Fighting at Machine Speed: AI and U.S. Army Counterfire Under the Law of War – Part I

A strategic shift in national security priorities from counterinsurgency operations to great-power competition and conflict has exposed significant technological and capability gaps in the U.S. military. Among these are lack of guaranteed air superiority and peer adversaries’ field large artillery forces. Both of these likelihoods require significant improvement in U.S. capabilities to detect and target enemy indirect fire systems, or as they are known in military circles, counterfire operations.

Artificial intelligence (AI) presents significant opportunities to decrease response time and increase lethality in counterfire operations. Three AI systems (MSS, TITAN, and ASTARTE) are designed to do just that, raising legal and operational questions that require evaluation under the core principles of the law of war. This post outlines the doctrinal foundations of counterfire and examines how emerging AI technologies accelerate targeting and engagement. Part II of this post then analyzes the use of these capabilities under the law of war.

U.S. Army Counterfire

Field artillery is a critical component of large-scale combat operations (LSCO), consisting of cannons, rocket launchers, and radars. Commanders use artillery to shape the battlefield and, critically in contested airspace, to target enemy air defense and artillery capabilities. The enduring imperatives that “God fights on the side with the best artillery” and the adage that removing “the enemy’s artillery is older than a cannon” drive continuous advances in counterfire technologies and tactics.

Ensuring one’s artillery out-survives the enemy’s is therefore crucial; effective counterfire contributes to that very outcome. The U.S. Army defines counterfire as “an operations function that aims to accurately engage enemy indirect fire systems once acquired.” Essentially, counterfire employs friendly indirect fire and fire support assets to destroy enemy indirect fire systems.

While destroying an adversary’s artillery is an objective as old as the technology itself, modern counterfire doctrine emerged only after artillery evolved from a defensive, direct-fire weapon into an offensive system capable of indirect fire (fire that does not trace a line of sight from a weapon to its target). That shift accelerated during the First World War, when indirect fire became the Army’s primary means of offensive fire and new techniques like sound and flash ranging, aerial observation, registration, and centralized artillery intelligence were developed to locate enemy guns and reduce response times. Across subsequent conflicts, from the Second World War through Korea, the Army refined these methods while expanding radar, forward observers, and longer-range fires, reinforcing the principle that defeating enemy artillery depends on speed, accuracy, and effective target acquisition.

During the Cold War and Vietnam era, computers and battlefield sensors further compressed counterfire timelines by reducing human decision-making delays. Although digital fire control systems and rocket artillery proved decisive during the 1990-91 Desert Storm campaign, counterfire receded during 21st century counterinsurgency operations conducted under U.S. air superiority.

A Counterfire Revival

With the return of peer competition and conflict, the Army has refocused on counterfire, acknowledging its importance to success. This focus frames how the U.S. Army organizes and executes counterfire in LSCO. At the most general level, the Army conducts counterfire operations against many levels of enemy command but concentrates effort where enemy fire support systems are most active. At the operational level, counterfire operations fall into two categories: proactive and reactive. Proactive counterfire targets an adversary’s indirect fire system before they can engage U.S. forces, while reactive counterfire responds after those systems are acquired, usually having fired and revealed their position.

Although AI enhances both approaches, because reactive counterfire hinges on speed and precision, it provides the clearest entry point for AI integration. This distinction between proactive and reactive counterfire both structures and constrains AI integration during the targeting process.

The Targeting Process

Roughly speaking, targeting refers to the act of deciding what to strike and when. The Army conducts two types of targeting: deliberative and dynamic. The former uses a cycle of distinct, though often overlapping stages including: decide; detect; deliver; and assess. This cycle can be time-consuming and is primarily used to plan proactive counterfire operations. Reactive counterfire, by contrast, most often occurs in a dynamic targeting construct, where detection, target development, and engagement decisions must be made in minutes.

There is no fixed time standard for counterfire operations. Instead, three variables drive response times including: the enemy’s time to displace after acquisition; the weapon’s time-of-flight to the target; and the time required to send target data (also known as the sensor-to-shooter loop) and orders to the firing unit (often referred to as the “execution loop”). Of these variables, only the last two are meaningfully reducible. The other variables are largely fixed by physics and adversary capability.

That constraint defines the operational problem the Army seeks to solve through AI-enabled counterfire. Consider the following scenario: a target detected 50 kilometers from a counterfire rocket system requires roughly one minute of flight time and, at best, another one to two minutes for detection validation, deconfliction, approval, and fire transmission before the enemy displaces. Under those conditions, even modest delays in processing or coordination can render counterfire ineffective. Examining the sensor-to-shooter architecture reveals where time is lost, and where time can be saved.

The Sensor-to-Shooter Network

Modern counterfire depends on two complementary information collection, or as they are often referred to, “sensor” regimes that collect target data and feed the targeting enterprise: 1) weapons-locating radars designed to detect and locate incoming projectiles for reactive counterfire, and 2) broader intelligence, surveillance, and reconnaissance (ISR) detection systems that feed proactive counterfire. Each regime produces different kinds of target data and timelines, and therefore different doctrinal and legal expectations about how rapidly commanders may act.

