Towards democratic intelligence oversight: Limits, practices, struggles

Abstract

Despite its common usage, the meaning of ‘democratic’ in democratic intelligence oversight has rarely been spelled out. In this article, we situate questions regarding intelligence oversight within broader debates about the meanings and practices of democracy. We argue that the literature on intelligence oversight has tended to implicitly or explicitly follow liberal and technocratic ideas of democracy, which have limited the understanding of oversight both in academia and in practice. Thus, oversight is mostly understood as an expert, institutional and partially exclusive arrangement that is supposed to strike a balance between individual freedom and collective security, with the goal of establishing the legitimacy of and trust in intelligence work in a national setting. ‘Healthy’ or ‘efficient’ democratic oversight then becomes a matter of technical expertise, non-partisanship, and the ability to guard secrets. By analysing three moments of struggle around what counts as intelligence oversight across Germany, the UK, and the US, this article elucidates their democratic stakes. Through a practice-based approach, we argue that oversight takes much more agonistic, contentious, transnational, and public forms. However, these democratic practices reconfiguring oversight remain contested or contained by dominant views on what constitutes legitimate and effective intelligence oversight.

Introduction

I would definitely describe my work holding governments and intelligence agencies to account as a form of oversight … Activism, advocacy, litigation; it’s just a different language to talk about the same thing; it’s all various forms of oversight.¹

In 2013, the revelations by whistleblower Edward Snowden that intelligence agencies were routinely gathering and sharing data on citizens precipitated a crisis of legitimacy for the bodies in charge of holding these agencies to account. The reason why the disclosures were seen as subversive by many oversight and intelligence actors was not so much because, as many have claimed, they threatened ‘national security’. Even former NSA director Michael S. Rogers downplayed the impact of the Snowden disclosures on national security, saying that the ‘sky did not fall’ as a consequence of his actions.² Rather, as some observers have pointed out, the leaks made clear the structural failures of institutional oversight.³ From this perspective, the practices that Snowden documented were not an ‘aberration’, but a form of systemic abuse to which oversight structures were – at least to some degree – complicit.

How can we then understand the meaning of ‘democratic oversight’ amid such systemic failure? In this article, we propose to approach this question through the analysis of struggles around three elements of intelligence oversight. Firstly, who is included as a democratic protagonist of oversight? Should oversight be confined to formal bodies legally tasked with that role (parliaments, courts, or administrative agencies)? If not, who should be included? As one of our interlocutors pointed out, the forms of advocacy, activism and litigation in which they were engaged as a member of civil society in the aftermath of the Snowden disclosures were for them a de facto form of overseeing intelligence agencies. Secondly, what role does ‘secrecy’ as a defining feature of intelligence policy – and particularly the fact of being positioned within the so-called ‘ring of secrecy’– play? By consequence, can forms of ‘radical transparency’⁴ like public whistleblowing be seen as a means of legitimate democratic oversight, enabling disclosure, visibility and public debate? Thirdly, how does contestation shape how ‘democratic oversight’ is practised? To what extent does oversight rely on consensual practices, trust and impartiality?

The article takes these questions as a point of departure to interrogate three sites of non-official intelligence oversight – litigation, whistleblowing, and advocacy – to identify the dominant ways of construing ‘democracy’ and ‘oversight’, and challenges to those. Academic discussions of intelligence oversight have mainly taken place within the field of intelligence studies, a field historically derived from a strong Anglo-American state-policy lineage and grounded within functionalist, state-centric epistemologies.⁵ As Hager Ben Jaffel, Alvina Hoffmann, Oliver Kearns, and Sebastian Larsson have argued in a recent article, this sociological context has narrowed the scope of the field to the promulgation of ‘theories for, rather than of, intelligence’.⁶ Although the intellectual genealogy of intelligence oversight is slightly more heterogeneous than this diagnosis suggests, we argue that these debates have been underpinned by certain normative assumptions, centring on liberal, functionalist, and technocratic views at the expense of more radical and agonistic understandings of democracy.⁷ As we unpack further down, these assumptions in turn limit the range of democratic practices deemed to enact legitimate forms of intelligence oversight.

While the discipline of International Relations (IR) has seen numerous debates about theories and practices of democracy, an engagement with the kinds of democracy practised through intelligence oversight has been largely absent from intelligence studies.⁸ We propose to unpack the versions of democracy ‘at work’ in intelligence oversight in order to understand what other versions of democracy are silenced, left unthought, or excluded. Rather than starting with a taxonomy of democracy, we draw out the normative assumptions about democracy by attending to practices, thus bringing to the fore the limits of liberal democracy and the struggles for other forms of democracy.

The dominance of liberal understandings of democracy in intelligence studies can be seen as the consequence of two factors: the proximity of intelligence scholars to the agencies (either former intelligence officers or policymakers) and a wider trend in academia to increasingly focus on professional skills. A community of practitioner-scholars or ‘pracademics’ emerged from the common socialisation of intelligences scholars and practitioners.⁹ What Ben Jaffel and Larsson call ‘endogamy’ has come to structure the limits of the field of intelligence studies, either through formal outreach mechanisms such as the CIA’s Office for Academic Affairs,¹⁰ or through the establishment of visiting professorships on prominent intelligence programmes. Taken together, these dynamics explain the lack of ‘critical distance’ required to break with the pre-given notions of the intelligence field.

Most scholarly discourse on intelligence oversight has thus led to the disqualification of more agonistic critiques of intelligence, and more radical modes of oversight. To this day, the dominant frame of oversight as a well-ordered, institutionalised, and secret arrangement often masks how the history of intelligence oversight has largely been driven by scandals unleashed by whistleblowers, and struggles by activists or investigative journalists, with new oversight structures often being created in response to public pressure and following the delegitimation of intelligence agencies’ practices. An early prominent example is the establishment of the so-called Church Committee in the US, following several press revelations in the early 1970s. 1975 – often termed the ‘Year of Intelligence’ – marked a moment when intelligence oversight was institutionalised through various pieces of legislation and formal bodies. In the literature on intelligence oversight, the Church Committee is typically seen as having curtailed the power of US intelligence, setting the standard for other countries to follow. As Loch Johnson has put it, ‘the Church Committee did nothing less than revolutionise America’s attitudes toward intelligence supervision.’¹¹

However, the Church Committee cannot be seen in isolation from a decade-long series of scandals and radical opposition to the work of and abuses by intelligence actors.¹² In the view of such radical opposition – one that has been largely overlooked by intelligence studies –, the Church Committee also sought to re-establish consensus around intelligence through a legal framework that supposedly guaranteed that intelligence would now stick to the rule of law. But the new oversight professionals populating these structures came to view their work primarily as abiding by secrecy and representing intelligence within parliament. Their insertion in the realm of secrecy displaced the boundary between intelligence and its critics, excluding more radical engagements that appeared less legitimate. Soon enough, they could be co-opted by the executive branch to help build consensus around intelligence policy, passing regressive intelligence reforms and codifying expansive intelligence powers, construing their oversight role as a matter of checking conformity of rule by law, rather than rule of law.¹³ Although strong differences remain in the national histories of intelligence oversight, similar processes of scandal-driven institutionalisation took place in other countries like the UK, France, and Germany from the 1970s and 1980s onwards.

