Introduction

The idea of the kill chain that characterised the “global war on terror”, problematic as it was, was based on the principle of “analyse-reflect-act” (Brose 2020). It allowed and necessitated new practices of targeted killing that were seen as morally plausible (Statman 2004) and more accurate. Yet, the myth of accuracy and objectivity of emerging technology-assisted targeted killing has been debunked by many scholars over time (Fereidooni and Heidt 2024), and the practices of these “kill lists” make the “messy targeting process more opaque and less traceable” (Weber 2016, p. 107). On the other hand, the “kill cloud” worsens this opacity and messiness with its characteristic dependence on vast amounts of data, increased autonomy of machines, and machine learning techniques. The idea of the kill cloud was first introduced by Lisa Ling and Cian Westmoreland to denote a rapidly growing networked infrastructure with a global reach, whose principal aim is domination in every spectrum of warfare, “including space, cyberspace, and the electromagnetic spectrum” (Bazzichelli, 2021; Ling and Westmoreland 2021).

This commentary summarises the discussions at the “Investigating the Kill Cloud: Information Warfare, Autonomous Weapons & AI” conference that took place at the Kunstquartier Bethanien in Berlin, Germany, from the 29th of November to the 1st of December 2024. As a celebration of the Disruption Network Lab’s tenth anniversary, the Disruption Network Lab cooperated with the “Meaningful Human Control: Autonomous Weapons Systems Between Regulation and Reflection” (MEHUCO) research network to bring together whistleblowers, artists, civil society, and scholars from a variety of disciplines including legal studies, science and technology studies (STS), International Relations (IR), political theory, media and communication studies, sociology, computer science, and others. The conference, curated by Tatiana Bazzichelli in collaboration with Jutta Weber, was also aimed at providing space to share the results of the investigations on the impact of artificial intelligence (AI) on new technologies of war, automated weapons, and networked warfare by the Disruption Network Institute’s affiliated research fellows. These research fellows (Fig. 1) were Lisa Ling (whistleblower, former Technical Sergeant, US Air Force Drone Surveillance Programme, USA), Jack Poulson (Executive Director—Tech Inquiry, USA), Naomi Colvin (Programme Director—Blueprint for Future Speech, UK), and Joana Moll (Artist & Researcher, ES). The three-day event also included roundtable discussions aimed at fostering discussion, knowledge sharing, and collaboration among experts from various disciplines.

The MEHUCO research network, led by Paderborn University, consists of five institutions of higher education (University Bonn, Leibniz University Hannover, Hamburg University, Ostfalia University of Applied Sciences, and Paderborn University). The “Investigating the Kill Cloud” conference was organised by the Disruption Network Lab together with the MEHUCO subproject “Swarm Technologies – Control and Autonomy in Complex Weapons Systems” hosted at Paderborn University. Although the discussions and papers presented were numerous, wide-ranging, and specialist, this paper presents a synthesised report of these different talks. It focuses on a few themes that came out of the conference while suggesting areas of further research. All the presentations and performances are available online for the reader to immerse themselves in more detail.

AI, the kill cloud, and whistleblowing

Contemporary war and related military information practices rely on secrecy and classified data. Yet, with the increased automation of warfare, this secrecy is also closely constitutive of impunity, war crimes, and arbitrary killing of innocent populations from afar. The role of declassified data in “peering through the fog of war” has become urgent and indispensable (O’Loughlin et al. 2010). It is within this context that Jesselyn Radack argued that “whistleblowing is critical in the context of drone war because we would have no idea what’s occurring in the most secret and most lethal government programme without the perspective of insiders.”¹ Without whistleblowers, Radack pointed out, it would be impossible to know how governments undercount civilian deaths, regarding thousands of lives as collateral. Thus, the revealing work of people like Chelsea Manning, Julian Assange, Brandon Bryant, Daniel Hale, Christopher Aaron, and others has been instrumental in exposing the ugly truths about remote warfare. Jack Poulson, whose presentation was on the methods of database investigation of Western intelligence and special operations, a major concern of data-centric warfare is the propensity of governments (in his case the US government) to delete evidence from the public while having legal justifications to conceal any data that would expose the military’s injustices. ²

In addition to exposing what is already taking place, Thomas Drake (Fig. 2) posited that whistleblowing helps warn the public about the threats of data-centric warfare. ³ For Drake, whistleblowers pay an incredibly high price, risking everything to expose injustices for the public's good. In data-centric warfare, it is essential to “hold AI to its own mirror” (ibid). It is crucial to expose the dangers of AI as it expands the killing fields, increases the “murder ratio”, and becomes a narrative for the permissibility of inhuman methods of warmongering from a distance. Whistleblowing is crucial in exposing the false beliefs held by the public regarding “sanitised warfare,” which is touted as the advantage of data-centric warfare (Lisa Ling). ⁴

