The label of the “first AI war” obscures the reality that Operation EPIC FURY is still a conflict in which human judgment remains central to targeting. Artificial intelligence (AI) has not replaced human operators, but it has redefined how human judgment functions. The contemporary battlefield is now shaped by rate of fire, targeting accuracy, AI-enhanced cognition, and the real transformation that machine learning has introduced into modern warfare. 

One of the most persistent misunderstandings about AI-assisted targeting is the claim that humans have somehow been removed from the loop, decisions are made solely on AI recommendations, or strikes are approved in seconds without meaningful review. The human factor has not disappeared. Humans remain indispensable to targeting. What has evolved is not the elimination of human involvement, but the rapid synthesis of intelligence with target acquisition. That distinction matters because many assume there is little or no review before a strike is done. However, no systems make decisions independently. They supplement human decision-making by sorting and ranking information to generate recommended outcomes that must still meet rigid criteria. Analysts verify intelligence, legal teams conduct reviews, commanders make the final decision, and human beings remain responsible for the outcome. 

Critics often point to ambiguity in strategic-level U.S. directives. The Department of War Directive 3000.09 attempted to regulate certain AI-enabled systems, though the technology at the time was far less sophisticated than it is today. Military doctrine undermines the myth of autonomous targeting as well. The Army’s FM 3-60 frames targeting as an iterative command process and states that commanders remain the final approval authority for targeting activities and acceptable levels of risk. Machines may assist with detection, but they do not inherit command responsibility. The result is that humans remain in the loop because targeting is still a command process, not an autonomous one. Military doctrine frames targeting as a cycle of deciding, detecting, delivering, and assessing, but commanders retain authority over acceptable risks. AI can compress and organize the data, but it cannot make strategic or moral judgments. 

AI models remain central to identity-based targeting and advanced decision support. Open-source reporting indicates that sophisticated models, such as Anthropic’s Claude, integrated into systems such as Palantir designed Maven Smart System, have enabled rapid conversion of vast amounts of intelligence, surveillance, and reconnaissance (ISR), signals intelligence, and behavioral data into target packages. 

Human productivity has also increased. Tasks that once required weeks and a large staff can now be completed in minutes with fewer personnel. However, speed and efficiency do not mean AI independence. It does not change what Clausewitz described as the “grammar of war.”  

The most consequential shift in conflict today is the compression of time within the targeting cycle and its integration into intelligence. In the past, high-value or high-payoff targets were often missed because manual processes relied heavily on human operators and overwhelming amounts of ISR data. Past conflicts reflected these limitations. During Operation Desert Storm, Iraqi mobile Scud launchers exploited delays by firing and relocating before U.S. forces could strike them. Kinetic precision still frequently exceeds intelligence fidelity. A munition could hit its coordinates perfectly while the underlying intelligence remained flawed. The use of intelligence to target enemy combatants predates modern technology. AI did not invent decapitation strategies; it made them more data-driven and less dependent on purely human intelligence sources. The Information Age once overwhelmed operators with data. AI now provides a way to navigate that environment. This is why cyber intelligence and persistent access are essential to modern targeting. Pattern-of-life targeting relies on multiple streams of surveillance and behavioral data. AI’s greatest strength is its ability to combine these streams on a scale that would overwhelm most military units. Yet the central question remains unresolved by algorithms alone: should the target be neutralized? That decision is legal, moral, political, and strategic. 

A clear example of AI-integrated intelligence limits came on the first day of the 2026 Iran War, when a U.S. missile struck an elementary school in Minab, Hormozgan province, killing civilians, including children, in one of the war’s deadliest civilian incidents. The incident underscored a basic truth: AI-enabled targeting is only as dependable as its data. Here, the system likely relied on outdated intelligence that missed the school’s proximity to an Islamic Revolutionary Guard Corps compound.  

AI was not the likely source of failure. More likely, flawed intelligence and the fog of war were to blame. Human operators still validated the strike with satellite imagery and intelligence reviews, even though the target was effectively co-located with the school.  