Taking as a point of departure practices of oversight rather than its policy representation, the article unpacks the everyday struggles involved in the constitution of ‘democratic intelligence oversight’. It reveals ways in which practising ‘oversight’ can take much more agonistic, contentious, transnational, and public forms than most of literature on oversight suggests, or that applicable policy frameworks acknowledge. As with the theoretical discussions on oversight, however, these heterogeneous practices of oversight are similarly contested or contained by dominant views on what constitutes legitimate oversight.

Methodologically, we approach oversight practices in situations of struggle in the wake of scandals about the activities of intelligence agencies. The analysis includes different national contexts (the US, Germany, and the UK), formal oversight bodies, and civic practices of disputing and challenging intelligence powers. Our choice of empirical sites is driven by attention to three modes of agonistic practices – litigation, whistleblowing, and advocacy – that bring to light three distinct limits and practices of liberal democracy: exclusion/inclusion, security/publicity, and contestation/consensus. While situated against national backgrounds, these practices emerged in the wake of the transnational circulation of the Snowden disclosures, public concern, and mobilisation about mass surveillance.¹⁴ The article also brings together authors with different disciplinary backgrounds and working on distinct empirical fields, which means that the methods pursued in our empirical research reflect this heterogeneity, combining archival analysis, textual analysis of legal documents, and oral history interviews with key actors involved in contesting surveillance legislation.

To trace what democracy does in these practices and struggles, we proceed in three steps. First, we examine the tension over what we call the ‘dual exclusion’ of who is regarded as a legitimate actor of oversight, and whose communication is deemed to deserve oversight. To unsettle these lines of inclusion/exclusion, we investigate strategic litigation by a transnational constellation of German non-governmental organisations (NGOs) and foreign journalists against the foreign intelligence service of Germany, the Bundesnachrichtendienst (BND). Second, we revisit the struggle over secrecy and publicity in democratic oversight by placing Snowden’s public disclosures of routine mass surveillance in relation to current whistleblowing and national security laws in the US. Third, we turn to the tension between contestation and consensus in what is considered legitimate and effective democratic oversight. The analysis of an advocacy campaign by a coalition of NGOs against the UK’s 2016 Investigatory Powers Act reveals how this tension plays out in practice. We conclude with a set of reflections on the practices of ‘democratic oversight’ and how our approach to oversight and democracy as practices offers both an agenda for research for intelligence studies and IR more broadly, and a political intervention in debates about meanings and practices of oversight.

Inclusion and exclusion: Pluralising oversight protagonists

The dual exclusion of actors and non-citizens

Who are the protagonists of oversight? By attending to who is seen as a legitimate oversight actor and who is not, we can unpack how oversight and democracy are intertwined through the dynamics of inclusion/exclusion. Different understandings of representative or participatory democracy underpin various oversight architectures and constellations of actors. These may vary with regards to the involvement of civil society actors and their potentially more contentious oversight practices, or the centrality of professional and legally established overseers.

The institutionalisation of oversight went along with another, generally accepted, form of exclusion. In most Western democracies, data collection on foreigners has been either completely or largely excluded from established regulations of intelligence and oversight structures.¹⁵ By concentrating intelligence laws and their oversight on national citizens or national territory, it not only determined who is entitled to make claims against the surveillance of their communication, but also who and whose communication deserves to be subjected to oversight to begin with. Being considered ‘fair game’, non-nationals have been the primary targets of large-scale surveillance but not subjects of protection through institutional oversight.

This dual exclusion is by and large mirrored in the academic literature on intelligence oversight.¹⁶ Oversight scholars have indeed remained largely silent on the foreign-national distinction in intelligence practices and oversight, replicating a territorialised understanding of liberal democracy. After the Snowden disclosures, it was legal scholarship that either opposed the exclusion of foreigners from oversight by promoting a human rights approach,¹⁷ or asked for granting at least some protection to foreigners.¹⁸ With respect to the exclusion of unofficial oversight actors, many intelligence scholars iterate the historical trajectories of oversight by emphasising formal, public bodies as oversight protagonists.¹⁹

Some authors locate civil society as external to the official oversight system of legislative, executive, and expert bodies, stating that civil society organisations and the media play some role in oversight, without further definition.²⁰ Here, democracy veers towards representative institutions and balance of powers, with oversight understood in terms of scrutiny or control. Other intelligence scholars employ a broader understanding of oversight, integrating practices of civil society actors as ‘public oversight’,²¹ ‘informal oversight’,²² or ‘civil society oversight’.²³ Civil society actors are envisaged to provide input for official oversight, restrain intelligence agencies’ power and offer a secondary accountability mechanism to scrutinise the overseers’ activities.²⁴ In this literature, oversight comprises different levels or layers, with so-called ‘soft’ or ‘informal’ oversight framed as the outermost layer, while remaining within the same space of formal oversight institutions. These informal overseers become intermediaries acting on behalf of and being responsive to a wider (and rather abstract) public.²⁵ Others even see them as a means for the political participation of citizens.²⁶ It is this ‘acting for’²⁷ that necessitates the inclusion of leverage under the direct influence of the public, namely through elected representatives.²⁸

More recently, several authors have criticised the focus on representative institutions as too narrow and bureaucratic in comparison with wider democratic ‘accountability’, which can entail ‘assertive verification in advance of proposed action, or report or correction once an action has been taken’.²⁹ As the transnationalisation of intelligence agencies has limited formal accountability structures in many countries,³⁰ Richard J. Aldrich has described civil society as the sole locus of transnational efforts to control the agencies, noting that ‘accountability now seems to flow from a globalised network of activists and journalists, not from parliamentary oversight committees’.³¹ Understood in this broader sense of ‘accountability’, oversight mobilises a wider array of actors. Such flexible conceptualisations see an ‘informal network of researchers, journalists, and lawyers in civil society’ acting to some extent in symbiosis with formal oversight,³² positioning civil society in a loose configuration with official oversight actors, while entrenching a temporal dimension in which formal oversight comes first. Consequently, it cannot but act as a ‘fire alarm’ to, as opposed to a ‘police patrol’ of, intelligence practices.³³ This temporal distanciation is supplemented through a spatial distanciation, as these actors are situated on the margins of the social space inhabited by formal oversight actors, which means that actors acting from outside oversight institutions are only partially included – or putting it differently, they continue to be partially excluded. Ultimately, representative democracy comes first, with participatory and agonistic forms of democracy as supplementary and secondary.