Experimental, business-based warfare

For STS scholars, who were part of the aptly named “Disarming the Kill Cloud” ⁵ panel (Fig. 3), data-centric warfare is based on “sloppy military data science” relying on the principles of correlation (Jutta Weber), with increased algorithmic intensification that does not align with the truth of the lived realities of those on the ground (Lucy Suchman). Suchman critiqued the idea of “ground truth” in military data practices, arguing that together with the datasets from which it is built, ground truth presupposes and requires operations of datafication—rendering worlds of interests as numbers (also see Suchman 2020). For Eric Reichborn-Kjennerud, data-centric warfare represents military logic, not the reality of the physical world. For Elke Schwartz, this military-business logic is underpinned by a logic of trial and error, a high-risk-high-reward business model that encourages the capitalist thinking of “move fast and break things” and “done is better than perfect”. This military philosophy mandates mistakes and valorises error in search of high financial rewards. Marijn Hoijtink’s work on this panel expanded these ideas to include the idea of “the platform” (see Hoijtink 2022; Hoijtink and Planqué-van Hardeveld 2022) which embeds the idea of perfect knowledge, the ability to link everything to everything, to systematise, and naturalise the logic of targeting as exemplified by the US Department of Defense’s (DoD) Joint All Domain Command and Control (JADC2) that links all US military services through data sensors, platforms, and other communication technologies. Hoijtink’s presentation proved the impact and influence of platform owners in contemporary warfare, which transforms the logic and practices of warfare. Naomi Colvin’s presentation showed how the UK’s policy positions on AI safety and risk are intricately linked to ideas from AI labs and Silicon Valley (Colvin 2024).

The work by Joana Moll, which suggests going beyond the techno-economic business model views of digital advertising (Ad Tech), showed how the mechanisms of Ad Tech are grounded in “critical relationships to bodies, emotion and cognition, natural resources, warfare, ideologies, and past and present politics” (Moll 2024, p. 5). Thus, Moll’s performance and presentation showed these intricate relationships while exposing how Ad Tech has become a major enabler of dangerous forms of surveillance that have fed into the mechanics of the kill cloud.

The coloniality of algorithmic violence

The study of algorithmic warfare and autonomous weapons systems has largely focused on legality, state positions and narratives, imaginaries, governance, ethics, and the problems of meaningful human control. Yet, power, historical context, and colonial violence are at the core of these systems' practices, governance, and conceptualisation. In a joint presentation covering their work with the Airspace Tribunal, an ongoing project that is developing a proposed new human right to counter the colonisation of the sky, Shona Illingworth and Anthony Downey’s talk characterised the “pathologies of algorithmic warfare” as an offshoot of and central to neocolonialism. A number of the presentations at the conference exposed the coloniality of algorithmic violence—that is, a set of values, attitudes, epistemologies, and power structures that provide rationalisation and justification of colonial forms of violence and dominance (see Quijano 2007). For Donatella Della Ratta (discussion contribution), AI simply provides a narrative through which machines can be blamed for colonial violence perpetrated by human actors. Rather, algorithmic warfare is mainly targeted at lives that do not matter (see Butler 2009). For Khalil Dewan, algorithmic violence is necropolitical—it ascribes the right to death for those whose lives are deemed losable and disposable (Mbembe 2019). ⁶ Shona Illingworth drew on evidence gathered through the Airspace Tribunal’s development of the proposed new human right to protect the freedom to live without physical or psychological threat from above, that highlights how the sky, especially for the postcolonial subject (see Jabri 2012 on the postcolonial subject), is associated with imprisonment and death (Mohammed 2024), causing long-term psychological trauma and physiological harm (Illingworth et al. 2024).

The realities of algorithmic warfare are inadequately dealt with by the existing legal debates. For Anthony Downey, there is a need to question the historical function/context of AI, its operative logic in warfare, and its systemic functioning. AI is used to provide justifications for algorithmic violence, and it actualises threats it was created to predict (Downey 2024), which has real ramifications on real lives. These ramifications are even more threatening for populations in the Global South. For example, the assault on privacy that people lament in the West is nothing compared to the assault on privacy on people who live in places with the most violent foreign surveillance practices, like Gaza, Afghanistan, and elsewhere (Lisa Ling). ⁷ It is therefore essential to question and critically analyse It is therefore essential to question and critically analyse the epistemological underpinnings, power structures, and colonial injustices created by the use of AI and related technologies in warfare.

Automated surveillance and targeted killing in Gaza

The outbreak of the war in Gaza, characterised by the International Court of Justice’s (ICJ) ruling on genocide, ⁸ United Nations (UN) and several other human rights organisations as a genocide, exposed several aspects related to algorithmic warfare. In the first instance, it brought to the public’s attention the realities of algorithms and AI in warfare. Secondly, it forced the discourse on autonomous weapons systems to move from theoretical imaginations of how AI and machine learning techniques could influence warfare to discussions on how AI has facilitated genocide in practice (Amnesty International 2024). Thirdly, the war transformed the discourse that had largely been focused on autonomous weapons systems to now include other aspects of information warfare like Decision Support Systems that are equally problematic. The “Automated Surveillance & Targeted Killing in Gaza” panel (Fig. 4), ⁹ which consisted of Matt Mahmoudi, Sophia Goodfriend, and Khalil Dewan and was moderated by Matthias Monroy, considered how surveillance technologies and practices of targeted killing operated in Gaza.