The episode showed both the limits of AI models and the need for human review. Systems like Maven Smart System’s Target Workbench sort, correlate, and reveal intelligence, but humans still approve of final actions. AI can aid target validation, but legal review and command authorization remain essential. 

CONCLUSION 

The effectiveness of any algorithm depends entirely on the intelligence architecture and data supporting it. AI does not create certainty; it produces probability. If the underlying data is manipulated, incomplete, stale, or inaccurate, the output will reflect those flaws. The greater danger is not the removal of the human in the loop, but the compression of human judgment into groupthink. AI-generated recommendations can create an aura of probabilistic certainty that encourages agreement instead of scrutiny. Human operators may still make the final call, but the risk is that they increasingly validate model logic rather than independently challenge it. 

Humans remain in the loop today, but intelligence is now sorted at machine speed while generative systems provide recommendations to reviewers within the targeting cycle. Doctrine should evolve to ensure that human judgment takes precedence over AI-generated recommendations. The defining feature of this so-called first AI war is therefore not the replacement of human agency, but the intensification of human responsibility to judge, restrain, and decide at machine speed. 

Lieutenant Colonel Matthew J. Fecteau is an information operations officer working with artificial intelligence, and a PhD researcher at King’s College London. The views expressed in this report are those of the author and do not necessarily reflect the official policy or position of the Department of the Army, the Department of War, or the US Government. 

Why the “First AI War” is Still a Human Struggle was originally published on Global Security Review.

As the world gears up for the 2026 Nuclear Non-Proliferation Treaty (NPT) Review Conference, a new and multifaceted factor is complicating the global strategic calculus: Artificial Intelligence (AI). The “nuclear-AI nexus” has evolved from a niche technical interest to a prominent feature in global security discussions, with implications for every aspect of the NPT’s three pillars: non-proliferation, disarmament, and peaceful uses of nuclear energy. However, as experts recently cautioned at the “Atoms for Algorithms” webinar, we need to cut through the speculative “AI hype” to ensure this technology remains a means for peace, not an avenue for unintentional escalation.

To regulate the nuclear-AI connection, we first need to understand the technology. As a United Nations Institute for Disarmament Research researcher, Dr. Yasmina Fina suggests, AI is not a monolithic entity, but a “construct” and a “system of systems,” made up of code, software, data, hardware, and sensors. It is a system used to fulfill certain functions, not a threat capable of usurping human decision-making. The risk is the “speed and scale” of AI, which can have myriad implications for performance and strategy. Moreover, Fina warns that comparing nuclear governance to AI governance is unhelpful because the technologies are not the same; nuclear materials are limited and tangible, whereas AI is ubiquitous and digital.

In the context of non-proliferation, AI is touted as a game-changing verification tool. The International Atomic Energy Agency (IAEA) could potentially use “AI agents,” semi-autonomous machines capable of processing large data streams and satellite images to verify the accuracy of declarations by states at a pace humans cannot match. However, Dr. Ian Stewart, Executive Director of the CNS Washington, states that AI will not help states develop nuclear weapons they could not otherwise build, for two physical reasons: AI cannot “magic up” fissile material, and there is no evidence that large language models can transfer the “tacit knowledge” necessary for weaponization.

When it comes to AI and nuclear weapons, political concerns are high. States have been reluctant to allow the IAEA to use open-source data or “black box” algorithms. Should an AI detect an event, the absence of “explainability” or how the machine arrived at its decision could lead to a crisis of political legitimacy for the safeguards system.

The most fraught part of the nexus is disarmament. We are now seeing a “race to adopt” AI in military strategies due to the perceived speed advantage it offers. AI can speed up threat identification and data integration, potentially freeing up more time for decision-making (or, on the other hand, reducing decision-making cycles to the point that humans are simply rubber-stamping decisions).