At the tactical level, the sensor-to-shooter loop traces a consistent pattern: a sensor detects a target, a fires cell validates it, a fire direction center refines the firing solution, and the gun line shoots. The efficiency of this architecture depends on how quickly and accurately information moves between these nodes. Every handoff introduces delay and potential for error. When communications are stable and firing authority is effectively delegated between units, a fire mission can be generated and transmitted within minutes. However, degraded networks or multi-level approval requirements can rapidly extend that timeline. Also, because sensors deliver data in different formats and with varying confidence levels, deconfliction and validation (while essential for safety) remain the principal constraints on the speed of counterfire.

A further constraint in the sensor-to-shooter environment is airspace control. Army doctrine conceptualizes battlefield airspace as a shared domain, requiring that fire missions be coordinated and deconflicted with multiple airspace users including aviation assets. Fires cells do not clear fire missions themselves or in isolation. They must consult and deconflict against airspace control measures put in place by and applicable to others. When conflicts arise, fire missions may require modification or delay to avoid fratricide and unintended effects. In congested or contested environments, those demands slow counterfire execution greatly, particularly when airspace data is fragmented across systems or updated manually.

Although the sensor-to-shooter system and airspace deconfliction requirements constrain both proactive and reactive counterfire, the sensors that support each type of counterfire generate different timing and integration challenges. Those challenges shape where emerging AI-enabled systems can most improve counterfire performance. Proactive counterfire is inherently anticipatory, relying on an intelligence and surveillance network that collects, correlates, and predicts enemy firing patterns before rounds are in flight. In execution, collection and fires systems often remain poorly integrated, limiting the Army’s ability to anticipate and preempt enemy fires. However, emerging programs seek to close these seams by enabling predictive, lawfully grounded proactive targeting.

Reactive counterfire, by contrast, is triggered only after an adversary fires, relying primarily on weapons-locating radars (most commonly the Q-53) which provide near-instantaneous point-of-origin data. When communications and commanders function effectively, this process can allow a counterfire mission before the enemy artillery unit that fired moves.

The Q-53’s speed and geolocation precision make it the centerpiece of reactive counterfire, but its acquisition is transient: the data transmitted from the radar to its controllers shows where a weapon fired from at one moment, not the surrounding context or whether the weapon still sits at that firing point. Without corroborating intelligence or pattern analysis, a raw radar-derived origin may propose an engagement in a restricted area or on against a location the target has moved from. These constraints bear directly on legal assessments and on whether immediate engagement is lawful. This imbalance between speed and context is what drives recent investment in AI systems to improve data accuracy and shorten engagement time.

Three systems, Maven Smart Systems (MSS), Tactical Intelligence Targeting Access Node (TITAN), and Air Space Total Awareness for Rapid Tactical Execution (ASTARTE), are designed to close the gap between rapid detection and reliable engagement. MSS accelerates early target development and pattern recognition; TITAN pushes coordinated and corroborated intelligence to the low-level units; and ASTARTE accelerates deconfliction in complex air-ground environments. These systems aim to accelerate decision timelines while preserving contextual clarity, improving the quality and tempo of all targeting decisions. Further publicly available details about each system illuminate the goals and limits of these gap-closing efforts in counterfire operations.

MSS

Maven Smart Systems is a customizable, software-based decision-support node that ingests and processes data from multiple battlefield sensors into a common operating picture: “put simply, MSS is an AI-based decision support system.” It fuses and scores sensor data with machine learning (ML)/rule-based algorithms to create a prioritized queue of targets and pre-populated targeting packages. MSS is designed to hand off nominations to fire-control systems or to present them to operators for human validation. Its primary role is to accelerate the sensor-to-shooter timeline, not to function as an autonomous lethal system.

MSS is ideal for integration into proactive counterfire, increasing the efficiency of planning and response times. MSS provides “easy access to sensor and image data from commercial and military satellites,” and, critically, facilitates post-strike battle damage assessment, a step that is often delayed or omitted.

Recent reporting confirms that MSS has moved beyond development into sustained use. During recent Army training exercises, MSS integrated satellite imagery and intelligence feeds from all services of the armed forces and across security classification levels, accelerating target identification and validation within existing approval processes. The unit participating in the exercises achieved the same level of efficiency as targeting cells in Operation Iraqi Freedom (OIF), considered the most efficient to date, “with roughly 20 people in its targeting cell, whereas the OIF cell benefitted from more than two thousand staff members.”

MSS has also been employed operationally, including at a U.S. airbase in Qatar. The Qatar-based U.S. lethal fires element using the technology noted “[it] is not replacing the Intelligence Analyst’s job but simply speeding up the processes while enhancing workflow and efficiency.”

