Making pushbacks public: Secrecy,material witnesses and devices of dis/appearance

Abstract

In early 2021, the European Parliament established the Frontex Scrutiny Working Group (FSWG) to monitor all aspects of the functioning of the European Border and Coast Guard Agency (Frontex). The FSWG organized a series of public hearings and carried out a ‘fact-finding’ investigation to gather information and evidence about pushbacks of refugees in the Aegean Sea. By unpacking some of the controversies that emerged during the hearings of the FSWG, I explore how secrecy was practised and strategically employed to obscure the responsibility of Frontex for the reported pushbacks, and how it was contested through the presentation of related evidence. I explain how secrecy and related controversies and struggles over making pushbacks public involve a variety of actors that enrol and interact with a heterogeneous set of objects, including digital, visual and archival traces of violence at sea, as well as databases used to record information about maritime incidents. I argue that secrecy regarding pushbacks is not just about keeping information about people and objects involved in security operations hidden. Secrecy is also produced through the selective recording, (mis)categorization and circulation of information in the name of transparency.

Introduction

In early 2021, the European Parliament (2021a) established the Frontex Scrutiny Working Group (FSWG) to monitor all aspects of the functioning of the European Border and Coast Guard Agency (Frontex). During the first four months of its activities, the FSWG organized a series of public hearings and conducted a ‘fact-finding’ investigation to gather information about fundamental rights violations in the Aegean Sea. This investigation was launched after the publication of a media report claiming that Frontex had knowledge of – and in some cases was directly involved in – illegal pushbacks of would-be refugees in Greek territorial waters during Joint Operation Poseidon (Bellingcat, 2020). Justified on the basis of longstanding security-humanitarian rhetoric that represents migrants as being at risk and as potential risks to European security (Aradau, 2004; Garelli and Tazzioli, 2018; Pallister-Wilkins, 2022; Perkowski, 2018), Operation Poseidon was launched to support Greek authorities with border surveillance and search and rescue (SAR) operations. Amid accusations of covering up fundamental rights violations and scandals of financial mismanagement, Frontex’s executive director, Fabrice Leggeri, resigned in April 2022 (Loschi, 2022). His resignation however, has not, been followed by any official acknowledgement of the agency’s involvement in pushbacks, while representatives of the Greek government have repeatedly dismissed allegations as ‘fake news’ (Nielsen, 2021).

In this article, I explore how secrecy was practised and strategically employed to obscure the responsibility of Frontex for the reported pushbacks and how it was contested through the gathering and presentation of related evidence. I do so by unpacking some of the central controversies that emerged during the public hearings and fact-finding investigation of the FSWG. Hearings are important research sites for understanding how secrecy is made public through knowledge claims and the presentation of evidence about covert security processes (see De Goede, 2014; Gros et al., 2017; Ingram, 2019; Walters, 2014). Indeed, controversies that emerge during hearings ‘[re]distribute the sensible’ (Birchall, 2021: 31), since they make it possible to appreciate how specific incidents relevant to security, like pushbacks at sea, are debated and rendered (in)visible.

I argue that secrecy regarding pushbacks is not just about keeping information about people (e.g. coast guards, migrants), objects (e.g. patrol boats, helicopters) and security operations hidden. Secrecy is produced through a complex set of practices that also involve the selective recording, storage and circulation of information in the name of transparency. Building upon arguments emerging in the literature that examines the complex interplay between secrecy and transparency (e.g. Birchall, 2021; Bratich, 2006; Hansen and Flyverbom, 2015), I challenge the idea that simply making information about state violence available leads to those responsible being held accountable. Evidence is never self-evident (Schuppli, 2020). We will see that making pushbacks public involved carefully piecing together and presenting information that was strategically registered and (mis)categorized in databases managed by Frontex.

This article contributes to the literature in critical security studies that examines secrecy not as something hidden, pre-given and fixed, but as being reiteratively performed in practice (e.g. Balmer, 2004; De Goede and Wesseling, 2017; Kearns, 2017; Walters, 2021). It does so by elaborating an analytical framework attentive to ‘secrecy controversies’ (De Goede and Wesseling, 2017) and introducing a conceptual toolbox with which to study sociotechnical processes of concealment and exposure. Inspired by science and technology studies and material semiotics (e.g. Callon et al., 2001; Latour and Weibel, 2005; Venturini and Munk, 2022), my analysis focuses on the issues, present-absent actors, material witnesses and devices of dis/appearance implicated in the controversies revolving around pushbacks. ‘Material witnesses’ (Schuppli, 2020) refers to objects like USB sticks, videos, images and geolocation data that embody knowledge about security operations and state violence. During the hearings of the FSWG, investigative journalists mobilized material witnesses to expose pushbacks. Still, the credibility of both journalists and material witnesses was questioned by actors who attempted to hide Frontex’s involvement. Conversely, ‘devices of dis/appearance’ refers to mechanisms that produce secrecy even though they are deployed to make the state more transparent. The device of dis/appearance I will focus on is the so-called Serious Incident Reporting (SIR) mechanism. While this was designed to enhance the transparency of Frontex’s operations, it was used to hide pushbacks by selectively reporting information about them. Attending to material witnesses and devices of dis/appearance implicated in controversies enables us to understand how transparency and secrecy are intertwined.

This article also speaks to the emerging literature in border and migration studies that investigates the production of non-knowledge at borders and other sites where mobility is controlled, filtered or contained. Scholars explore how migrants’ passages and bureaucratic trajectories are rendered opaque through their non-registration by state authorities (Rozakou, 2017; Tazzioli, 2020). Others inquire into how non-knowledge results from omissions, errors and failures in the processing and sharing of data about migration (Aradau and Perret, 2022; Scheel and Ustek-Spilda, 2019). Focusing on non-knowledge as a form of resistance and contestation (see Stel, 2016), scholars also investigate how migrants attempt to evade the surveillant gaze of authorities and exercise their ‘right to opacity’ (Glissant, 1997) by, for instance, submitting fake documents when applying for visas (Scheel, 2017), and even by burning their fingertips to avoid biometric identification (Glouftsios and Casaglia, 2023). What the present article adds to this literature is an explicit consideration of how secrecy and, paradoxically, transparency work together to produce non-knowledge about border violence. Non-knowledge is produced by various actors for different purposes: authorities not registering migrants to avoid dealing with asylum claims, migrants trying to resist and contest state control, researchers and investigative journalists withholding information to protect their sources. Focusing specifically on state secrecy and how it feeds into non-knowledge is an important analytical exercise that allows us to unveil actors, practices and objects that render border violence invisible, resulting in the evasion or denial of responsibilities for migrants’ fundamental rights violations.

