Artificial intelligence does not yet autonomously guide U.S. missiles, but it has already deeply entered the process leading to the identification, selection, and engagement of a target. The key name is Maven Smart System (MSS), the platform developed by Palantir as an evolution of Project Maven, the program launched by the Pentagon in 2017 to accelerate the analysis of images, videos, and data collected from drones, satellites, and other sensors.
Over time, Project Maven has become one of the symbols of the digital transformation of modern warfare. Born to help the military analyze vast amounts of data more quickly, today it represents a much more complex system: a platform that integrates geospatial intelligence, ISR images, information flows from the battlefield, and artificial intelligence tools capable of supporting the decision-making process.
Maven is managed within the orbit of the National Geospatial-Intelligence Agency (NGA) and is designed to be used across the different branches of the U.S. armed forces. Its goal is not only to “see” the battlefield better but also to drastically reduce the kill chain times, that is, the cycle from discovering a target to deciding to strike it.
In recent months, the platform has returned to the spotlight for its alleged use in the context of U.S. operations against Iran. On this point, it is important to distinguish between what is documented and what emerges from journalistic reconstructions: it is confirmed that Maven is a significant component of the American military architecture and that new generation AI tools have been integrated into operational workflows; more delicate, however, is the precise confirmation of tactical details related to individual operations.
The essential point, however, remains clear: Maven Smart System is not software that directly pilots weapons, but a platform that helps collect, sort, analyze, and correlate data, providing commanders with a faster and more structured informational basis for decision-making.
How it works: data, images, AI, and target selection
From a technical standpoint, Maven Smart System operates on multiple levels. The first is data collection and normalization: satellite images, drone videos, tracks, ISR reports, and operational intelligence are integrated into a common environment. The second is automated analysis, where computer vision algorithms and AI models help identify objects, suspicious movements, military vehicles, sensitive infrastructures, or anomalies on the ground.
The third level is the most delicate: decision support. Once the data is processed, the system can present operators with a series of options, priorities, and correlations useful for building the operational picture. In practice, it does not “fire,” but suggests. It can help narrow down a list of possible targets, link a vehicle to a logistics network or a military structure, estimate the relevance of a target, and even indicate which asset might be most suitable for a potential engagement.
It is in this step that AI truly changes the way of fighting. Not because it replaces the human decision-maker, but because it increasingly influences the speed, field of view, and priority hierarchy. A system of this type allows operations that once required hours, distinct teams, and different software to be compressed into a few minutes.
Even DifesaNews has highlighted this aspect, explaining how Maven's evolution aims to overcome traditional information silos between sensors, drones, command centers, and tactical systems. The value of the system lies precisely in its ability to transform the battlefield into a continuous network of shared data, updated in real-time and immediately usable by decision-making levels.
Making this transformation even more significant is the integration of Claude models from Anthropic within the Palantir ecosystem. Generative artificial intelligence is not only used to classify images but also to facilitate natural language queries, intelligence synthesis, complex data analysis, and operational planning support. In other words, the leap is not only technological: it is also cultural. The military no longer interacts only with maps and feeds but also with interfaces capable of responding, summarizing, connecting, and suggesting.
The risks: errors, responsibility, and the future of automated warfare
If Maven Smart System represents one of the most advanced systems available today for targeting support, its use inevitably raises enormous questions. The first concerns reliability. No AI system is infallible, especially in high-intensity scenarios, with incomplete data, delayed updates, degraded images, or ambiguous operational contexts. A geolocation error, a wrong classification, or an incorrect correlation can have devastating consequences.
For this reason, proponents of these platforms insist that the final decision remains human. But it is precisely here that the most controversial issue lies. When a system reduces decision time, orders priorities, proposes alternatives, and filters available information, the risk is that human control increasingly becomes a quick ratification of a chain of suggestions produced by the machine.
In the case of operations against Iran, public attention has focused precisely on this: not so much on the idea of a completely autonomous weapon, but on the growing dependence on digital ecosystems that transform vast amounts of data into operational targets. The decision to strike formally remains entrusted to a human being, but the context in which that decision matures is now deeply shaped by artificial intelligence.
The issue is not only military. It is also political, industrial, and strategic. Those who control these platforms also partly control the pace of war. This is why companies like Palantir, cloud providers like AWS, and AI model developers like Anthropic have become increasingly relevant players in the contemporary defense system.
The Maven case thus shows a truth destined to weigh more and more in the years to come: the war of the future will not only be made of missiles, drones, and satellites, but of platforms capable of transforming data into decisions. And the most delicate point is not to determine if AI is already pulling the trigger, but to understand how decisive it has already become in choosing where to aim it.
Comments
No comments yet. Be the first!