By: Prof. Branislav Radeljić
This piece is largely inspired by the author’s 2026 study, “Genocide by Algorithm in Gaza: Artificial Intelligence, Countervailing Responsibility, and the Corruption of Public Discourse,” Genocide Studies and Prevention, 20(1), 4–37, https://doi.org/10.5038/1911-9933.20.1.2074.
The growing use of AI in warfare is changing not only how wars are fought, but how violence is classified, justified, and governed. AI systems can sort populations into categories of threat, identify potential targets, assess collateral risk, and process vast amounts of surveillance data. Their apparent objectivity can give algorithmic decisions an authority that may obscure the human assumptions and political choices behind them. Nowhere has this transformation been more evident than in Israel’s conduct in Gaza, where AI-driven targeting systems have been used to automate the identification of targets, including individuals, residential blocs, and civilian infrastructure.
In June 2025, the United Nations Special Rapporteur on the situation of human rights in the Palestinian territories described the “dark side” of Israel’s technology sector and argued that “[r]epression of Palestinians has become progressively automated.” The report referred to systems including Gospel, Lavender, and Where’s Daddy?, which it said were used “to process data and generate lists of targets.” It also identified major technology companies—including IBM, Amazon, Google, Microsoft, Palantir, and Hewlett Packard Enterprise—as providing services that facilitate Israel’s operations in Gaza. With this in mind, and given Israel’s settler-colonial approach and deployment of advanced technologies, the case of Palestine represents a sort of “laboratory,” where the system in place is best described as the “genocidal surveillant assemblage.”
This raises a question that extends beyond any individual technology: what happens to responsibility when decisions about life and death are increasingly mediated by algorithms? The concept of “genocide by algorithm” is useful not because algorithms themselves possess genocidal intentions, but because they can transform how violence is organized and how responsibility is distributed. In other words, the use of algorithms can create distance between an individual decision-maker and the consequences of that decision. This complicates one of the central questions of genocide law: intent.
While the arguments on AI’s role in genocidal violence are still emergent, important connections are being made, including the notions that (a) AI or algorithmic systems may shift blame or facilitate denial, since state actors sometimes frame civilian deaths as accidents, collateral damage, “errors” in targeting, etc.; (b) algorithmic systems could amplify or automate surveillance, the enforcement of blockades, and decisions about movement/displacement or which civilian infrastructures to spare or destroy; and (c) discourse plays a dual role—to legitimize action and shape international response, largely through the rhetoric of “subhumans” or “enemy civilianization,” thus framing civilian casualties as inevitable and contributing to desensitization and normalization. In this context, algorithmic targeting is not merely a matter of technological efficiency; it becomes a tool of systematic erasure.
Michael Harmon’s work on administrative responsibility provides a useful framework for thinking about this problem. Harmon distinguishes between political, professional, and personal responsibility—three forms of responsibility that can overlap and sometimes conflict. Applied to AI-enabled warfare, these categories help reveal that accountability cannot stop with the person who ultimately authorizes a military strike.
Political responsibility concerns decisions made within systems of authority. For governments and militaries, this means responsibility for ensuring that military operations comply with international humanitarian law and other applicable legal obligations. If AI systems are used to facilitate unlawful attacks or systematic destruction of civilian life, the question is not simply whether an algorithm made an error; it is also whether political and military authorities designed, approved, deployed, or failed to regulate a system in ways that made such outcomes possible.
The same question extends beyond Israel. Foreign governments that provide weapons, intelligence, technology, or other forms of assistance must consider what their support enables. If companies are effectively arms of the state, then their actions fall into the realm of political responsibility; put differently, if these companies assist or enable genocide through algorithmic systems, they share in political responsibility by virtue of their alignment with state‐mandated action. Political responsibility therefore reaches across borders and institutions rather than ending at the point where an algorithm is deployed. In other words, if the hierarchical apparatus of foreign administrations permits Israeli actions or algorithmic infrastructure that facilitate genocide, then they have political (and legal) responsibility.