To be clear, my goal is not just to emphasize secrecy as a constitutive characteristic of migration management. Besides recognizing that ‘exclusion requires silence and concealment’ (Mountz, 2011: 122; see also Moran and Gill, 2016), I demonstrate how silences and concealments, taking the form of secrecy, are enacted and contested. Also, instead of approaching secrecy as a challenge that various actors – researchers, journalists, oversight bodies – deal with when investigating the injustices of contemporary borders (Cleton, 2022; Klimburg-Witjes et al., 2022; Wissink, 2019), I study secrecy as a construction: as emerging from a set of practices that go beyond hiding.

The remainder of this article is structured as follows. In the first section, I reflect upon the complex interplay between secrecy and transparency. In the second section, I explain why public hearings are important for secrecy research. In the third section, I elaborate the methodological framework that I employ to unpack controversies relevant to the hearings and fact-finding investigation of the FSWG. Finally, in the last two parts of the article, I attend to these controversies and the different actors, material witnesses and devices of dis/appearance that performed and contested secrecy about pushbacks.

Secrecy, exposure and transparency

Secrecy is performative (Glouftsios, 2023a; 2023b): it is crafted within different sociotechnical contexts and produces specific effects relating to those contexts. Walters (2021: 9) emphasizes the need to understand how state secrecy is (un)made by engaging with its formal and informal practices, the identities and hierarchies it produces, and the paradoxes, contradictions and controversies that it may engender. In other words, he examines secrecy ‘in situ’ (Walters, 2021: 7) by advancing a research programme that interrogates how it is enacted, governed and contested. This implies that secrecy is not something universal. What constitutes the secret is in flux: mutating, taking multiple forms and meanings, and exceeding any clearcut definitions representing it as something obscure and invisible. Similarly, Birchall (2021: 8) invites attention to the material practices, institutions and technologies through which secrecy is constructed, problematizing its dominant but simplistic conceptualization as hiding and concealment. What is important to understand about secrecy is, as proposed by De Goede and Wesseling (2017: 257), ‘less its hiding per se, and more the way in which it structures social relations, regulates communication, and distributes political power’. Secrecy structures the relation between those who know and those who do not (Horn, 2011: 109), and conditions their respective practices.

The questions driving research on secrecy go beyond the ‘hermeneutics of the secret’ (Birchall, 2021) – that is, the revelation of the meaning and ‘truth’ of states’ dirty secrets. Examples of such secrets are surveillance infrastructures (Gros et al., 2017), targeted assassinations (Ingram, 2019), drone strikes (Walters, 2014) and the use of torture in counter-terrorism (Stampnitzky, 2020). Instead of merely trying to reveal such secrets, scholars explore how secrecy is performed through practices like document classification and redaction (Curtin, 2014; Luscombe and Walby, 2017), the implementation of ineffective oversight mechanisms that result in organizational ignorance and non-knowledge (Aradau, 2017; Rappert, 2012), and the manufacturing of secrecy as a ‘spectacle’ (Bratich, 2006; Van Veeren, 2011) in media, arts and the entertainment industries.

As I will further explain below, during the hearings of the FSWG, representatives of Frontex and other actors framed migration as being associated with ‘hybrid threats’ to the national security of EU member-states. They also emphasized the ‘security-sensitive’ nature of Frontex’s mandate and that releasing information on its operations would jeopardize its work as a ‘law enforcement’ agency. On the basis of such claims, secrecy in border and migration management was justified along with what Thomas (2020: 78–79) describes as the ‘securitization of information’. This implies the acknowledgement that information disclosure may threaten European security without, however, recognizing the risks that secrecy may pose to the protection of fundamental rights and democratic accountability more broadly. The securitization of information feeds into what Pozen (2005: 630; see also De Goede and Wesseling, 2017) outlines as the ‘mosaic theory’ in intelligence gathering, according to which the disclosure of certain pieces of seemingly non-sensitive information (e.g. which Frontex assets were deployed in the Aegean Sea during the reported pushbacks) can be potentially used by an ‘adversary’ (e.g. trafficking networks) to gain knowledge about and potentially circumvent security measures. The problem with the mosaic theory and wider securitization of information is that they open ‘a discretionary space for actions that do not have to be accounted for’ (Horn, 2011: 106): they can lead to the concealment of information that could function as evidence of Frontex’s involvement, or complicity, in pushbacks.

It is important to clarify that disclosing relevant information does not equal the revelation of states’ secrets. As Stampnitzky (2020: 598) explains, while ‘exposure’ refers to the overall process of ‘making information publicly available’, ‘revelation’ is about the ‘collective recognition that there has been a significant change in what is publicly known’. For information to function as evidence, it should become evident and credible to the audience in front of which it is presented. The production of evidence is achieved through the careful ‘public staging’ (Aradau and Huysmans, 2019: 49) of information and knowledge about secrets across different social domains that may involve policymakers, legislators, experts and the wider public. This also requires much representational effort and skills on the part of those who make objects (e.g. documents, videos, photographs, coordinates) act as material ‘traces’ of certain events, like deaths at sea (see Pezzani and Heller, 2019) and ‘material witnesses’ (Schuppli, 2020) that render state violence visible and sense-able.