Aliche Sultini, senior research lead at the Rhode Island School of Design, explains that AI systems create new levels of uncertainty. If a state cannot grasp how an adversary’s AI operates in its decision-making, it might fall into worst-case scenarios, reinforcing alert postures that prevent disarmament. To support disarmament, AI must be used to enhance technical verification and confidence, not to shorten the path to war. Possibly the most disruptive aspect of this interaction is the NPT’s third pillar: peaceful nuclear energy applications. According to Mr. Shota Kamishima of the IAEA, an “affinity” is emerging between nuclear power and AI. We are now moving into a world where energy-hungry AI data centers need the clean, scalable, and reliable power offered by nuclear power, and where nuclear generation and maintenance are improved through AI.

This alliance is especially important for the rollout of Small Modular Reactors (SMRs), for which AI-optimized construction schedules and supply chains are critical. By enhancing predictability and avoiding cost overruns, a major issue for nuclear construction, AI could make nuclear projects more “bankable” and thus more attractive for the global shift towards clean energy. Despite technological progress, the experts agree: human responsibility is essential. Whether it is a safeguards inspector at the International Atomic Energy Agency (IAEA) or a human commander in a nuclear-armed nation, humans are the “last line of defence”.

“Black box” systems are incompatible with a strong safety culture. Governance policies must ensure that AI is implemented with transparency, traceability, and “explainability”. We must also be alert to the potential for AI to be employed by “agents” to monitor sites, which could result in disinformation and the distortion of threat perception through “anomaly detections.” As the 2026 Review Conference draws near, policymakers’ mission should be to “denoise.” Nuclear policy decisions must not be based on science fiction or fear of competition. Rather, we should prioritize “lower stakes” opportunities where AI can help us now, such as employing AI agents to navigate the overwhelming output of the NPT process — making it searchable and highlighting inconsistencies in delegation positions.

The relationship between nuclear and AI is not a bug to be fixed with more software, but a circumstance to be managed by the international community. By prioritizing evidence-based policy and law, human-in-the-loop systems, and the common ground of the peaceful “Atoms for Algorithms” alliance, we can ensure that the digital revolution supports, rather than undermines, the global nuclear order. In the end, the fate of the NPT will not be determined by algorithms, but by human intelligence.

Muhammad Shahzad Akram is a Research Officer at the Centre for International Strategic Studies, AJK. He holds an MPhil in International Relations from Quaid-i-Azam University, Islamabad. He is an alumnus of the Near East South Asia (NESA) Centre for Strategic Studies, National Defence University (NDU), and Washington, DC. His expertise includes cyber warfare and strategy, arms control, and disarmament. Views expressed in this article are the author’s own. 

Navigating the AI and Nuclear Nexus was originally published on Global Security Review.

Whether the MSS is indicative of an unfolding RMA remains a subject of debate. At a minimum, it represents a significant leap in how modern militaries sense, decide, and act in combat. From a scholarly perspective, RMAs are not defined by single technological breakthroughs but by clusters of innovations that fundamentally transform the conduct of warfare.

They typically involve shifts in doctrine, tactics, organization, culture, and technology. Unlike broader military revolutions, which reshape societies and political systems, RMAs are confined to the military sphere—and they often unfold quietly, only recognized in hindsight.

Several RMAs were identified in the past, providing a framework to anticipate future ones. In The Dynamics of Military Revolutions: 1300–2050, MacGregor Knox and Williamson Murray outline five significant military revolutions in the West since 1618. Each one, they argue, set off a chain of revolutionary changes in military affairs.

These include the emergence of the modern state with its standing armies, the political and social upheavals brought on by the French Revolution, the industrialization of warfare in the 19th century, the era of total war in the 20th century, and the transformative impact of nuclear weapons. If a new RMA is underway, we may not fully recognize it until it has already matured.

The concept of RMA has drawn justified criticism for being abstract, amorphous, and debated to the point of analytical paralysis. After the Gulf War, the DoD’s fixation on identifying the “next RMA” often overshadowed the operational impact of emerging capabilities. Scholars frequently focus on definitional purity rather than assessing real battlefield transformation.

