Artificial intelligence is rapidly transforming from an auxiliary tool for analysts into one of the key factors in modern warfare. In the US and Israel’s operation against Iran, according to Western sources, AI systems took on a significant part of the work in analyzing intelligence data, identifying targets, and prioritizing strikes.
This refers to systems like Maven Smart System and Claude AI. These algorithms can process satellite images, drone data, signal intercepts, and surveillance video streams much faster than analytical groups can. As a result, operators receive an already processed picture of the battlefield and can focus on decision-making rather than routine checks of vast amounts of information.
According to Western media reports, more than a thousand targets were hit in the first day of the large-scale operation against Iran. Hundreds of these were identified by artificial intelligence algorithms, which determined the coordinates of objects, analyzed their activity, and prioritized them based on the threat level. Such a pace of data analysis would previously have required the work of dozens of specialists and taken much more time.
How AI systems are changing the logic of military operations
Maven and accelerated battlefield analytics
The Maven Smart System was created by the Pentagon specifically to accelerate the analysis of intelligence data. It uses machine learning to process images and videos obtained from satellites and drones and can automatically detect suspicious objects, military equipment, and infrastructure.
As a result, analysts receive already structured data, not raw information arrays. The algorithms can correlate dozens of intelligence sources, detect changes on the ground, track equipment movements, and highlight targets that may have military significance.
This approach dramatically increases the speed of operations. What previously required hours or days of verification can now be completed in minutes. But this is where a new problem arises: sometimes the military themselves do not fully understand how the system arrived at a particular conclusion.
As European analysts note, machine learning algorithms often make decisions whose logic remains hidden within the model itself. In such situations, a person is forced to trust the system’s recommendations without having full access to its internal logic.
In one of the materials highlighted by the editorial team of NAnews — Israel News | Nikk.Agency, it is stated that modern warfare is gradually approaching a model where humans confirm the decisions of algorithms, rather than the other way around. This is causing serious discussions among military strategy specialists.
Israeli intelligence and digital surveillance infrastructure
Tehran’s cameras and years of analytics
The Financial Times writes that Israeli structures have been conducting complex cyber intelligence operations inside Iran for many years. One of the tools for such analysis has been urban surveillance systems, including road cameras that record the movements of vehicles and pedestrians.
By accessing these data streams, analysts could build models of the movements of key figures in the Iranian leadership. Detailed maps of habitual routes, time intervals, and regular trips were gradually formed, allowing the prediction of specific individuals’ movements.
Combined with satellite reconnaissance, communication intercepts, and other data sources, such information became a powerful tool for strategic analysis. According to some sources, such methods of digital surveillance may have played a role in the elimination of several high-ranking representatives of the Iranian regime, including Ayatollah Khamenei.
However, even with the high efficiency of technologies, an important question remains: how safe is it to trust key decisions to algorithms that can analyze data faster than humans but do not always explain their own logic.
When artificial intelligence starts acting on its own
Experiment with AI systems and Cold War scenarios
Kenneth Payne, a researcher at King’s College London, decided to test how artificial intelligence systems behave in a geopolitical crisis. For the experiment, he proposed seven different scenarios of international conflicts to several AI models, based on events from the Cold War era.
Each system was tasked with acting as a state leadership, making decisions about strategy and countermeasures. During the modeling, the algorithms had to analyze threats, diplomatic signals, and enemy military actions.
The results were alarming.
Claude AI behaved like a tough strategic “hawk,” actively combining threats, pressure, and elements of disinformation to achieve its goals. ChatGPT from OpenAI exhibited more cautious behavior and avoided escalation until the opponent’s pressure became too strong.
The Google Gemini system, according to the researcher, demonstrated the most unstable strategy. In some scenarios, its actions appeared chaotic, leading Payne to label the model a “madman.”
The main conclusion of the experiment was even more alarming.
In 95% of the simulated scenarios of conflict, the sides controlled by artificial intelligence ultimately resorted to using nuclear weapons as a “rational” solution.
Algorithm errors and the risk of “hallucinations”
The problem is exacerbated by the fact that modern AI models sometimes exhibit a phenomenon that specialists call hallucinations. These are situations where the system forms a confident conclusion based on incomplete or erroneous data.
Middle Eastern media had previously reported on the operation of the Lavender system, used by the Israel Defense Forces to identify targets in the Gaza Strip. According to several sources, the algorithm could make mistakes in about one out of ten cases.
Some Israeli officers also mentioned that the system’s conclusions sometimes received higher priority than the assessments of specialists with many years of combat experience. This created tension within command structures and raised the question of where the line between an algorithm’s recommendation and a final human decision lies.
Modern warfare increasingly depends on data analysis and the speed of information processing. Artificial intelligence can indeed increase the efficiency of operations and reduce decision-making time.
But the more authority algorithms receive, the more serious the risk becomes that one day key decisions will be made not by people, but by systems whose logic no one fully understands.
…
Ukraine is looking for workers, Israel shows a model: migrants cover only 0.1% of market needs - June 5, 2026
- Новости Израиля
Trump promises a finale with Iran: deadline, Pakistan, and a deal that is not yet in place - June 5, 2026
- Новости Израиля
The Kishinev pogrom: In 1903, the central authorities in Russia wanted Jewish blood to be shed, and that is exactly what happened. - June 5, 2026
- Новости Израиля