I want to emphasize the role that different non-state actors play in publicizing security processes and violence. In the case of pushbacks, the role of investigative journalists was crucial as they engaged in meticulous ‘epistemic work’ (Davies et al., 2023) that involved gathering, analysing, disseminating and presenting information to hold Frontex accountable. This information was not always hidden from the public’s view. We will see that journalists assembled and exposed information about pushbacks by consulting, among others, publicly accessible websites used to monitor marine and air traffic. The work of journalists and other actors, such as whistle-blowers and anonymous collectives like WikiLeaks (see Birchall, 2021), should be seen as resisting and contesting state secrecy.

Yet exposure, disclosure and the careful staging of evidence do not always result in revelation, while the transparency of state institutions and agencies does not necessarily translate into less secrecy. For Birchall (2021: 4, 9), secrecy and transparency are ‘conjoined twins’. This means that transparency can produce secrecy effects. The ‘project of exposing the hidden’, as Hansen and Flyverbom (2015: 873) put it, ‘often ends up concealing more than is revealed’. Transparency can function as ‘a strategy of public perception management’ (Bratich, 2006: 498), serving the goal of selectively and strategically releasing information. For example, as Luscombe and Walby (2017) demonstrate in their research on government transparency, freedom of information (FOI) often operates as a mechanism for state secrecy. This is evident when authorities heavily redact documents or provide only printed material instead of electronic files that are easier to search (Luscombe and Walby, 2017: 383). Transparency can also be productive of secrecy when events are reported by eliminating contextual information or adding irrelevant information that produces distortion (Hansen and Flyverbom, 2015). As we will see later in this article, monitoring and reporting mechanisms can, in theory, make Frontex operations more transparent. However, when these mechanisms are implemented in practice, they are not always effective and can be used for concealing, instead of disclosing, information about pushbacks. The paradox is that transparency – or, better, the practical work and mechanisms that render an organization, process or event transparent – may ‘be part of secrecy’s game’ (Bratich, 2006: 494).

Studying secrecy through public inquiries and hearings

Public inquiries and hearings in parliaments are important sites in which to study how secrecy is performed and contested. Security and secrecy do not always emerge as forms of anti-political exception: they can be constitutive aspects of ‘normal’ parliamentary work and professional political life (Neal, 2019; see also Hagmann et al., 2018). Inquiries can bring security issues characterized by secrecy into public deliberation. This is done through the presentation of evidence, the expression of political positions and responsibilities, and the forwarding of measures to enhance the transparency of state institutions and agencies (Gros et al., 2017). Inquiries and hearings can offer fruitful avenues for research into secrecy since they function as forums where political struggles and controversies unfold (Aradau, 2017; De Goede, 2014; Thomas, 2020), allowing researchers to understand how secrecy is debated, rationalized and put into discourse (Walters, 2021: 101–102).

In research on security, borders and migration management, scholars can encounter challenges related to field access when investigating controversial policies and practices: extradition of terrorist suspects (Kapoor, 2018), deportation programmes (Cleton, 2022; Walters, 2018; Wissink, 2019), border violence and pushbacks (Davies et al., 2023; Karamanidou and Kasparek, 2022). Such challenges are often due to state secrecy, which is directly linked to the tactical production of non-knowledge to avoid ‘repercussions of inconvenient evidence’ and liability for malpractices (McGoey, 2019: 2–3). Hearings and inquiries are useful in this context because they allow researchers to follow debates during which state actors are invited to give an account of previously concealed practices, and where secrecy is both actively contested (e.g. through questioning and the presentation of evidence) and maintained (e.g. through arguments that justify state secrecy or contest previously presented evidence). Hearings and inquiries can also help researchers gather empirical material that would be otherwise difficult to access as they result in the creation of rich audio-visual and documentary records available to the public. As Walters (2021: 101) explains, hearings also allow for the ‘translation’ of complex, technical and obscure security-related information to the non-insiders. This is done through the interventions of various actors who either participate in hearings or comment and present aspects of them to the public. To use Birchall’s (2021: 31) words, inquiries and hearings can ‘[re]distribute the sensible’ – that which is visible, audible, sayable and knowable about secrecy – which then conditions political (re)actions vis-a-vis state secrets.

What matters, methodologically and analytically, is the way that secrecy is (un)made through controversies, discursive struggles and wider narratives of concern that revolve around issues pertinent to security (De Goede, 2014; Gros et al., 2017; Walters, 2014; Walters and D’Aoust, 2015). Engaging with controversies is a way to observe security politics and secrecy in action. By focusing on the controversies that emerged during the hearings and fact-finding investigation of the FSWG, one can understand how secrecy regarding pushbacks was practised and contested. Pushbacks should be understood here as ‘events’ (Ingram, 2019) that engender controversies about whether and how they happened, events that formed the basis upon which the work of the FSWG unfolded to assemble knowledge that would be credible enough to function as evidence of Frontex’s (non-)involvement in fundamental rights violations.

Controversy-oriented research has emerged within science and technology studies, where scholars have been studying the development and closure of technoscientific disputes and the variegated processes through which knowledge claims come to be accepted or rejected (e.g. Barry, 2001; Bijker and Law, 1992). Research in science and technology studies has also explored how ‘matters of concern’ (Latour and Weibel, 2005; Marres, 2012) engender disagreements that bring together different publics. As Latour and Weibel (2005: 23) put it, ‘we don’t assemble because we agree . . . but because are brought by divisive matters of concern into some neutral, isolated place in order to come to some sort of provisional makeshift (dis)agreement’. Matters of concern engender and sustain controversies that involve various actors who attempt to contest each other’s authority and knowledge claims.

In this literature, controversies are approached as dynamic and fluid situations that unfold in space and time (Callon et al., 2001: 26). The dynamism and fluidity of controversies mean that their substance (e.g. the issues around which they revolve) and configuration (i.e. the actors participating in them) are subject to change (Venturini and Munk, 2022: 35). Another characteristic of controversies is that they are never just ‘technical’ or subject entirely to expert knowledge. As Callon et al. (2001: 24) note, to declare a controversy purely technical is effectively a practice that removes it from the influence of the public debate. For example, to say that reported pushbacks ought to be investigated only by experts and practitioners in border management with the necessary technical skills and operational knowledge is a particular kind of ‘boundary work’ (Gieryn, 1983) through which the knowledge claims of others (e.g. journalists, NGOs) are discredited.