Despite acknowledging that actors from civil society play a role in intelligence oversight, there have been no studies of the practices of actors such as the media or NGOs through the lens of oversight. While the oversight literature has acknowledged that courts play an increasingly important role in intelligence oversight, and that these judicial corrective practices depend on litigation to get underway,³⁴ there has been almost no attention to how legal challenges to the dual exclusions we have located can contribute to oversight as democratic practice. Yet, strategic litigation – the collective mobilisation of law by civil society actors – can perform a watchdog function, advocating for marginalised groups, and stirring public discourse.³⁵ In the following section, we analyse strategic litigation by two German NGOs to account for practices of oversight that challenge the dual lines of inclusion/exclusion outlined above.

Struggles over inclusion/exclusion in practice: The case of litigation

In 2017, two German NGOs, Gesellschaft für Freiheitsrechte e.V. (GFF) and Reporters without Borders (RSF Germany) as well as six investigative journalists from different countries lodged a case against the intelligence law that authorises the BND to conduct foreign intelligence. The plaintiffs challenged the hitherto largely unchecked surveillance of foreigners and particularly the attachment of protection against surveillance to nationality and state territory. They argued that what was at stake was not only the privacy of communications but also press freedom, since the BND law did not foresee protective mechanisms for journalists and their sources.³⁶ Referring to international law, the collective of NGOs and journalists argued that foreigners may claim fundamental rights vis-à-vis German authorities, since the latter are bound by constitutional law when acting on behalf of the German state, no matter where these actions take place. The German government and the BND defended the opposite view, claiming that the scope of applicability was limited to national territory.³⁷ In May 2020, the constitutional court ruled against the government, requiring reform of the intelligence law. This decision against the status quo was made despite former intelligence officers publicly disqualifying the claimants as ‘fools’ (‘Hansel’)³⁸ and framing the lawsuit as a national security threat.³⁹ The case entails two related struggles over inclusion/exclusion: who is allowed to participate in oversight and who deserves protection, and thus oversight.

For decades, the rationale ‘we only spy on foreigners’ has become entrenched in the logic of mass surveillance in liberal democracies. This rationale helped intelligence agencies justify blanket data collection and enabled the partly automated exchange of indiscriminately collected data of foreigners among intelligence agencies of different countries.⁴⁰ As a consequence of foreigners’ exclusion from protection, their data has become a currency in the transnational economy of surveillance. Rather than being clearly articulated in a legal statute, the distinction between foreigners and nationals has operated as a taken-for-granted legal interpretation and as a practice that has been institutionalised in the organisation of intelligence, technical surveillance infrastructures, and oversight regimes. Therefore, most of the surveillance conducted by agencies like the British GCHQ, the German BND, and the French DGSE is subject to much looser independent oversight than domestic surveillance.

Following the Snowden disclosures, a discourse emerged through transnational constellations of actors that transformed the blanket interception of foreign data from a silent, institutionalised practice of the field to a publicly contested principle. In the immediate aftermath of the first Snowden leaks, a transnational network of NGOs published a working version of the ‘Necessary and Proportionate’ principles, providing an analysis of international human rights and its application to communications surveillance. The text made clear that human rights applied ‘extraterritorially’, meaning that they covered both domestic surveillance conducted on the residents and foreign nationals whose data might be scooped in the context of large-scale surveillance.⁴¹ The text was presented at the UN Human Rights Council in Geneva in September 2013 and subsequently endorsed by more than 400 organisations across the world.

After this initial push, 2014 was marked by several ‘critical moments’,⁴² that is, moments of dispute that required justification and contributed to breaking the silence on the previously taken-for-granted assumption of foreigners being ‘fair game’ for mass surveillance. In the US, the Obama administration adopted the ‘Presidential Policy Directive 28’ (PPD28), which promised to grant privacy protections to all humans ‘regardless of their nationality or wherever they might reside’.⁴³ Following an initiative by Germany and Brazil and the engagement of various NGOs, the UN published its resolution ‘Right to Privacy in the Digital Age’, which reconfigured data subjects as data citizens through the language of human rights.⁴⁴ In Germany, the critical moment took place through a confrontation of the BND’s practices with the legal discourse of constitutional lawyers in the parliamentary ‘NSA inquiry panel’. The right to privacy, Article 10 of the German constitution, protects all humans, the lawyers claimed.⁴⁵

Yet, after its reform in 2016, the revised BND law did not take this constitutional and human rights perspective into account. With a growing transnational discourse on the foreign-national distinction providing momentum, GFF and RSF Germany challenged the foreign-national distinction through the litigation process. While the question of who enjoys privacy rights and who deserves oversight could be relevant for everyone in a transnationally connected world, they chose to focus on consequences for journalists. Due to their work on sensitive issues like corruption and other forms of abuse, the plaintiffs suspected that they had been of interest to intelligence agencies – not necessarily by the BND itself, but its foreign partner agencies. They were concerned about the consequences of uncontrolled data sharing among intelligence agencies from different countries, both for their own safety and the protection of their sources. By tackling this mix of unregulated gathering and sharing of foreign communication by the BND, the claimants illustrated the transnational implications of data collection and expanded the democratic values that are at stake under the BND surveillance, from privacy as a universal right to press freedom.

The ruling of Germany’s constitutional court was a landmark decision. The judges acknowledged the vulnerable situation of journalists and lawyers abroad, demanding a quasi-judicial oversight body to authorise the surveillance of these professional groups as well as the sharing of insights about them with foreign state authorities. However, regarding the foreign-national distinction, the judges adopted an ambivalent interpretation. On one hand, they agreed that neither privacy nor press freedom are bound to nationality. On the other hand, the ruling ultimately legalised the distinction foreign-national, as it justified lower standards of protections for foreigners.