Professional responsibility in the context of Israel’s “genocide by algorithm” applies to military planners, tech corporations, algorithm designers, intelligence analysts, and regulatory agencies. While it is reasonable to believe that in Israel’s military and intelligence apparatus, there are authorities who must act in accordance with professional standards (training, international law, and ethical codes of military conduct), it is equally important to state that when such authorities either design or approve systems which enable automated targeting and mass civilian casualties, they breach the professional responsibility.
The central issue is therefore not simply whether a system can automate targeting, but whether professionals involved in its development and deployment have adequately questioned what the system is being used to accomplish. Therefore, big tech companies, including their professionals in AI governance, machine learning, and systems engineering, carry professional responsibility to ensure their algorithms are not used to facilitate human rights violations. Failure to do so means they are complicit via professional neglect. Finally, US professionals, including regulators, military lawyers, and intelligence ethicists, also hold professional responsibility to ensure that algorithmic systems provided to allies (for example, Israel), or developed by US‐based companies, do not facilitate mass extermination or systematic destruction of populations.
Personal responsibility is also key since algorithms do not make moral choices in the same way people do. They do not possess conscience, political commitments, or an independent understanding of human dignity. The people who design, approve, deploy, maintain, and authorize them do. The crucial question is therefore not whether an algorithm intended an outcome; it is which humans created the conditions under which that outcome became possible, foreseeable, or acceptable—and what capacity they had to intervene.
For example, a military commander in Israel who approves an AI‐targeting algorithm has personal responsibility to step back and reflect if that is consistent with their moral commitments, their understanding of human rights and humanity. If they simply follow orders or if they actively enable algorithmic destruction without reflexive moral consideration, they breach personal responsibility. Similarly, a big tech engineer working at a company whose algorithm is used for targeting civilians in Gaza has a personal responsibility to act with moral commitment, resist or blow the whistle if the system facilitates genocide. If they remain silent or complicit, they fail their personal responsibility. Finally, a US official (or representatives of any other government) authorizing military aid, or a civilian oversight official, also holds personal responsibility as to whether their actions align with moral commitments to protect human life. Their silence or support may constitute personal complicity.
This is where AI creates a particularly difficult challenge. Traditional systems of accountability tend to look for identifiable actors: Who gave the order? Who carried it out? Who violated the law? Algorithmic systems complicate this chain. A decision may emerge from data collected by one institution, a model developed by another, software integrated by a third, intelligence analysed by a fourth, and a military decision ultimately authorized by a fifth. Responsibility can consequently become fragmented across an entire sociotechnical system.
But fragmentation does not necessarily mean absence of responsibility. An algorithm does not become a moral shield simply because many people contributed to its operation. If anything, the complexity of AI systems makes it more important to establish accountability throughout their entire lifecycle—from data collection and model training to system design, deployment, monitoring, and review. This also means that the debate cannot be reduced to whether AI is inherently good or bad. The more important question is who controls these systems, whose interests shape them, what constraints govern their use, and who can be held accountable when they cause harm.
The significance of Gaza therefore extends beyond one conflict or one set of technologies. It raises broader questions about AI-enabled militarism and the future of accountability. The increasing alignment between political and corporate elites has blurred the boundaries between public interest and private gain. As power continues to concentrate, the influence of unelected actors has become impossible to ignore. As noted elsewhere, “[b]illionaires are now effectively making national security decisions on behalf of the United States and using their social media companies to push right-wing agitprop and conspiracy theories.” While AI may be programmed to be aware of the concerns of all, its functionality is inseparable from the capital invested in its development and the political or corporate interests influencing its priorities.
The implications extend beyond warfare. As surveillance systems, recommendation algorithms, data monopolies, and automated decision-making become embedded in everyday political life, citizens increasingly encounter institutions through technological systems they cannot easily see or challenge. Democratic systems, once thought to provide the tools to hold elected leaders accountable, now appear increasingly fragile under the weight of surveillance capitalism, data monopolies, and digital authoritarianism.
The language of “genocide by algorithm” ultimately points toward a broader problem: the danger of allowing technological systems to become substitutes for moral and political judgment. The central issue is therefore not whether machines are responsible for violence. It is whether humans will allow machines—and the complexity surrounding them—to make responsibility so diffuse that nobody can be held accountable.