Unpacking controversies: Issues, actors, objects

I followed a threefold research strategy to unpack the controversies that emerged during the fact-finding investigation and hearings of the FSWG. First, I identified the main issues that composed controversies. According to Venturini and Munk (2022: 65–66), distinguishing controversies from issues allows researchers to break down controversies into smaller questions and systematically explain the stakes in a debate. Different issues are interdependent within controversies and, as these evolve, new issues may appear while the relevance of others may diminish. There have been multiple controversies linked, in one way or another, to pushbacks in the Aegean Sea, but not every controversy revolved around secrecy. For instance, in some FSWG hearings, controversies concerned the legal uncertainties that presumably make it difficult to determine whether pushbacks are (il)legal (see Glouftsios 2023b). In this article, I focused on two specific controversies directly linked to attempts to obscure information that could function as evidence of Frontex’s involvement in or awareness of pushbacks.

To identify the issues that composed controversies, I listened to, transcribed, coded and subsequently analysed thematically eight public hearings, each roughly one hour long, that took place during the four-month fact-finding investigation of the FSWG. I first analysed the transcript of the hearing during which investigative journalists presented their research. I wanted to understand how they assembled relevant information about Frontex’s attempts to cover up pushbacks, how they presented this information in front of the FSWG, and, importantly, what reactions and questions their presentations triggered. I then analysed the remaining hearings chronologically, paying particular attention to emerging disputes and how knowledge claims were framed and contested. For example, I explored how Frontex representatives tried to discredit journalists’ claims and how members of the European Parliament tried to question arguments raised by Frontex about its non-responsibility for and lack of knowledge about pushbacks. I aimed to discern the central issues and ‘matters of concern’ (Latour and Weibel, 2005) that composed each secrecy controversy. Studying the hearings was essential to detect disagreements, tensions, and convoluted practices of secrecy and exposure that were rendered somewhat invisible in the final report published by the FSWG. I also analysed publicly available documents to find further information about the issues at the core of each controversy. These included legislative texts and manuals governing Frontex operations, reports that resulted from the investigations of the FSWG and a working group established within Frontex’s management board, reports produced by oversight bodies like the European Ombudsman and the European Anti-Fraud Office (OLAF), as well as letters exchanged between the then executive director of Frontex and national authorities. Finally, I conducted four semi-structured interviews with members of the European Parliament (or their assistants) who participated in the hearings and an interview with one of the journalists who contributed to the investigation and testified in front of the FSWG. These interviews were not recorded and did not form the basis of my analysis. They provided some additional clarifications about the hearings and the issues that emerged.

Second, besides the issues that composed controversies, I identified the main actors involved. ‘Actors’ typically refers to humans and non-humans who make a difference in controversies (Venturini and Munk, 2022: 80; see also Callon et al., 2001). However, for clarity reasons, I reserved the category of the actor to subjects, their practices and knowledge claims, while I inquired into the role of objects in controversies separately. My goal was not just to make an inventory of actors relevant to each controversy but to understand how they tried to expose, conceal or obscure information about pushbacks. Each controversy concerned a variety of actors who were both present and absent from the hearings. Among them were the parliamentarians from the different political groups who were raising questions relevant to pushbacks and actors invited to provide answers and present supporting evidence. These included representatives from the European Commission, Frontex, national border and coast guard authorities, and the European Ombudsman, along with legal experts and journalists. During the hearings, these actors formed alliances by supporting or collectively disputing the claims of others. Through representation, they also enrolled other actors who were absent from the hearings in controversies, such as migrants subjected to pushbacks or frontline border guards. Yet, as Latour and Weibel (2005: 26) remind us, representation is not the same as re-presentation: the ways in which practices and experiences of absent actors were made present in controversies often involved provisional assertions, omissions and secretions, as well as disputes and negotiations about evidence.

Third, besides issues and actors, I attended to the double role that objects played during the hearings and fact-finding investigation of the FSWG. Objects like videos, images, geolocation data and social media posts emerged as ‘material witnesses’ (Schuppli, 2020) that embodied knowledge about practices related to pushbacks. Such objects do not speak of themselves and, when they emerge as material witnesses, they do not automatically produce evidence. It is the human actors involved in controversies who open up ‘the expressive potential of things’ (Schuppli, 2020: 9) by speaking on their behalf and making meaningful the knowledge that they embody. As Walters and D’Aoust (2015: 54) put it, objects ‘require mediators and translators who speak (or claim to speak) for them in public forums’. This means that knowledge and evidence are not just inter-subjective, but also inter-objective (Venturini and Munk, 2022: 99). Indeed, the knowledge that objects embody is often framed as more credible than the testimonies and claims of humans – for example, victims and perpetrators of state violence – that are not materially substantiated.

Besides acting as material witnesses, objects that participated in the controversies about pushbacks also functioned as ‘devices of dis/appearance’. This is different from what Walters (2021: 88) describes as ‘devices of dis/closure’ to explain how ‘certain forms of closure may be a condition of possibility for processes and practices of disclosure’. A device of dis/closure can be, for example, a consent form that we sign with our research participants to, among other things, guarantee their anonymity. By agreeing not to disclose their identity, we support them in disclosing information they may deem sensitive. Differently from devices of dis/closure, devices of dis/appearance imply that certain modes or mechanisms implemented to make practices, organizational processes and events (e.g. pushbacks) apparent can, in practice, produce distortion and secrecy. Devices of dis/appearance trouble the presumed opposition between secrecy and transparency, pointing to the fact that mechanisms designed to make the state more transparent can be tactically used to hide its dirty secrets, sometimes in plain sight (Birchall, 2021; Hansen and Flyverbom, 2015). Specifically, devices of dis/appearance imply that through the appearance of certain elements, others disappear and become invisible (see Bellanova and González Fuster, 2013). As I explain in the following sections, databases that record information related to incidents at sea linked to potential violations of fundamental rights were used to make such information dis/appear. ‘Dis/appearance’ here does not just refer to situations where information is not recorded or deleted. It also refers to cases where information is registered but under categories irrelevant to fundamental rights violations. Pushback incidents appeared in databases managed by Frontex but, at the same time, they disappeared through their categorization as non-pushbacks.

