I
n the 21st century, wars have begun to take shape not only through purely physical conflicts but also through systems integrated with cognitive, cyber and artificial intelligence (AI) technologies. We can say that modern conflict and battlefields encompass operations targeting information, perception and the human mind, alongside the use of military force. This new form of warfare is referred to in the literature as cognitive warfare. The fundamental objectives of cognitive warfare can be described as influencing the adversary’s decision-making processes, steering the psychology of the target population(s), and weakening political will. NATO’s 2025 Cognitive Warfare report highlights that contemporary conflicts and wars are increasingly behavior-centric, and that the decisive arena is no longer geographical; rather, it is based on the strategic reality of how individuals and groups perceive, interpret, decide and act.
Within this framework, it can be argued that today’s mass media, the increasing use of AI-enabled technologies, and the undeniable influence of social media are effectively forming the new operational front of war. Cognitive warfare extends to social foundations such as trust, shared epistemic standards and institutional legitimacy. From a force development perspective, AI platforms are utilized alongside elements of cognitive warfare, both for threat/attack purposes and as a countermeasure. In other words, techniques that enable enemy influence can also be utilized through reverse engineering to support defensive cognitive security (such as detecting anomalies in influence networks and pattern recognition of coordinated deceptive behavior).
At this stage, whilst neuroscience and neurotechnologies are emerging as tools for influencing the mind by affecting the brain, it can be said that AI platforms have become the primary means of influencing the mind by manipulating the information ecosystem. Consequently, the operational concern is not merely that AI produces persuasive content. It also lies in AI’s ability to: Optimize narratives in real time across all channels, automate social media interaction (using bot or hybrid actor swarms), generate artificial credibility (deepfake videos, fake expert profiles, fabricated evidence), exploit cognitive biases and psychological vulnerabilities (the salience effect, fear conditioning, polarization dynamics).
In this context, cognitive warfare can be regarded as the process of preparing the groundwork prior to a physical attack. In NATO documents, the cognitive domain is discussed as the sixth operational domain following land, sea, air, cyber and space. The tensions between the US and Iran, and particularly the military conflicts of recent years, have clearly demonstrated just how significant cognitive warfare has become in modern war doctrines. As of 28 February, the attacks launched by the US against Iran have involved not only conventional military force but also intensive use of cyberattacks, cognitive warfare integrated with information manipulation, and AI-supported, target-focused strikes. The key question here is: how did the US utilize AI-integrated systems, which combine data-driven targeting with AI-integrated cognitive warfare, as a strategic tool throughout Operation Epic Fury, the offensive launched against Iran?
Military AI encompasses a wide range of systems and tasks. On this basis, we can state that it is fundamentally divided into two main categories: autonomous weapon systems and decision support systems. Autonomous weapon systems have the capability to select and/or engage targets independently. Decision support systems, on the other hand, are regarded as a key focus in the technology-driven structure of many modern armies today. These systems can be considered software applications that provide intelligence and planning information to personnel.
Admiral Brad Cooper, Commander of CENTCOM, stated during the US-Iran conflict that the US military was using AI systems and that AI assisted in data processing, but that final decisions were made by humans. It is known that the US military utilizes AI systems based on large language models (LLMs) for logistics, administrative support, intelligence gathering and analysis, as well as decision support systems on the battlefield. Claude, developed by Anthropic, is cited as an example of a decision support system rather than a weapon.
Claude is integrated into the intelligent system known as Maven, which is widely used by military, intelligence and law enforcement organizations. Maven uses AI algorithms to identify potential targets from satellite and other intelligence data. In this context, Maven accelerates the military’s offensive capabilities by suggesting and prioritizing targets. Whilst these AI systems have a relatively long history, the US military’s ability to utilize them did not emerge overnight. We are also aware that many Iraqi and Afghan civilians lost their lives as a result of analytical errors and cultural biases within the US military.
Most recently, there is strong evidence that the use of AI systems led to a Tomahawk cruise missile striking a girls’ school adjacent to an Iranian base, contributing to the deaths of approximately 175 people, the majority of whom were students. This attack could be interpreted as a failure of US intelligence or as an indiscriminate strike. Indeed, the effective use of automated systems depends on comprehensive infrastructure and personnel capable of operating them. In this context, it is also worth noting that the US’s ability to utilise AI in warfare today is the result of decades of investment and experience.