For the panellists, the technologies used for control and killing in Gaza are used to justify the ongoing injustices. Sophia Goodfriend argued that the algorithmic violence taking place in Gaza was largely based on three key political decisions: (1) the intentional targeting of civilian homes, (2) raising the threshold of civilian casualties, and (3) encouraging soldiers to over-rely on AI decisions (i.e. automation bias). For Khalil Dewan, the conscious and deliberate political move towards the automation of violence in Gaza has a long history, with Israel being the first country to have a targeting policy in the year 2000.

In practice, this targeted killing depends on a massive infrastructure of surveillance technologies that include the Red Wolf facial recognition technology, Blue Wolf (Goodfriend 2023), MABAT 2000, Checkpoint 56, and others, leading to what Matt Mahmoudi characterised as “automated apartheid” (also see Aizeki et al. 2024 on surveillance, control, and violence). This automated violence is done with impunity, particularly with many actors and infrastructures operating from outside the warzone, including satellites, developers, and so on (Dewan). For Khalil Dewan, the Law of Armed Combat (LAC) faces a key problem in the Gaza case because the perpetrators of violence cannot be attacked back. In addition, Dewan argued, the legal understanding of “hostile intent” of the victims in Gaza is erroneous because the entire population is labelled as hostile. Gaza has shown us that there is a need for new critical approaches that question the injustices of the laws of war that are biased against Global South populations.

Conclusion

As the effects of network-centric warfare become more and more real, and as the numbers of unaccounted—for victims pile up, it has become essential to expand the study of algorithmic warfare and the action to constrain its excesses. As the “Investigating the Kill Cloud” conference showed, the role of whistleblowers in exposing these excesses cannot be overstated. Secondly, as suggested by Anthony Downey at the conference, there is an urgent need to develop fit-for-purpose interdisciplinary methodologies for thinking “from within” the apparatuses that animate algorithmic warfare rather than merely representing or reflecting on them (see also Bazzichelli, 2021 on the idea of disruption from within). Jack Poulson’s work presents a big step in doing this from a whistleblowing perspective, while the majority of scholars represented in this report are largely critical of algorithmic violence. Thirdly, the frames of thinking on algorithmic warfare have been largely limited in scope, leaving most of the existing scholarship unable to question the conditions of injustice and power both at governance and practice levels. Thus, subversive theorisation, including Global South and postcolonial frameworks, can help unsettle the existing systemic injustices necessitated and encouraged by emerging technologies of war. This subversive thinking can open up spaces for gender-sensitive, vulnerability-focused, and environmentally conscious streams of questioning that have the potential to unsettle systemic violence and domination. My own work has so far questioned these injustices and coloniality at the governance level (Bhila 2024a, 2024b). The work of the Stop Killer Robots Campaign and like-minded organisations has exposed the fallacies of algorithmic warfare and targeted killing, but there is still a need for collaborative efforts towards the critique of the “kill cloud” at all levels, and by various communities of practice and thought.

Notes

1.
Jesselyn Radack, Investigating the Kill Cloud Conference, 29 November 2024. Accessible at https://www.disruptionlab.org/investigating-the-kill-cloud
2.
Jack Poulson, Investigating the Kill Cloud Conference, 29 November 2024. Accessible at https://www.disruptionlab.org/investigating-the-kill-cloud
3.
Thomas Drake, Investigating the Kill Cloud Conference, 29 November 2024. Accessible at https://www.disruptionlab.org/investigating-the-kill-cloud
4.
Lisa Ling, Investigating the Kill Cloud Conference, 29 November 2024. Accessible at https://www.disruptionlab.org/investigating-the-kill-cloud
5.
Panel—“Disarming the Kill Cloud” at the Investigating the Kill Cloud Conference, with Lucy Suchman, Erik Reichborn-Kjennerud, Marijn Hoijtink, and Elke Schwarz, moderated by Jutta Weber. https://www.disruptionlab.org/investigating-the-kill-cloud
6.
Khalil Dewan, on Panel “Automated Surveillance & Targeted Killing in Gaza”, Investigating the Kill Cloud Conference, 30 November 2024. Accessible at https://www.disruptionlab.org/investigating-the-kill-cloud
7.
Lisa Ling, Investigating the Kill Cloud Conference, 29 November 2024. Accessible at https://www.disruptionlab.org/investigating-the-kill-cloud
8.
Summary of the Order of 26 January 2024 | INTERNATIONAL COURT OF JUSTICE, 2024).
9.
Panel—“Automated Surveillance & Targeted Killing in Gaza” at the Investigating the Kill Cloud Conference, with Matt Mahmoudi, Sophia Goodfriend, and Khalil Dewan. Moderated by Matthias Monroy.