Secrecy and material witnesses

One of the main controversies that emerged during the fact-finding investigation and hearings of the FSWG revolved around investigative reporting that exposed how secrecy is practised during Frontex joint operations. According to Bellingcat’s (2020) article that sparked the debates on Frontex’s complicity in pushbacks, and according to the information journalists presented during the FSWG hearings (European Parliament, 2021e), Frontex assets involved in pushbacks in the Aegean Sea attempted to hide their traces. To expose operational hiding, journalists enrolled, represented and spoke on behalf of material witnesses (videos, photos, geolocation data, etc.) in their publications and during the hearings. Yet the credibility of their knowledge claims was contested through various strategies that Frontex’s representatives employed to deny the accusations.

Exposing operational hiding

Bellingcat attempted to determine Frontex’s involvement in pushbacks by identifying and tracking the movements of assets used in Joint Operation Poseidon, including vessels, helicopters and airplanes. They used various sources of information, such as Frontex’s website, social media posts by ship-spotters, and publicly available real-time data on vessel and aircraft locations from sites like Marine Traffic and Flight Radar 24. To clarify, following Chapters IV and V of the 1974 International Convention for Safety of Life at Sea (SOLAS), as well as related technical standards adopted in later years, passenger, cargo and government vessels can be tracked through the so-called Automatic Identification System (AIS). Specialized equipment installed in vessels transmits AIS messages – identification codes, geolocation data, and course and speed information – to terrestrial stations established across coastlines or dedicated satellites (Pezzani and Heller, 2019). Bellingcat found that while some vessels had their AIS equipment turned on and could be tracked, others only transmitted data when in ports.

Hiding by turning off AIS equipment or using vessels without such equipment is a common practice used by organized crime networks involved in smuggling and trafficking (see Glouftsios and Loukinas, 2022). Whether Frontex intentionally turned off AIS equipment to avoid detection during pushbacks or if this is a standard operating practice remains unclear. In November 2020, one month after Bellingcat’s publication, the then executive director of Frontex, Fabrice Leggeri, asked the Romanian border police for explanations about a vessel they deployed on behalf of Frontex in Operation Poseidon (Frontex, 2020e). According to Bellingcat, this vessel was involved in one of the pushbacks and had its AIS equipment turned off. The Romanian border police replied that, as a government vessel, they have the option to operate in ‘visible mode’, where position and movement parameters are available to all civilian web platforms monitoring maritime traffic, or ‘invisible mode’, where access to position and movement parameters is only available to accredited users like Frontex (Romanian Border Police, 2020). They also stated that Greek authorities had requested them to operate in invisible mode during the patrol mission relevant to the pushback incident.

Turning off AIS equipment made it harder for Bellingcat to determine which Frontex assets were involved in pushbacks. Bellingcat (2020) obtained more detailed data from vessel- and flight-tracking companies and cross-referenced these with information collected by NGOs and coordinates sent by migrants. Concerning the pushback incident involving the Romanian vessel, a journalist from Lighthouse Reports who participated in the Bellingcat investigation stated during an FSWG hearing that they had obtained videos from the Turkish coast guard, who claimed to have rescued the migrants after the pushback (European Parliament, 2021e). It also emerged during the hearings that the Turkish authorities had sent videos to the FSWG via a USB stick in a diplomatic pouch (European Parliament, 2021d). The journalist demonstrated in his presentation that some of these videos showed the Romanian vessel blocking a rubber dinghy used by migrants and performing dangerous manoeuvres to create waves. Additionally, one of the videos contained encoded information about the time and location where it was recorded.

Tracking data and videos, understood as material witnesses, did not automatically produce evidence. The knowledge they embody had to be made evident and credible during the FSWG hearings. For example, to substantiate his claims, the journalist explained the methods they used to assess the authenticity of the videos provided by the Turkish authorities (European Parliament, 2021e). He explained how they forensically managed to confirm the location of the incident by comparing the landscape in the background of the videos with the landscape of the exact location seen through Google Earth. They also verified that the videos were filmed on the same date by comparing several videos in which the migrants in the rubber dinghy appeared to be wearing the same clothes. Notably, a NATO ship near the pushback appeared in the videos with its AIS equipment turned on. This was another piece of information that confirmed the incident’s location, date and time.

AIS data exposing vessels’ location and movement parameters, Google Earth photos, and landscapes and other elements appearing in videos were only some of the material witnesses that archived and made visible knowledge relating to the involvement of Frontex assets in pushbacks. These material witnesses did not speak of themselves. Investigative journalists re-presented them in their publications and interventions during the FSWG hearings. Secrecy performed through operational hiding was contested and exposed by journalists’ materially substantiated knowledge claims. To make these claims credible and render evidence of pushbacks evident to the FSWG, the journalists had to enrol several material witnesses and explain in detail the methods they used to gather, validate and synthesize information. Despite this careful staging of their findings, other actors who were also invited to testify before the FSWG contested their claims.

Contesting knowledge claims

The exposures made by investigative journalists did not result in an official acknowledgement of Frontex’s involvement in pushbacks. Indeed, Frontex’s Executive Director Fabrice Leggeri emphasized during the hearings that ‘there was no substantiated or evidence-based facts or conclusions saying that Frontex participated or carried out violations of fundamental rights’ (European Parliament, 2021b). This was not just a blunt rejection of the knowledge claims put forward by actors reporting pushbacks: it involved the practical contestation of such claims by discrediting the information embodied by material witnesses and presenting counter-evidence.