The judges deduced the theoretical inclusion and practical exclusion of foreigners from protective safeguards in two steps. First, they followed the claimants’ line of argumentation to include foreigners in the realm of basic rights by anchoring the applicability of the German constitution not in the location of the object of state action, but the state actor itself. Like a leash, there is a binding effect of fundamental rights for state actors acting beyond borders, including intelligence agencies, with a claim to democratic legitimacy.⁴⁶ Nonetheless, the judges argue that, in contrast to foreigners, German citizens are to a greater extent exposed to interventions by German state authorities, and therefore more easily subjected to follow-up action when surveilled by the BND.⁴⁷ Regarding foreigners abroad, the BND’s lack of ‘operational powers’ in conjunction with spatial distance is seen as a buffer that lessens the potential impact of communications surveillance on foreigners abroad.⁴⁸ What follows from this line of argument is a differential treatment that leaves foreigners with little benefit from their newly granted right to private communication. For instance, in contrast to nationals, foreigners are not deemed ‘notification worthy’ since they could not seek legal remedy and their notification would allegedly not foster democratic discourse on communications surveillance.⁴⁹

The litigation showed how established boundaries and mechanisms of exclusion embedded in intelligence oversight can be partially re-opened and re-negotiated. First, the exclusion of foreigners from oversight was declared unconstitutional. Yet, boundaries are not dissolved but displaced, as the foreigner-national distinction is upheld and legalised. Second, civic collectives used strategic litigation to circumvent established oversight bodies, becoming themselves oversight actors against delegitimising strategies of intelligence and government officials. The transnational constellation of claimants – NGOs and affected journalists from abroad – transcended boundaries of who was allowed to make claims about surveillance and reversed the primacy of representative over participatory democracy. In 2015, a lawsuit by RSF Germany had been rejected on the grounds that, as national claimants, they were not able to construct a case of affectedness. Foreigners alone had also not been able to challenge BND’s surveillance. The strategic alliance of nationals and foreigners was able to address a vacuum of oversight that transnational surveillance by intelligence agencies had created and upheld. This transnational push can be seen as part of what Alvina Hoffmann has described as ‘a new imaginary and set of resistance practices’, in which civil society facing surveillance by intelligence agencies has started to make claims with reference to universal rights, ‘not just as citizens of their own country’.⁵⁰ These practices are connected to a second tension, that between secrecy and publicity, to which we now turn.

Secrecy and publicity: Whistleblowing as democratic practice

The ‘circle of secrecy’ in oversight and intelligence

Intelligence is a field of state secrecy par excellence, where secrecy reinforces the exclusion of outsiders by depriving them of knowledge about the reality of intelligence work. This secretive nature is at odds with demands for control of governmental conduct and publicity, rendering intelligence oversight a special oversight case.⁵¹ In intelligence studies, the answer out of this conundrum of democratic values then usually becomes that of a ‘balance’ to be reached between secrecy and publicity.

This idea of ‘balancing’ secrecy and publicity has, in recent years, been subjected to greater scrutiny within critical IR literature on security.⁵² Within these conversations, secrecy is conceptualised not in opposition to publicity, but as a mutable and fluctuating category of international politics, able to reorganise sociopolitical relations in particular ways.⁵³ What William Walters has tentatively described as a ‘secrecy turn’ in the study of security practices has seen secrecy examined as a social space,⁵⁴ in relation to subjectivity,⁵⁵ as a form of non-knowledge,⁵⁶ and as a terrain to navigate in relation to methods and methodology.⁵⁷ Far from seeing secrecy as antagonistic to publicity, this literature reconceptualises the relation between the two ‘poles’ as a dynamic terrain of contested knowledge practices.⁵⁸ Following Walters’ invitation to ‘inject mobility, struggle, and material transformation into the way we theorize secrecy’,⁵⁹ we approach whistleblowing as struggles over the relations between secrecy, publicity, and democracy. ‘Public’ whistleblowing is a democratic practice whereby insiders with access to secret information ‘go public’, bringing that special knowledge – for example, knowledge about abuse committed by intelligence professionals – to the media, NGOs, or lawyers who can then further investigate, advocate, and/or litigate.

At least since the 1960s, whistleblowing has been a key driver in enabling public debate around intelligence. But where special laws were adopted to regulate whistleblowing in the field of national security (first in the US and much more recently in France), it was confined to institutional channels, effectively limiting the role of whistleblowers to that of ‘organisational defenders’ rather than public advocates against intelligence abuse.⁶⁰ Such reports to institutional oversight institutions, however, have proved unable to generate reform or address abuse, which might be partly explained by the fact that these entities address governmental institutions, not publics.⁶¹ In other words, struggles about whistleblowing in intelligence affairs constitute another stage where the opposition to agonistic and more participatory democracy plays out.

Therefore, governments have tried to delegitimise or suppress public whistleblowing under laws around state secrecy and counterespionage. Take, for instance, a piece of legislation adopted in the US: the 1998 Intelligence Community Whistleblower Protection Act. This law was introduced not to protect insiders going public but ultimately to reinforce the protection of secrecy, making public whistleblowing a federal crime for those with access to classified information. This piece of legislation was sponsored by Porter Goss, former CIA agent who served as chairman of the House Permanent Select Committee on Intelligence between 1997 and 2004. In his congressional role, Goss can be seen as what intelligence scholar Loch K. Johnson calls a ‘cheerleader’ for the agencies, by which Johnson means a type of overseer who promotes the rationale, demands, and overall interests of intelligence agencies.⁶² After being appointed by George W. Bush to head the CIA, Goss would stick to the same line regarding whistleblowing:

those who choose to bypass the law and go straight to the press are not noble, honorable or patriotic. Nor are they whistleblowers. Instead they are committing a criminal act that potentially places American lives at risk. It is unconscionable to compromise national security information and then seek protection as a whistleblower to forestall punishment.⁶³

One might object that Goss and others have simply sought to defend legislation that regulates official secrets. However, as many journalists have pointed out, elected officials have no problem leaking sensitive information as long as they obtain political credit from it,⁶⁴ with the US Department of Justice showing far less interest in these leaks. These efforts to criminalise public whistleblowing undermine the ability to denounce fundamental rights violations.

In the academic literature on intelligence oversight, several authors have offered more nuanced justifications for this historically sedimented structure of exclusion. While acknowledging systemic oversight failure, Mark Phythian bemoans that, with public whistleblowing, ‘we are dependent on a moral compass emanating from sources outside of government.’⁶⁵ Recently, Melina Dobson has also expressed worry about a ‘continuing trend to “publish and be damned”’.⁶⁶ Johnson adopts a similar view. While acknowledging that ‘far too much information – some 85 percent – is unnecessarily classified by intelligence and military bureaucrats in the first place’ and that ‘the significance and danger of leaks have been exaggerated’,⁶⁷ he still frames public whistleblowing as illegitimate. For him, the priority lies in improving internal whistleblowing channels so as ‘to make sure whistle-blowers have a chance to make their case in a responsible manner, without having to go to jail or abandon their country’.⁶⁸

Some intelligence scholars, however, recognise that public whistleblowing can address oversight failures. Damien Van Puyvelde writes that ‘whistleblowers are particularly valuable because they provide alternative sources of information that, by definition, are not controlled by the government.’⁶⁹ For Claudia Hillebrand too, ‘media outlets can provide a channel for leaking information that might not have been taken into account by formal oversight bodies, or when individuals felt unable to approach formal oversight bodies and instead approached journalists.’⁷⁰ Finally, Aldrich points out that ‘regulation by revelation’ has seen activists and media pressure groups performing a de facto, albeit problematic oversight role, in that ‘these organisations have no democratic mandate and are not concerned with effectiveness.’⁷¹