In the context of the US–Iran conflict, one of the prominent features of modern warfare is the increasing role of AI-supported decision-making systems in attacks. Applications emerging in active war and conflict zones, particularly in the Middle East, demonstrate the potential consequences of these technologies in terms of both operational effectiveness and ethical responsibility.
Furthermore, the AI systems utilised by the US in its military and intelligence operations against Iran constitute a significant part of this transformation. Systems such as Maven, developed within the US Department of Defence, play a significant role in intelligence analysis concerning Iran’s military infrastructure, nuclear facilities and regional militia networks. In this context, discussions regarding AI-supported target identification systems used in Gaza are important for understanding the cognitive and technological warfare practices that may emerge in future high-intensity geopolitical rivalries such as that between the US and Iran.
The AI system named “Lavender”, used by Israel in Gaza, was developed to identify individuals believed to be linked to Hamas by analysing vast datasets. The system generates a score regarding an individual’s likelihood of being a Hamas member by analysing numerous data points, including telephone data, communication networks, address changes, social connections and intelligence records. However, although it is known that these algorithmic assessments are not definitive and contain a certain margin of error, there is evidence that the lists generated by this system were used in large-scale target identification processes during the first weeks of the war.
Whilst the Lavender system flagged tens of thousands of people as potential targets in the early stages of the war, some military sources have stated that this figure reached approximately 37,000. It emerged that a significant proportion of these individuals were not high-ranking military figures, but rather people of lower rank or with indirect connections.
The ‘Where’s Daddy?’ system, used in conjunction with the Lavender system, is designed to track the locations of designated targets. This system specifically aims to identify the moment the target returns home in order to determine the most opportune time for an attack. The practical consequence of this approach has been that many attacks have taken place in residential properties where the target was present alongside family members. In research, some intelligence officers have noted that military operations are often carried out not whilst individuals are engaged in military activities, but whilst they are at home, and that this leads to an increase in civilian casualties.
Furthermore, it is stated that the use of these systems has significantly lowered the accepted ‘collateral damage’ thresholds in military operations, particularly during the early stages of a conflict. In general, it has been noted that even when targeting a low-ranking militant, a certain number of civilian casualties is deemed acceptable for the operation. This situation forms the basis for serious debates regarding AI-assisted target identification systems, not only in terms of military efficiency but also in relation to international humanitarian law and the ethics of war.
The ethical debates surrounding AI-assisted target identification systems demonstrate that cognitive warfare affects not only societies’ perceptions but also the mental processes of decision-makers. In this context, the debates surrounding systems such as Lavender and Where’s Daddy, used in Gaza, serve as a significant example for understanding the human–machine relationship and how decision-making processes are transforming in modern warfare. In particular, the fact that these systems classify potential threats by analyzing extensive datasets during target identification processes highlights that the role of human decision-makers on the battlefield is becoming increasingly complex.
From a cognitive warfare perspective, this situation can be viewed not merely as a technological advancement, but also as a factor influencing military personnel’s perceptions, risk assessments and ethical boundaries. Consequently, we can state that the large-scale target identification processes generated by the Lavender system’s lists make a significant contribution to military planning.
However, despite the awareness that algorithmic assessments may contain a high margin of error, data indicates that these outputs are often approved with limited human oversight due to the need for rapid decision-making in operational processes. This situation suggests that recommendations generated by AI could exert a powerful cognitive influence on human decision-makers, leading to the algorithmic output effectively becoming a decision-making mechanism in practice. The fact that the ‘Where’s Daddy’ system focuses specifically on identifying the moments when target individuals return to their homes has resulted in operations frequently taking place in residential areas.
This situation has given rise to significant debates not only from a military strategic perspective but also regarding the ethical dimensions of warfare. Indeed, the fact that target identification processes rely heavily on algorithmic analysis leads military personnel to place greater trust in the recommendations of technological systems when making decisions, which in turn can narrow the scope of human assessment.
(Originally published in Turkish by Kriter)
Stay informed by subscribing to
our weekly newsletter.