More specifically, Leggeri attempted to discredit the information presented by Bellingcat, arguing that Frontex could not authenticate the videos provided by the Turkish coast guard (European Parliament, 2021g). However, he failed to provide a detailed explanation as to why forensic analysis of the videos was not possible. Instead, he repeatedly referred to ‘hybrid threats’, ‘geopolitical complexities’ and the ‘instrumentalization’ of migration by Turkey. Although the FSWG’s work focused on concrete allegations of fundamental rights violations, Leggeri attempted to shift the focus to Turkish authorities, emphasizing that they cannot be trusted. He also claimed that, in other videos, Turkish vessels appeared ‘harassing’ Greek, Finnish and Swedish ones (European Parliament, 2021g). Similarly, on the basis of information from Frontex’s internal reporting mechanisms, the acting director of Frontex’s Operational Response Division, Ana Cristina Jorge, referred to evidence about ‘aggressions towards Frontex by Turkey’ that happen ‘very regularly’ (European Parliament, 2021c). Some parliamentarians from the European People’s Party (EPP) and Identity and Democracy (ID) right-wing groups also emphasized hybrid threats and geopolitical complexities. This was an attempt to shift the focus of the hearings from concrete allegations of pushbacks to Turkey’s instrumentalization of migration and to imply that information provided by Turkish authorities and later reproduced by investigative journalists should not be considered credible evidence.

Besides this strategy, Leggeri tried to substantiate his conclusion about the non-involvement of Frontex in pushbacks by referring to the work of a specialized working group that was established within Frontex’s management board to investigate, in parallel to the work of the FSWG, the ‘allegation of so-called pushbacks in the Eastern Mediterranean’ (Frontex, 2020a). Frontex’s management board consists of a representative from each member-state and two representatives of the European Commission’s Directorate General for Home Affairs. The working group comprised representatives and experts from the European Commission and eight member-states, including Greece and Romania (Frontex, 2020a). In March 2021 – more than one year after the publication of the report by Bellingcat and while the FSWG hearings were still taking place – the working group published its final report, which concluded that there was no clear evidence of Frontex’s involvement in fundamental rights violations (Frontex, 2021). In the following sections, I will discuss in more detail the report’s findings and its conclusions. It will suffice here to raise two points.

First, on the basis of some contradictory statements made by national authorities involved in pushbacks, the working group emphasized the uncertainty and ambiguity of the ‘facts’ presented by the journalists. We will see that Greek authorities and operators of Frontex assets disagreed about the precise location of certain pushback incidents. In other cases, Frontex assets claimed that they could not observe the Greek coast guards’ actions because they were asked to depart from the location of reported incidents and continue their patrol elsewhere. Contradictory statements and claims about the impossibility of verifying all the parameters of specific pushback incidents contributed to the production of uncertainty and its tactical mobilization by Frontex’s internal working group to substantiate the claim that there was no clear evidence of the agency’s involvement in, or knowledge of, pushbacks. Uncertainty should not be seen here as just a limitation of the investigation conducted by the working group, but as actively produced and promoted (see also Best, 2008) to evade liability for malpractices.

Second, uncertainty and ignorance about Frontex’s involvement in pushbacks also emerged owing to the inadequate investigatory methods employed by the working group. Indeed, we will see later in the article that the working group’s conclusions were met with scepticism as they were based exclusively on testimonies and information provided by Frontex staff and the concerned national authorities. Not considering sources other than those coming internally from Frontex and the member-states is a particular kind of boundary work that enacts secrecy by disregarding information that could function against Frontex’s interests. Secrecy, in this context, is not simply about hiding information but includes maintaining ignorance by not considering alternative, ‘unofficial’ sources and knowledge claims, like those of investigative journalists and NGOs working with migrants. Ignorance, as I explain below, was also fabricated through the strategic use of mechanisms implemented to report incidents at sea that could be potentially related to fundamental rights violations.

Transparency and devices of dis/appearance

The second major controversy that emerged during the hearings and fact-finding investigation of the FSWG concerned the ways in which knowledge relevant to specific patrolling missions of assets deployed on behalf of Frontex and Greece was obscured – knowledge that could function as evidence of the agency’s involvement, or complicity, in pushbacks and fundamental rights violations. While in the previous part of the article we saw how secrecy was performed through operational hiding and the contestation of the journalists’ knowledge claims, here I will explain how secrecy was, paradoxically, practised by using mechanisms that are supposed to enhance the transparency of Frontex operations. Secrecy resulted from the dis/appearance of information about pushbacks. Let me provide some background information useful for understanding the stakes of this controversy.

Under Article 80 of the 2019 European Border and Coast Guard Regulation (Official Journal of the European Union, 2019), Frontex must effectively monitor and guarantee the protection of fundamental rights during its activities. This means that Frontex should, in principle, be aware of fundamental rights violations in the area where it operates and performs border surveillance. To achieve such awareness, besides the deployment of monitoring infrastructure – for example, airplanes, helicopters, drones and coast guard ships equipped with cameras, radar and other surveillance equipment (see Dijstelbloem et al., 2017; Glouftsios and Loukinas, 2022; Tazzioli, 2018) – Frontex shall make effective use of reporting mechanisms. One example is the Serious Incident Reporting (SIR) mechanism, which allows participants in Frontex operations to report security-related events and situations of suspected violations of fundamental rights. Crews of vessels and aircraft patrolling sea borders should report serious incidents to the Frontex Situation Centre in Warsaw within two hours of their detection (Frontex, n.d.: 3). This is a two-step process. Serious incidents should be reported immediately by using all available means of communication (e.g. email, mobile phone) and then registered through the Joint Operations Reporting Application (JORA) (Frontex, n.d.: 4; Meijers Committee, 2021: 5; see also Tazzioli, 2018).

In JORA, incidents are divided into four categories (Frontex, n.d.: 1–3). Category 1 incidents are ‘situations of high political and/or operational relevance especially with potential effect on border management’, such as a terrorist attack or a border conflict between a member-state and a third country. Category 2 incidents occur ‘in Frontex activities/Joint Operations and not related to Frontex staff, or any other participant in such activity’, such as ‘unexpected massive arrivals of irregular migrants’ or incidents that involve authorities of third countries. Category 3 incidents are relevant to ‘participating actors in a Frontex coordinated activity’, such as serious accidents involving Frontex staff or deployed officers from national authorities. Finally, Category 4 incidents – the category most relevant to pushbacks – refer to ‘situations of suspected violations of Fundamental Rights’, such as the right to asylum and the violation of the principle of non-refoulement. Incident information includes details of the location, time, involved actors, operating assets and actions taken (Frontex, n.d.: 4).