While some authors advocate granting oversight bodies the power to declassify information and to hold public hearings on intelligence matters,⁷² many of those we interviewed dismissed the desirability and even the legitimacy of these oversight bodies going public. Overall, the focus lies in ensuring better access for oversight agencies to information exchanged within intelligence agencies. This is the Janus face of access: it expands the inner circle of secrecy and closes off that space for everyone else. Thus, oversight officials become the guardians of secrets. This idea is a common, even if sometimes subtle, thread in oversight literature. This is, for instance, the case when authors write that ‘in the USA the requirement that all involved [in oversight bodies] sign non-disclosure agreements neuters most effective exposure should wrongdoing be detected’,⁷³ or that ‘the ability of the Committee to demonstrate the basis for its conclusions is restricted by the fact that it operates within the ring of secrecy, and does not itself have the competence to declassify secret information.’⁷⁴ Generally, secrecy is normatively defended (e.g., ‘the committees have proven themselves reliable keepers of the nation’s highest secrets’;⁷⁵ ‘after September 11, 2001, it became widely acknowledged that a legitimate requirement for secrecy exists’).⁷⁶

The struggle over secrecy and publicity: The Snowden case

When Snowden decided to go public, he knew that he would be subjected to many attacks on the part of US state actors and media. The cases of Chelsea Manning, Thomas Drake, and other intelligence whistleblowers who had faced prosecution provided a clear indication that he would be demonised and even risk incarceration. Along with these recent cases, official warnings by intelligence professionals had long signalled what would happen to those who break the rule of secrecy. Take, for instance, the warning by Goss in 2006 to ‘investigate these cases [of unauthorised disclosures] aggressively’.⁷⁷

While attacks by prominent intelligence officials were to be expected, a more surprising reaction was that of many leading media who attacked Snowden, particularly in the US and the UK. There were two main kinds of disqualifications: those that accused Snowden of endangering national security ‘beyond repair’, and those that rather focused on particular personality traits, insisting on his pathological solitary character⁷⁸ or his ‘narcissistic’ personality.⁷⁹ The common thread to both types of attacks was clear: Snowden was no hero and deserved no praise because he did not follow the proper channels to raise his concerns.

These efforts to delegitimise Snowden were echoed among intelligence scholars. When six members of the Church Committee wrote a letter to then President Barack Obama asking him to pardon Snowden, arguing that the ‘lack of disclosure can cause just as many, if not more, harms to the nation than disclosure’,⁸⁰ one of its most visible representatives, Loch K. Johnson, refused to sign. While recognising Snowden’s contribution to public debates, Johnson considered that he did not use the proper institutional channels and, more importantly, went too far in releasing ‘granular details about intelligence budgets and very sensitive programs’.⁸¹

In more than a dozen interviews that our team conducted, intelligence scholars argued that Snowden endangered national security and that whistleblowers should have gone through official channels. They considered it too dangerous to defer to the whistleblower’s moral standards and to journalists to decide what was of legitimate public interest. However, such disqualifications of public whistleblowing by virtue of the unreliability of their individual values contradicts the same scholars’ insistence that, at the end of the day, the effectiveness of oversight institutions largely lies in overseers’ own personality and ethics.⁸²

Snowden’s disclosures did not only undermine the legitimacy of intelligence agencies. They also, albeit less directly, delegitimised oversight bodies like parliamentary committees, secret courts like the Foreign Intelligence Surveillance Court (FISC), and the people working in these institutions. If, as most intelligence scholars argue, overseers have only worked as reactive ‘fire fighters’ instead of preemptive ‘police patrols’, Snowden started a fire that forced these officials to face their own shortcomings. As Snowden recalls in his memoirs, he had sought to report these abuses internally only to be turned down by his superiors.⁸³ Therefore, his case proved that insiders worried about the vast expansion of surveillance powers could not hope to see these practices reformed without relying on external public pressure. It highlighted the structural failures of institutional oversight.

Long-standing attempts at containing whistleblowing to institutional channels neglects the history of public investigations and their role in keeping intelligence in check. With no press, there would have been no ‘Year of Intelligence’ in 1975, nor many other fundamental public debates about intelligence policy, violence and abuse. This is why, in contrast to dominant views in the field of intelligence oversight and against governments’ attempts at suppression, the right to public whistleblowing has been claimed time and time again. When Daniel Ellseberg leaked the Pentagon Papers in 1971 and the Nixon administration sought to prevent their publication, Supreme Court Justice Hugo Black stressed that:

[i]n the absence of the governmental checks and balances present in other areas of our national life, the only effective restraint upon executive policy and power in the areas of national defense and international affairs may lie in an enlightened citizenry – in an informed and critical public opinion which alone can here protect the values of democratic government.⁸⁴

A few weeks after Snowden’s first disclosures, the European Court of Human Rights also stressed the importance of the right for whistleblowers to ‘go public’ with public-interest information. The case at hand related to an intelligence agent who revealed in the press widespread practices of illegal political surveillance of communications by the Romanian intelligence service.⁸⁵

Further initiatives took place transnationally, as NGOs and academic experts issued the 2013 Tshwane Principles on Transparency and National Security,⁸⁶ which were reproduced by the Council of Europe in a resolution on ‘national security and access to information’.⁸⁷ Principle 37, for example, recalls the need to protect whistleblowers for reporting a wide range of abuses and other ‘wrongdoing’ that they witness, both in the context of internal procedures (Principle 39) as well as in the context of public disclosures – for example via the press. This is particularly the case when, following an internal alert, the ‘person has not received reasonable and appropriate results within a reasonable time’ or where ‘the person has reasonable grounds to believe that there is a significant risk that an internal disclosure and/or disclosure to an independent oversight body will result in the destruction or concealment of evidence, interference with witnesses or retaliation against the person or a third party’ (Principle 40).

Current laws surrounding whistleblowing in national security contexts remain disconnected from these international standards, effectively creating a chilling effect for potential whistleblowers. Similarly, we have not seen these principles taken up, or even discussed, in intelligence studies. This gap underlines the fact that, although it is arguably one of the most important forms of oversight over intelligence abuse, public whistleblowing remains a contested practice, one that is effectively repressed and delegitimised by dominant approaches to intelligence oversight. The struggles around whistleblowing illustrate a clash between agonistic democratic claims and the defence of the prevailing consensus in intelligence affairs, one that most intelligence studies and state officials work to protect.

Whistleblowing asks us to revisit democratic tensions between secrecy and publicity and reformulate publicity beyond the discourse of the balance between security and privacy, secrecy and transparency. By approaching whistleblowing as a practice of ‘going public’, we have shown how democratic publics are not pre-given or limited to electoral moments, but enacted by challenging the boundaries of secrecy and revealing the failures of oversight institutions.