Whether the MSS fits a textbook definition, adopted by the DoW or derived from historical theory, is less important than its functional impact. If an RMA is indeed emerging or approaching, there should be tangible real-world consequences. Otherwise, theory becomes disconnected from practice. In this light, the MSS may serve as a bridge between the long-unfolding information RMA and a new, AI-driven transformation.

The MSS could be indicative of another significant shift in command and control (C2). While the US Army’s command post computing environment (CPCE) already integrates legacy systems into a modular, cloud-capable architecture for multi-domain operations, the MSS pushes these capabilities toward revolutionary real-time situational awareness.

While initially developed to automate drone feed analysis, the MSS has evolved into an AI-powered battlefield intelligence engine. It fuses intelligence, surveillance, and reconnaissance (ISR) data, enables real-time targeting, and supports distributed decision-making. As with the telegraph in the 19th century, the MSS may redefine the military’s relationship with information and time.

Historically, C2 was slow and fragmented. Commanders relied on flags, runners, and direct observation, limited by geography and transmission delay. The Industrial Revolution began to change this. Introduced in 1793, Claude Chappe invented the optical telegraph which allowed faster coordination across long distances. It was Samuel Morse’s electrical telegraph, patented in 1837, that truly revolutionized communication.

AI is reshaping combat just as electricity once did. Electricity transformed communication by creating the foundation for critical innovation, like the internet. The harnessing of electricity for industrial use itself was not an RMA, but it was the essential prerequisite for one. Without it, the revolution in communication that began with the telegraph would not have been possible. AI may not constitute a full RMA on its own, but it is the enabling foundation for one.

During the Crimean War and the American Civil War, the telegraph enabled real-time command for the first time. In the US, President Lincoln relied on the War Department telegraph office to direct Union forces and enforce strategic decisions. Strategic-level C2 became possible, and expectations for real-time situational awareness took hold. The rise of the steam-powered printing press and the expansion of railways accelerated this transformation, making war reporting nearly instantaneous—a precursor to modern information warfare.

Similarly, Project Maven, initiated in 2017, began as a machine learning initiative to automate drone video analysis. Since then, the MSS has grown to integrate cloud computing, ISR fusion, and targeting. The MSS delivers intelligence to the tactical edge at machine speed on enterprise cloud infrastructure. It processes unfathomable amounts of data in milliseconds— augmenting analysts and automating portions of the workflow.

Just like the electric telegraph centralized control and supported linear commander decisions, the MSS introduces machine learning, machine inference, and adaptive analytics to take command and control. The MSS provides a picture of the theater that is not merely quantitative, but qualitative.

A true RMA requires more than new technology. It demands operational adaptation, organizational restructuring, and doctrinal evolution. The MSS checks many of these boxes. Technologically, the MSS merges AI, edge computing, and cloud infrastructure in a holistic fashion. Operationally, it uses human-machine teaming to accelerate kill chains. Organizationally, it catalyzed the creation of institutions such as the Joint AI Center (JAIC) and the Chief Digital and Artificial Intelligence Office. Doctrinally, it promotes shifts toward algorithmic and mosaic warfare, which are adaptive, data-driven models of conflict.

The MSS could signal a broader shift in military operations, much like the telegraph reshaped communication in the 19th century. By combining intelligence, surveillance, and reconnaissance (ISR) with artificial intelligence at operational speed, the MSS is changing how armed forces interpret the battlespace, make decisions, and coordinate action—all while improving the shared situational picture. Yet without a corresponding cultural shift, even the best tools can fail to yield a true RMA. Whether the Department of War can fully adapt its doctrine and institutions to leverage the MSS remains to be seen.

Lieutenant Colonel Matthew J. Fecteau is an information operations officer working with artificial intelligence. The views expressed in this report are those of the author and do not necessarily reflect the official policy or position of the Department of the Army, the Department of War, or the US Government. 

Signals of a New Revolution: Maven Smart System and the AI-RMA Horizon was originally published on Global Security Review.