In theory, serious incident reporting can function as a mechanism that renders border management processes more transparent by exposing, for example, information about pushbacks. In practice, however, serious incident reporting can also function as a device of dis/appearance that hides more than it reveals through the strategic categorization of incidents as not related to fundamental rights.

Hiding through (non-)reporting

During a public hearing of the FSWG, a journalist from Der Spiegel presented details of an incident that occurred near the north part of Lesbos island at around midnight on 18–19 April 2020 (European Parliament, 2021e). This incident was detailed in the final report of the working group established within Frontex’s management board (Frontex, 2021: 6–7), and the relevant SIR can be found online (Frontex, 2020b). The incident involved a Frontex surveillance aircraft that observed an empty rubber boat being towed by a Greek coast guard vessel towards Turkish territorial waters. On board the Greek vessel, there were approximately 20–30 migrants who were then transferred to the rubber boat at the Greek–Turkish border. The Greek coast guard left the scene and later stated that two Turkish coast guard vessels took over responsibility. This claim, however, could ‘not be corroborated and documented by the Frontex Surveillance Aircraft sightings due to its merely partial involvement in the incident’ (Frontex, 2021: 7). Crucially, in the photographs taken by the aircraft, the rubber boat appears to have no engine. Also, when it was abandoned with migrants on board at the Greek–Turkish border, there were no Turkish vessels present at the location. The incident was registered as Category 2 (i.e. not related to Frontex staff and participants in operations) instead of Category 4 (i.e. fundamental rights–related). This categorization contradicted the conclusion of the SIR, according to which migrants ‘were on board the patrol vessel and then placed back in the rubber boat, put again in a distress situation and left adrift without means of propulsion.... [T]here is a strong belief that presented facts support an allegation of possible violation of Fundamental Rights or international protection obligations such as the principle of non-refoulement’ (Frontex, 2020b).

In this case, serious incident reporting functioned as a mechanism of dis/appearance. The details of the incident appeared in JORA but at the same time disappeared and were rendered invisible as it was categorized as not related to fundamental rights. While serious incident reporting could expose potential violations of fundamental rights, we see here the convoluted ways in which transparency, via the implementation of reporting mechanisms, can produce secrecy effects when it is employed strategically to make information and evidence dis/appear. Nevertheless, besides dis/appearance enacted through the wrongful categorization of incidents, during the hearings of the FSWG there also emerged cases of attempts to register false information or bluntly conceal incidents through non-reporting.

For example, during the hearing on 15 March 2021, a superintendent of the Swedish border police argued that it is ‘partially true’ that Swedish officers faced resistance in registering a SIR. He had previously stated at a meeting of the Frontex management board that ‘there may be a problem in investigating allegations based on incident reports that do not exist because they had never been lodged, because there is an understanding from the operational management that SIRs should not be done’ (Frontex, 2020c). These statements referred to an incident on 30 October 2020, when a Swedish coast guard vessel intercepted a rubber boat with migrants inside Greek territorial waters (see Frontex, 2021: 10–11). After handing the incident over to the Greek coast guard, the Swedish crew monitored the situation through radar and observed a singular echo on their screens, despite two vessels moving towards Greek territorial waters. According to the Swedish crew, they were initially hampered from lodging a SIR by the responsible Frontex officer. Eventually, they reported the incident, but it was inconsistently ‘categorized as Prevention of Departure despite the fact that the rubber boat entered Greek Territorial waters’ (Frontex, 2021: 10–11).

In another incident on 27 July 2020 (Frontex, 2021: 8), a Danish helicopter crew detected a rubber boat near the island of Chios in Greek territorial waters. They reported the sighting to the Greek coast guard, who intercepted the boat. However, the incident was classified as ‘prevention of departure’, suggesting that the Turkish coast guard (not the Greek) intercepted the boat in Turkish territorial waters. The Danish crew also claimed that Greek authorities asked them to change the reported location of the incident. Frontex’s executive director sent a letter to the Greek authorities seeking clarification on these allegations (Frontex, 2020d). In their reply, the Greek authorities criticized ‘the SIR reporter who unilaterally proceeded with his actions based on a personal assumption and without any prior feedback’ from the Greek coast guard (Hellenic Coast Guard, 2020), even though according to the guidelines for the registration of SIRs such feedback is not necessary as incidents should be reported immediately (Frontex, n.d.). It was later verified that the incident occurred inside Greek territorial waters, and ‘the Greek side stated that the post-operational communication between them and the Danish detachment was a misunderstanding’ (Frontex, 2021: 8).

These are not just single cases of reporting false information through SIRs. Investigative journalists – including those who were invited to present information in front of the FSWG (European Parliament, 2021e) – gained access to entries in JORA and found a pattern of hiding pushbacks by classifying incidents as ‘prevention of departure’ even though migrant boats are intercepted in Greek territorial waters (see Christides and Lüdke, 2022). Classifying incidents as prevention of departure and categorizing them as not related to fundamental rights means that Frontex’s fundamental rights officer – who is responsible for, among other things, monitoring Frontex’s compliance with fundamental rights (Official Journal of the European Union, 2019: Article 109) – is not informed about the relevant SIRs. This is indicative of attempts to withhold information even internally in Frontex.

(Im)partiality and failure in Frontex’s internal investigation

What raised disputes during the hearings of the FSWG was the (im)partiality of the investigation conducted by the working group established within Frontex’s management board. These disputes and overall scepticism about the investigation were raised because, although it became evident that reporting mechanisms were used to make information about pushbacks dis/appear, Frontex’s management board did not acknowledge the agency’s complicity in fundamental rights violations. The management board accepted information and justifications provided by Greek and other national authorities without question. For example, in relation to the case of the April 2020 incident where a rubber boat with migrants was left without an engine, the working group reiterated the Greek authorities’ claim that the engine was not visible on the photographic material produced by the surveillance aircraft ‘either due to the angle and distance of the aircraft or because the engine was temporarily unmounted’ (Frontex, 2021: 7). Instead of consulting sources beyond national authorities, including the journalists who conducted the investigation, the working group relied solely on reporting mechanisms and testimony from national authorities to investigate alleged pushbacks.