Consensus and contestation: Limits on civil society engagement

Consensus through impartiality and trust

Past scandals and ensuing legitimation strategies have led to a widespread view that radical critiques of intelligence agencies are illegitimate. For instance, this was the case in the US, where anti-war activist engagement with intelligence policies in the 1960s and 1970s came to be disqualified and delegitimised. However, the US has not been unique in this regard. In the UK, for example, with the establishment of the first Intelligence and Security Committee (ISC) in the Parliament in 1994, inaugural chairperson Lord Tom King described in his memoirs the need to ensure that parliamentarians (MPs) selected to serve on the committee were not ‘ideologically predisposed’ to an anti-agency viewpoint.⁸⁸

That state of play is by and large reflected in the academic literature on intelligence oversight, which tends to privilege a consensual view of democracy. Scholarship in intelligence studies often shares a presupposition that oversight bodies have to collaborate with and not confront the agencies.⁸⁹ As Anne Karalekas, author of one of the first books on the CIA, puts it, ‘The intelligence committees are heavily dependent on the agencies for the information required to execute their oversight responsibilities, creating strong incentives to establish cooperative relationships.’⁹⁰ Even when contestation is acknowledged, as in the conflict over the definition of democratic values like transparency, or between courts and intelligence agencies, it is integrated within an architecture of consensus and largely limited to the institutions of representative democracy.

This architecture of consensus takes two forms. One of these is trust understood as a mediator of relations between oversight bodies and intelligence agencies. Scholars writing at the intersection of intelligence studies and intelligence policy often hold this view. For instance, Anthony Glees and Phillip Davies, both university professors and frequent media commentators on intelligence matters, argue that the ISC must win ‘the trust of the secret agencies, and in particular their heads, in order to [be able to] “oversee” them’.⁹¹ This alleged need of trust suggests both that oversight is in a position of inferiority vis-à-vis the security and intelligence services (who can provide or deny access to their workings) and that a relation of companionship may arise between these two services. As echoed by Fred Schreier, the ‘critical issue of oversight is the balance between committee independence and criticism on the one hand, and the maintenance of a working relationship between the committee and the intelligence agencies on the other hand’.⁹² As we have seen with the delegitimation of whistleblowing as a challenge to the taken-for-granted ‘circle of secrecy’, trust promotes a consensual understanding of democracy, where contestation is seen as unproductive and conflict to be avoided.

Trust also represents oversight as politically neutral or impartial. This has led some authors on oversight to highlight and argue that cases in which oversight was politicised entailed negative effects. For example, Johnson describes the decade between 1992 and 2001 as the ‘partisan’ era, showing how power struggles of political parties have negatively affected the control of agencies.⁹³ In this vein, the political partisanship of oversight is assumed to undermine its effectiveness, since finding a common ground for investigations is harder and intelligence officials might doubt the intentions of political actors turned overseers.⁹⁴ As Gill points out, in the 1960s and 1970s there was a widespread fear that ‘legislatures would not be appropriate, for example because of their tendency to partisanship and to leak information for political advantage’.⁹⁵ It is against an agonistic understanding of democracy that understandings of neutrality, apolitics, or impartiality promote an aura of deliberation and came to be seen as desirable. Subsequently, these were supplemented by an emphasis on trust. In discussing congressional oversight in the US, Jennifer Kibbe goes as far as calling for ‘appointing intelligence committee chairs who are moderate, responsible, dedicated and committed to the notion of nonpartisan oversight’.⁹⁶ These arguments are also echoed in reference to European oversight, where scholars caution against the dangers of parliamentary scrutiny as the ‘security sector may be drawn into party political controversy – an immature approach by Parliamentarians may lead to sensationalism in public debate, and to wild accusations and conspiracy theories being aired under parliamentary privilege’.⁹⁷

Given these assumptions about political impartiality and the need for trust relations, it is not surprising that the understanding of democratic politics as consensual is extended to civil society. As we will see further down in the analysis of the ‘Don’t Spy on Us’ campaign in the UK, more conflictual forms of oversight come to be disqualified. This was also the case of the media, which was sometimes framed in a rather suspicious light as it might be leveraged for partisan power struggles.⁹⁸ However, we have seen that civil society actors can be a ‘surprisingly effective sentinel’ driving inquiries in intelligence activities and calls for public accountability.⁹⁹ While this may be true in some contexts, the stance that often dominates in the literature fails to problematise the limits of established oversight agencies, both theoretically and practically. Not only does it overlook the process whereby accountability is triggered (publicly politicising wrongdoing through a scandal); it also dismisses most conceptions of democracy and democratic politics as a locus for conflict. In so doing, these views merely reflect existing power relationships in the actual practice of intelligence oversight, as our final case illustrates.

Contestation and its constraints

The ‘Don’t Spy on Us’ Coalition came together to contest the new UK legislation, the Investigatory Powers Bill, which was subsequently passed into law in 2016. A coalition of NGOs campaigning for privacy, freedom of expression, and digital rights, Don’t Spy On Us made a series of recommendations for legislative overhaul following the Snowden revelations. The purpose of forming a coalition was to ensure that arguments were gaining maximum traction, that goals were aligned and strategically communicated, and that a consistent message was formulated. The campaign’s aims were twofold: raising public awareness of the harms of mass surveillance legalised and extended by the Investigatory Powers Bill, and lobbying parliamentarians to amend the bill along specific lines. However, after the bill was passed, the Don’t Spy On Us Coalition disbanded, leaving a sombre epitaph: ‘The UK Parliament has passed the Investigatory Powers Act, the most extreme surveillance law in our history.’¹⁰⁰

The advocacy practices of the Don’t Spy On Us coalition embodied both conflictual and consensual styles of democratic practice in its campaigns and within legislative struggles. Their practices shed light on another limit of what counts as ‘democracy’ in intelligence oversight, namely the role of more radical contestations in democracy. Drawing on interviews with actors involved in this coalition, as well as MPs, peers, and expert witnesses who engaged with these NGOs, we trace how dynamics of contestation transform into consensual practices through the foreclosing of debate around specific sites. As the campaign progressed, conflictual modes of engagement that resonate with agonistic and radical democratic approaches gave way to more consensus-based advocacy. This is partly because contestations of mass surveillance can be seen as constrained within certain dynamics, parameters, and ‘norms of sayability’, which dictate what could be accepted as ‘realistic’ or ‘legitimate’ critique by other actors, including members of civil society themselves.¹⁰¹

At the same time, these advocacy practices lent credibility to the idea, which intelligence services, the government, and official oversight actors endorsed, that the UK was setting a ‘global gold standard’ of surveillance legislation. At the start of the campaign, Don’t Spy On Us agreed on six demands for UK legislation on surveillance: no surveillance without suspicion; transparent laws, not secret laws; judicial not political authorisation; effective democratic oversight; the right to redress and a secure Internet for all.