TITAN

Tactical Intelligence Targeting Access Node is the U.S. Army’s next-generation, software-centric intelligence and targeting ground station built to ingest and fuse “sensor data received from space, high altitude, aerial, and terrestrial layers” using AI and ML. The goal of TITAN is to “ultimately reduc[e] the sensor-to-shooter timeline.” To achieve that goal, Palantir designed TITAN “to strengthen the connection between data-collecting sensors and the weapons and decision-makers on the ground, improving the accuracy and speed of long-range targeting.”

 TITAN fuses information from dispersed battlefield and global sensors into usable targeting data for modern fires. Within the Army’s targeting environment, TITAN’s relevance is immediate. As explained by targeting technician Jordin Katzenberger, integrating AI and ML “will improve the accuracy, efficiency and effectiveness of target acquisition and engagement,” decrease the chance of unintended harm, and “enhance … commanders’ comprehension of the operational environment in real time.” TITAN’s ability to collect, process, and share intelligence simultaneously across levels of units permits it to function as a “force multiplier,” permitting smaller units to achieve outcome previously attainable only by larger units. TITAN functions as a meeting and integration point for the counterfires cell, enabling more effective development counterfire missions from Q-53 radar acquisitions of enemy firing.

TITAN’s capacity to analyze vast amounts of sensor data has the potential to further streamline target assessment and convert the Q-53’s point-of-origin data into more timely and accurate targeting information. Presently, a counterfire cell relies primarily on such acquisition data to generate reactive counterfire missions. TITAN contributes wider context to those missions by locating and integrating intelligence beyond the initial point of origin.

Comparing MSS and TITAN clarifies their distinct operational implications. MSS is a comparatively mature, software-centric decision support platform that the Army has broadly deployed across intelligence and operational contexts to assist commanders in aggregating, prioritizing, and evaluating information beyond weapons employment. TITAN, by contrast, reflects a later stage in the Army’s integration of AI into command-and-control architectures focused specifically on supporting targeting. Army program materials describe TITAN as an AI-enabled intelligence and targeting ground station integrating hardware, software, and various sensor feeds. This choice narrows TITAN’s functional focus and places it in closer proximity to weapons employment, particularly in time-sensitive and reactive counterfire operations.

Unlike MSS, which primarily supports target nomination and information prioritization, TITAN is designed to operate at the seam between intelligence collation and processing on one hand and fire mission initiation on the other. TITAN can consolidate Q-53 radar acquisitions with corroborating intelligence, confidence assessments, and targeting parameters into a structured recommendation.

ASTARTE

The Air Space Total Awareness for Rapid Tactical Execution system is an AI-enabled software system designed to solve one of the most persistent friction points in fire mission processing: airspace management in crowded, contested battlespaces. ASTARTE integrates data from existing command and control systems to generate a live picture of airspace. Joint exercises have already demonstrated ASTARTE’s ability to effectively synchronize air and ground effects across the armed services.

ASTARTE supports both proactive and reactive counterfire. In proactive counterfire, the system supports by “integrating with existing command and control systems, predicting airspace usage, and improving speed by reducing time for planning and generating courses of action for joint fires.” Recalling, however, that reactive counterfire unfolds under compressed timelines and elevated risk, ASTARTE alerts potential firing units to critical constraints. The Defense Advanced Research Project Agency describes ASTARTE as “enabling safe, simultaneous operation of manned and unmanned aircraft, missiles, and artillery fire in the contested airspace above an Army division,” underscoring its role in allowing rapid fires execution in contested airspace.

Designers also built ASTARTE for broad accessibility. The system can integrate with existing command and control networks, suggesting accessibility for both relatively low-level staffs as well as higher headquarters, using compatible communications infrastructure. This design choice aligns with the Army’s recognition that airspace management is a cognitive and coordination problem, not merely a technical one. Army program materials emphasize that modern airspace management must integrate multiple levels of effort while converting large volumes of data at the pace of modern operations. As a DefenseScoop report confirms, the Army increasingly views AI-enabled airspace management as essential to reducing commanders’ “cognitive burden,” particularly in fast-moving fires scenarios. At the same time, this increased accessibility and fidelity alter the informational baseline against which commanders make decisions, narrowing the range of uncertainty that has historically justified accepting higher risk.

Conclusion

By providing commanders with real-time airspace awareness, automated deconfliction, and AI-generated courses of action, these systems not only accelerate counterfire but reshape the decision environment itself. They demonstrate how AI is transforming counterfire operations by accelerating targeting, reshaping command relationships, and expanding battlefield visibility.

Part II of this post will explore the limits of the law of war across the common and distinct features of each system.

***

Captain Megan Ezekannagha is a J.D. Candidate at The University of Texas School of Law and Field Artillery Officer in the Texas Army National Guard. 

The views expressed are those of the author, and do not necessarily reflect the official position of the United States Military Academy, Department of the Army, or Department of Defense.

Articles of War is a forum for professionals to share opinions and cultivate ideas. Articles of War does not screen articles to fit a particular editorial agenda, nor endorse or advocate material that is published. Authorship does not indicate affiliation with Articles of War, the Lieber Institute, or the United States Military Academy West Point.

 

 

 

 

 

 

 

 

Photo credit: U.S. Air Force