Scepticism about the investigation’s (im)partiality was also evident in the FSWG’s concluding report, which expressed ‘regret’ that the working group ‘limited its inquiry to information retrieved from within the Agency itself and from the Member States’ (European Parliament, 2021a: 13). During one of the hearings, parliamentarians repeatedly questioned the chairperson of the Frontex management board, Marko Gašperlin, about the independence of its investigation. The rapporteur of the FSWG, Tineke Strik (Greens/European Free Alliance Group), argued that ‘the conclusions raise some questions on the methodology’ regarding the sources consulted and asked why the investigation did not involve the journalists whose work was the reason for the inquiry in the first place, as well as NGOs and other actors working with migrants on the ground. Gašperlin evidently avoided answering the question, claiming he was not directly involved in the investigation but that his colleagues ‘used all available information’ (European Parliament, 2021c).

During the same hearing, Sira Rego (Left Group) criticized the lack of answers to the questions raised by the FSWG and questioned Gašperlin about whether it is ‘normal to be involved as an expert in an investigation which involves you’ [e.g. experts from Frontex and national authorities invited to testify about pushbacks conducted by their colleagues]. Other parliamentarians also questioned the methodology used by the working group to validate the testimonies and information provided by national authorities implicated in pushbacks, causing Gašperlin to ultimately admit that doubting official information coming from the member-states would be detrimental to future cooperation. In a separate hearing, Fabrice Leggeri was asked whether sending letters to national authorities and asking for explanations without proper investigation was sufficient. He argued that when asking ‘in writing the highest level of authorities at national level to launch an investigation and they come back with the result of this investigation it is difficult, at least within the powers of the executive director, to go further’ (European Parliament, 2021g).

The final report of the working group did not recognize Frontex’s involvement in pushbacks. It only identified ‘deficits’ in monitoring and reporting systems that resulted in a situation where it was not possible ‘to clarify completely’ some of the incidents that it was called to examine (Frontex, 2021: 16). The report also emphasized that necessary changes in reporting procedures ‘should be combined with a newly introduced culture, in which failure is acknowledged and addressed’ (Frontex, 2021: 5). During one of the hearings, the newly appointed fundamental rights officer also acknowledged the need to establish a procedure to assess if certain incidents are miscategorized, implying however that miscategorization may not always be intentional and emphasizing that Frontex should adopt a ‘culture of learning’ that recognizes deficits in reporting processes (European Parliament, 2021f).

It is important to critically reflect upon this recognition of ‘deficits’ and calls to adopt a ‘new culture’ in Frontex. The information found in serious incident reports indicates that Frontex knew about the pushbacks in the Aegean Sea but did not engage in any meaningful actions to prevent and respond to them. Despite this, during the hearings, representatives of Frontex have never acknowledged the pushbacks and the agency’s involvement in them. Instead, they emphasized how they are willing to learn from previous mistakes and failures and, to address these, make related changes to their reporting mechanisms. Their failure to effectively report pushbacks at sea was repeatedly framed as a lesson to be learned and an opportunity for improvement, instead of evidence of Frontex’s involvement in pushbacks (for a discussion on failure framed as an instructive experience and a means of success, see Lisle [2018]). For instance, when members of the European Parliament asked questions about why pushbacks were not reported adequately, Gašperlin answered rather vaguely by referring to the recommendations that Frontex put forward to ‘optimize’ the functioning of the SIR mechanism (European Parliament, 2021c). Avoiding answering questions about the lack of adequate reporting and framing the detection of deficits as an opportunity for learning and optimization was another strategy employed to evade Frontex’s accountability and liability for pushbacks.

Conclusion

In this article, I explored how secrecy concerning pushbacks of refugees in the Aegean Sea was maintained and contested. Focusing on migration management allows us to discern forms of secrecy that differ from what one may encounter in more conventional security domains like military and intelligence operations. Secrecy in those domains often takes the form of non-disclosure and is framed as a necessary condition to protect national security. In migration management, enduring criticisms about fundamental rights violations and calls to protect migrant lives push border agencies to adopt measures that, in theory, make their operations more transparent. Paradoxically, however, those same measures can be used to produce secrecy. Different from other domains of (inter)national security, where secrecy is typically enacted through concealment, migration management – at least in the EU context – is characterized by a peculiar mode of secrecy enacted through transparency: transparency mechanisms are used to hide information that could function as evidence of fundamental rights violations.

In conversation with the literature that emphasizes the value of following public hearings and inquiries to study secrecy (e.g. Aradau, 2017; Gros et al., 2017; Walters, 2021), I proposed attending to the issues, actors, material witnesses and devices of dis/appearance implicated in the controversies related to Frontex’s involvement in pushbacks. This made it possible to showcase how secrecy’s performativity entails practices of hiding, the strategic recording and (mis)categorization of data, and the meticulous epistemic, sociotechnical work that goes into making information stand as evidence.

Beyond the case of pushbacks, concepts like material witnesses and devices of dis/appearance are useful for capturing the contentious and thoroughly material politics of secrecy. Retracing how oversight bodies and civil society actors enrol and speak on behalf of material witnesses makes it possible to understand how knowledge claims become credible in the process of holding security actors accountable, as well as how their credibility is contested through the mobilization of materially substantiated counter-evidence. Conversely, paying attention to devices of dis/appearance helps us to critically examine transparency measures that, we are told, contribute to the openness and democratization of security. Thinking through the concept of devices of dis/appearance implies that we should be paying more attention to how security authorities hide malpractices by selectively and proactively disclosing information. Disclosure can be a trap: by making certain information and elements pertinent to security appear, it can make injustices and violence disappear. Ultimately, a focus on material witnesses and devices of dis/appearance makes it possible to understand how security is, at least partially, made public through controversies that entail the dis/appearance, staging, negotiation and contestation of information based on which knowledge claims are raised by various actors involved in the (un)making of secrecy.