explicit responsibility for it in its communications. According to Meta, those responsible for the crisis are the actors with bad intentions, those who inject and spread disinformation on the platforms and do so for political or economic reasons: “State actors and agitators around the world” (story 7, June 14, 2018, “How is Facebook’s fact-checking program working?”) or “state-controlled media outlets” (e. g., story 59 update 4, March 1, 2022, “Meta’s ongoing efforts regarding Russia’s invasion of Ukraine”), for example, and very often it simply refers to “bad actors” (e. g., story 28, November 7, 2019, “How Facebook has prepared for the 109 UK general election”) or “fake accounts which are responsible for misinformation” (story 7, June 14, 2018, “How is Facebook’s fact-checking program working?”) and thus not Meta itself. It also frequently refers to the fact that those devious actors with bad intentions continually adjust their strategies to try to circumvent Meta’s detection strategy, requiring the company to make continuous efforts to combat them. As it writes in October 2018 (story 9, “Removing additional inauthentic activity from Facebook”): “As we get better at uncovering this kind of abuse, the people behind it – whether economically or politically motivated – will change their tactics to evade detection. It’s why we continue to invest heavily, including in better technology, to prevent this kind of misuse.” With this, Meta seems to be employing the “shift the blame” strategy (i. e., put the blame on something outside the organization, see Benoit, 2005). Meta sells itself as a victim but at the same time stresses the fact that the company is defending itself at all costs. And it is doing that for all of us, it indicates. While, as mentioned, it does not feel responsible for the disinformation as such, it does communicate very regularly that it feels responsible for fighting the problem. In September 2021 (story 53, “What the Wall Street Journal got wrong”), for example, it writes: “These are serious and complex issues, and it is absolutely legitimate for us to be held to account for how we deal with them.” In other words, it does not seem to bear responsibility for the problem, but it does bear responsibility for the solution.

The interesting thing is that in stating this responsibility, it very often takes the opportunity to proclaim a positive message about Meta and its various platforms, thus seeming to use the strategy of bolstering as impression management. It does this in two ways. First, by highlighting some positive values and goals. For example, stories often begin with a general statement by which the company highlights the importance of free speech and accurate information. This is the case in this intro to a January 2019 post (story 14, “Working to safeguard elections in Thailand”), “Protecting the integrity of elections while making sure people can have a voice in the political process is a top priority for Facebook,” and in this one from March 2021 (story 44, “Mark Zuckerberg announces Facebook’s plans to help get people vaccinated against COVID-19”), “Building on our goal to promote authoritative information about Covid-19 vaccines, we have implemented (…).” A second way to give a positive message is to refer to the size of the company, to the many billions of users who use its platforms every day to connect with others or look for reliable information and for whom Meta therefore feels responsible. With this, Meta further emphasizes that its popular platforms are of interest to numerous stakeholders. In September 2020 (story 36, “Stepping up the fight against climate change”), for example, it sounds like this: “As a global company that connects more than 3 billion people across our apps every month, we understand the responsibility Facebook has and we want to make a real difference;” or in March 2020 (story 41 update 8, March 16, 2020, Working with industry partners”): “We’re helping millions of people stay connected while also jointly combating fraud and misinformation about the virus.”

Finally, Meta also employs a strategy of extending responsibility to all of society, emphasizing that disinformation is not a problem of social media platforms alone but of everyone. “It‘s not a new phenomenon” (story 2, April 6, 2017, “Working to stop misinformation and false news”), it emphasizes in its communication, and one that cannot be solved “with a single solution” (e. g., story 54, September 21, 2021, “Our progress addressing challenges and innovating”); “fighting misinformation is an ever-evolving problem and we can’t do it alone” (story 23, June 12, 2019, “Automation plus expert journalism: How Full Fact is fighting misinformation”), adding that it is a shared responsibility of not only tech companies but also news and media organizations and teachers.