Integrating AI/ML into low-observable technology presents a promising avenue for enhancing stealth capabilities, but it also comes with its own set of challenges. ML algorithms rely on large volumes of high-quality data for training and validation. Acquiring such data for low-observable technology is challenging due to the classified nature of military operations and the limited availability of real-world stealth measurements.

ML algorithms analyze vast amounts of radar data to identify patterns and anomalies that were previously undetectable. This includes the ability to track stealth aircraft and missiles with greater accuracy and speed. These advancements have significant implications for deterrence strategies as traditional stealth technology may diminish in its effectiveness as AI/ML-powered radar becomes more sophisticated, potentially undermining the deterrent value of stealth aircraft and missiles.

Stealth technology remains a cornerstone of deterrence, allowing military assets to operate relatively undetected. Radar, on the other hand, is the primary tool for detecting and tracking these assets. However, AI/ML are propelling both technologies into new frontiers. AI algorithms can now design and optimize stealth configurations that were previously impossible. This includes the development of adaptive camouflage that dynamically responds to changing environments, making detection even more challenging.

Furthermore, stealth technology encompasses a multitude of intricately designed principles and trade-offs, including radar cross-section (RCS) reduction, infrared signature management, and reduction of acoustic variables. Developing ML algorithms capable of comprehensively modeling and optimizing these complex interactions poses a significant challenge. Moreover, translating theoretical stealth concepts into practical design solutions that can be effectively learned by ML models requires specialized domain knowledge and expertise.

As ML-based stealth design techniques become more prevalent, adversaries may employ adversarial ML strategies to exploit vulnerabilities and circumvent the defenses afforded to stealth aircraft. Adversarial attacks involve deliberately perturbing input data to deceive ML models and undermine their performance. Mitigating these threats requires the development of robust countermeasures and adversarial training techniques to enhance the resilience of ML-based stealth systems.

Additional complexities are inherent in the fact that ML algorithms often operate as “black boxes,” making it challenging to interpret their decision-making processes and understand the underlying rationale behind their predictions. In the context of stealth technology, where design decisions have significant operational implications, the lack of interpretability and explainability poses a barrier to trust and acceptance. Ensuring transparency and interpretability in ML-based stealth design methodologies is essential for fostering confidence among stakeholders and facilitating informed decision-making.

Implementing ML algorithms for stealth optimization involves computationally intensive tasks, including data preprocessing, model training, and simulation-based optimization. As low-observable technology evolves to encompass increasingly sophisticated designs and multi-domain considerations, the computational demands of ML-based approaches may escalate exponentially. Balancing computational efficiency with modeling accuracy and scalability is essential for practical deployment in real-world military applications.

Integrating AI and ML into military systems raises complex regulatory and ethical considerations, particularly regarding autonomy, accountability, and compliance with international laws and conventions. Ensuring that ML-based stealth technologies adhere to ethical principles, respect human rights, and comply with legal frameworks governing armed conflict is paramount. Moreover, establishing transparent governance mechanisms and robust oversight frameworks is essential to addressing concerns related to the responsible use of AI in military applications.

Addressing these challenges requires a concerted interdisciplinary effort, bringing together expertise from diverse fields such as aerospace engineering, computer science, data science, and ethics. By overcoming these obstacles, AI/ML has the potential to revolutionize low-observable technology, enhancing the stealth capabilities of military aircraft and ensuring their effectiveness in an increasingly contested operational environment. On the other hand, AI/ML has the potential to significantly impact radar technology, posing challenges to conventional low-observable and stealth aircraft designs in the future.

AI/ML algorithms can enhance radar signal processing capabilities by improving target detection, tracking, and classification in cluttered environments. Analyzing complex radar returns and discerning subtle patterns indicative of stealth aircraft, these algorithms can mitigate the challenges posed by low-observable technology, making it more difficult for stealth aircraft to evade detection.

ML algorithms can optimize radar waveforms in real time based on environmental conditions, target characteristics, and mission objectives. Dynamically adjusting waveform parameters such as frequency, amplitude, and modulation, radar systems can exploit vulnerabilities in stealth designs—increasing the probability of detection. This adaptive approach enhances radar performance against evolving threats, including stealth aircraft with sophisticated countermeasures.