While mobilising key principles of liberal democracy around the rule of law, transparency and separation of powers, the framing of this initial contestation of mass surveillance was increasingly limited in two ways: first, around what claims were deemed ‘realistic’, and second, around what claims were deemed ‘legitimate’. One of the initial cleavages as the coalition came together was around formulating a strategic position: did the coalition want to engage and improve safeguards, or try to kill the entire practice of mass surveillance altogether? One member of the coalition we interviewed remembers this to be the single most contentious issue throughout the passing of the bill through Parliament.

For Don’t Spy On Us, the legislative struggle over the IPA came after a previous legislative victory of sorts against extending state surveillance. The 2012 Draft Communications Data Bill had been thrown out after being vetoed by then Deputy Prime Minister Nick Clegg. Within this campaign, NGOs had argued that older, much broader legislation was out of date and that new legislation was needed to better guard against abuse by the agencies. As one of our interlocutors explained,

Everybody called for the IP Act effectively; they called for a better version of RIPA [the (old) regulation of investigatory powers act]. So, you can’t scrap that. All you can do to my mind is improve it, improve transparency, improve oversight, improve mechanisms so that the wins are going to be very slight.¹⁰²

After the Snowden disclosures, a review of the use of bulk powers in the UK by David Anderson, the Independent Reviewer of Terrorism Legislation, deemed these powers of ‘vital utility’ to security and intelligence agencies, the use of which could not be matched by data acquired through targeted means.¹⁰³ His main issue with RIPA was that it was ‘incomprehensible to all but a tiny band of initiates’.¹⁰⁴ To occupy a position of trying to scrap the powers altogether was seen as somewhat unrealistic from the outset:

[There were] people [civil society activists] who felt that they would win the argument through the sheer conviction that they were right on a moral level, which anybody who’s worked in politics for more than a day knows is wrong.¹⁰⁵

When I went to Parliament to say to people who want to work in Parliament, do you want to obliterate every boulder? Or are you prepared to just chip away at the boulder so that you might be able to squeeze round it to get to the other side of the path? And anyone who said I want to obliterate the boulder I knew was not cut out for this.¹⁰⁶

In this demarcation of what was deemed a ‘realistic’ position for the government to engage with, opposing mass surveillance was reduced to tactical dimensions of safeguards and limitations. Some campaigners would refuse to engage with specific sections of the bill on bulk data collection, deciding instead to brief backbench MPs on specific language they could use to temper some of the more wide-ranging powers.¹⁰⁷

Struggles around what counted as ‘realistic’ also took place around public advocacy and campaigns, with different imaginaries of ‘the public’ enacted to mobilise public opinion against the bill. A widely circulated poster campaign likened then UK Home Secretary Theresa May to well-known dictators, such as Putin and Xi Jinping, calling on her to ‘stop giving [them] ideas’.¹⁰⁸ Appealing to critiques of surveillance based around the totalitarian-democratic binary was seen as ‘out of touch’ by fellow campaigners, who argued that a campaign based around government incompetence and fear of the ‘tax-man’ having access to this data would be more effective:

The poster campaign they ran was just inept! The public don’t respond well to being told that their government is like China and Russia, because it’s not, it’s nonsense. And I think it was just embarrassing that this went ahead.¹⁰⁹

Advocacy around the Investigatory Powers Bill embodied conflictual and agonistic understandings of democracy. It also raised questions about what mass surveillance means for understandings of democracy – does liberal democracy have the tools to hold it in check, or does it risk morphing into illiberal or even totalitarian forms?

However, contestation was also constrained by who or what was deemed to be a legitimate actor. For instance, in the evidence submitted to the Intelligence and Security Committee of the UK Parliament, former GCHQ Director David Omand deemed the reactions to the Snowden disclosures ‘a quite unnecessary moral panic over privacy’ and strove to clearly distinguish what he called ‘bulk access to the internet’ from ‘mass surveillance’.¹¹⁰ Omand’s play with categories and claim of ‘category error’ was successful to the extent that both Tory and Labour MPs came to reject ‘mass surveillance’ in debating the Investigatory Powers Bill only a few years later.¹¹¹

More agonistic understandings of democratic practice were thus reserved for public campaigning. However, actors who engaged in more public forms of advocacy were often deemed illegitimate by MPs, peers, and some expert witnesses. Campaigners who held a more radical message were delegitimated as ‘sensationalist’ or considered to instrumentalise the debate on behalf of the NGOs to gain more funding. In this vein, particular campaigning and highly visible strategies were deemed as ‘street theatre’ or ‘self congratulatory’ and lacking nuance.

I mean, you have obviously got people who are more active and rather keen on the publicity aspect of it. But there are others who are going to take a more nuanced and thoughtful approach. You know, that is that you have to speak to the detail of it.¹¹²

Rather than an integral aspect of agonistic democracy, publicity was equated with performance and spectacle. Parliamentarians involved in the Investigatory Powers Bill debates mentioned taking care with formulating their interventions in language which did not connect them to particular groups which were deemed ‘fringe’, which they argued would delegitimise their intercession. MPs and peers trying to limit these data collection and retention powers spoke about having more credibility with fellow parliamentarians if they adopted a position of being in dialogue with the needs of the security services rather than presenting arguments put forward by civil society, particularly civil society groups deemed too radical or extreme.¹¹³

The advocacy practices of the Don’t Spy on Us coalition show how more conflictual and agonistic versions of democratic practices become constrained within parameters which narrow the terms of engagement and reflect dominant understandings of what is considered ‘realistic’ and ‘legitimate’ in liberal terms of rule of law and institutional arrangements. They are also indicative of the fact that democratic contestation is not easily opposed to consensus, but that various actors operated at the interstices of more contestatory or more consensual politics. However, normative assumptions about consensual democracy, and the delegitimation of actors as radical or too ‘unrealistic’, limit the form and content of contestation. This resonates with understandings of effective oversight being seen as apolitical within much of the literature. Although members of the coalition took pride in making some gains (particularly around the inclusion of a judicial ‘double-lock’ mechanism before certain powers can be used), many took a more ambivalent stance, describing these struggles as a moment in time, part of the ever-shifting relations between freedom, democracy, and surveillance.

Conclusion

Taking as a point of departure the diverging answers to the key question of what makes intelligence oversight democratic, this article has focused on practices that contest mass surveillance by intelligence agencies across various national settings. Our aim has been to make a two-pronged contribution to critical approaches in intelligence studies and international relations more broadly.