Cognitive radar systems leverage AI/ML techniques to autonomously adapt their operation and behavior in response to changing operational environments. These systems learn from past experiences, anticipate future scenarios, and optimize radar performance adaptively. Continuously evolving their tactics and strategies, cognitive radar systems can outmaneuver stealth aircraft and exploit weaknesses in their low-observable characteristics.

AI/ML facilitates the coordination and synchronization of multi-static and distributed radar networks, comprising diverse sensors deployed across different platforms and locations. By fusing information from multiple radar sources and exploiting the principles of spatial diversity, these networks can enhance target detection and localization capabilities. This collaborative approach enables radar systems to overcome the limitations of individual sensors and effectively detect stealth aircraft operating in contested environments.

ML techniques can be employed to develop countermeasures against stealth technology by identifying vulnerabilities and crafting effective detection strategies. By generating adversarial examples and training radar systems to recognize subtle cues indicative of stealth aircraft, researchers can develop robust detection algorithms capable of outperforming traditional radar techniques. ML provides a proactive defense mechanism against stealth threats, potentially rendering conventional low-observable technology obsolete.

AI and ML enable the construction of data-driven models and simulations that accurately capture the electromagnetic signatures and propagation phenomena associated with stealth aircraft. By leveraging large datasets comprising radar measurements, electromagnetic simulations, and physical modeling, researchers can develop comprehensive models of stealth characteristics and devise innovative counter-detection strategies. These data-driven approaches provide valuable insights into the vulnerabilities of stealth technology and inform the design of more effective radar systems.

In the quest for technological superiority in modern warfare, the integration of AI and ML into radar technology holds significant promise with the potential to challenge conventional low-observable and stealth aircraft designs by enhancing radar-detection capabilities. AI and ML algorithms improve radar signal processing, optimize radar waveforms in real time, and enable radar systems to autonomously adapt their operation. By leveraging multi-static and distributed radar networks and employing adversarial ML techniques, researchers can develop robust detection algorithms capable of outperforming traditional radar systems. Moreover, data-driven modeling and simulation provide insights into the vulnerabilities of stealth technology, informing the design of more effective radar systems.

The rapid advancement of AI/ML is revolutionizing both stealth and radar technologies, with profound implications for deterrence strategies. Traditionally, deterrence has relied on the balance of power and the credible threat of retaliation. However, the integration of AI/ML into these technologies is fundamentally altering the dynamics of detection, evasion, and response, thereby challenging the established tenets of deterrence. Of further concern is the consideration that non-stealth assets become increasingly vulnerable to detection and targeting as ML-powered radar systems become more prevalent. This could lead to a greater reliance on stealth technology, further accelerating the arms race.

This rapid development of AI/ML-powered technologies could destabilize the existing balance of power, leading to heightened tensions and miscalculations. The changing technological landscape may necessitate the development of new deterrence strategies that incorporate AI and ML. This could include a greater emphasis on cyber warfare and the development of counter-AI and counter-ML capabilities.

The integration of AI/ML into stealth and radar technologies will be a game-changer for deterrence. To maintain stability and prevent conflict, policymakers and military strategists must adapt to this new reality of a continuous arms race, wherein both offensive and defensive capabilities are constantly evolving in pursuit of technological superiority. Continued investment in AI/ML research is essential to stay ahead of the curve and maintain a credible deterrent posture. International cooperation on the development and use of AI/ML technologies in military applications is crucial to limit the scope of a potential arms race that regularly shifts the balance of power and destabilizes global security.

Joshua Thibert is a Contributing Senior Analyst at the National Institute for Deterrence Studies (NIDS) and doctoral candidate at Missouri State University. His extensive academic and practitioner experience spans strategic intelligence, multiple domains within defense and strategic studies, and critical infrastructure protection. The views expressed in this article are the author’s own

The Double-edged Sword of Artificial Intelligence was originally published on Global Security Review.