By contrasting the dominant ways of construing ‘intelligence oversight’ as democratic in the academic and policy literature with three case studies of litigation, whistleblowing, and advocacy, we have shown how competing understandings of democracy play out in the everyday struggles of actors engaged in legitimising and contesting intelligence surveillance, highlighting how these practices were usually excluded from the remit, justifications, and modes of institutionalised oversight. Rather than starting from a taxonomy of theories of democracy, we looked at messy practices where different elements of what counts as ‘democracy’ co-exist, compete, or dominate. Moreover, our analysis of practices of litigation, whistleblowing, and advocacy suggests there isn’t a single model of democracy that informs these struggles – whether liberal, civic republican, deliberative, or agonistic. Rather, different elements are combined to challenge the exclusions, secrecy, and consensus that subtend practices of liberal democracy and its taken-for-granted dominance in academic and policy engagements with intelligence oversight.

The first case of the litigation against the BND has tackled two boundaries of liberal democracy: that of legitimate actors and territorial limits to the rule of law. We have shown how actors from civil society became meaningful oversight protagonists by collectively mobilising to litigate against the exclusion of foreigners from the purview of oversight. Here, although still constrained by dominant positions and views on what is needed to protect intelligence work, oversight can be seen as democratic through pluralising and including more actors in the process, thereby also extending oversight not just within but across borders. The second case on the contested practice of whistleblowing revisited tensions between secrecy and publicity, particularly the acceptance of ‘secrecy’ as a security practice in liberal democracies. By claiming ‘publicity’ and enacting ‘publics’, public whistleblowing practices simultaneously revealed the failures of institutionalised oversight and made the boundaries of secrecy more fluid, subject to mobilisation and struggle over the limits of knowledge. The third case about the UK’s Don’t Spy On Us coalition has illustrated how advocacy oscillates between conflictual and consensual styles of democratic practice, being channelled towards consensus through the delegitimation of critique that is deemed to go beyond what is accepted as ‘legitimate’ or ‘realistic’.

Of course, these localised instances of struggles inscribed in transnational networks are just three of the many that we could have investigated to show how clashing visions of democracy play out in intelligence oversight practices. Other sites could have been addressed – and should be considered in future research –, from open-source journalistic investigations such as those conducted by Bellingcat to the tensions surrounding the work of the United Nations in intelligence policy. What we hope to have conveyed is how meanings and practices of democracy that emerge through oversight practices move along a spectrum, from liberal and deliberative-functionalist understandings of democracy to participatory and radical-agonistic ones.¹¹⁴ In the former, legitimate actors of intelligence oversight are construed as ‘reasonable overseers’ who agree on the relevance and acceptability of intelligence agencies and state surveillance, and where bounded public discussions on intelligence affairs are supposed to help achieve a consensus around intelligence policy based on a stabilised ‘balance’ of values. In the latter, these functionalist views as well as the legal and institutional structures of exclusion giving them prominence are challenged by more radical and excluded actors hoping to convey a more systemic critique over the merits and motives of intelligence policy in democracy. What emerges out of these struggles are strategies of compromise, of tinkering and hybridisation, so that really-existing intelligence oversight remains heterogeneous and contested.

What then are the theoretical and political implications of these heterogeneous practices of oversight and meanings of democracy that our article has shown to be fundamental to the everyday practices of holding intelligence agencies to account? Firstly, our intervention comes as an invitation to reflect on the normative assumptions about democracy that underpin the practices of secret services and oversight actors. We have argued that moving towards plural democratic forms of intelligence oversight would require political imaginaries and policies to accommodate more radical claims and practices and better articulate the different actors engaged in oversight practices. In parallel to such a pluralisation of intelligence oversight practice, this article suggests that intelligence oversight scholarship needs to open up to a wider range of views and disciplinary approaches. As we have shown, the literature on intelligence oversight has tended to either explicitly or implicitly work with liberal and functionalist ideas of democracy that reproduce technocratic institutional arrangements, the rule of law within territorial boundaries, the necessity of secrecy to intelligence agencies, and the priority of consensus through representative institutions. In so doing, these implicit and explicit assumptions about what counts as ‘democratic’ shape and limit the understanding of oversight both in academia and in practice.

Addressing these limitations requires taking studies of intelligence practices and secret services beyond the confines of a field of study and connecting it with broader political questions of democracy, struggles, contestation, and agency, which have been at the core of critical approaches in international relations. These conversations have often taken place in subdisciplinary silos, fragmented and neatly delimited, precluding the construction of bridges or transversal social enquiry between these different imaginaries of democracy. Building on the analysis developed here, struggles around surveillance, intelligence and oversight can be reformulated in the broader terms of struggles around exclusion/inclusion, secrecy/publicity and consensus/contestation, paying attention to how these were formed and evolved in different national and transnational settings. Therefore, the boundaries of intelligence studies as a subfield need to be dismantled so that the theoretical and political concerns of international relations and interdisciplinary research come to reshape the questions, concerns and methods at work in the field.

Secondly, our practice-based approach to intelligence oversight can contribute to discussions of security, surveillance, and contestation in international relations more broadly. Oversight as a practice that limits and mediates security and intelligence practices has received little attention in IR. Oversight both overlaps with and slightly differs from control, scrutiny, or accountability, which constitute an important conceptual and practical apparatus of democratic practice that needs to be further unpacked. In constraining struggles over security, rights, and democracy, oversight is worthy of attention in its own right. When oversight and accountability are increasingly invoked in key sites of international politics, from borders to Artificial Intelligence, our analysis raises a cautionary note and offers a methodological investigation to both specify practices and analyse what a call for oversight means in relation to the multiplicity of democratic practices, meanings, and political subjectivities. Furthermore, as we have seen, oversight also mediates practices of legitimation and delegitimation. Future research will need to attend to practices of intelligence oversight as an important locus in the process of state-making and state legitimation as well.

Thirdly, our research recasts questions about democracy and IR. While democracy has been key to many theoretical approaches in IR, dichotomous conceptions of democracy have often been mobilised to unsettle the eirenic vision of liberal democracy and even dismantle its dominance: liberal/illiberal democracy, liberal/imperial, state/global, liberal/cosmopolitan, representative/participatory, antagonistic/agonistic, representative/deliberative, liberal/civic republican, and the list could go on. Through a practice-based methodology, we have shown that different elements that do not belong to one coherent theory or model of democracy are mobilised in struggles over the limits of democracy. In working through a set of dichotomies that are seen as constitutive of liberal democracy, we have shown how practices of ‘going democratic’ make these limits visible and challenge them. Rather than privileging a particular theory of democracy, we have proposed to take democracy seriously as ‘the paradoxical regime which – as much as possible – admits and accepts the risk of its own internal critique – in any case the critique of its own power-holders.’¹¹⁵ This is neither to revere nor reject certain versions of democracy, but to acknowledge practices that are messy, disputed, and replete with paradoxes.