A conflict born of violence, amplified by algorithms
On February 28, 2026, at 2:30 a.m. Washington time, Donald Trump announced in a video posted on Truth Social that the U.S. military had launched major combat operations in Iran. A U.S. official confirmed to Reuters that the strikes were being conducted by air and sea. This marked the start of Operation Epic Fury, the largest U.S. military intervention in the Middle East since the 2003 invasion of Iraq, and one of the first conflicts in which artificial intelligence has played a central operational role on both sides of the front.
Project Maven, the AI that guides bombs
At the heart of the U.S. military machine lies a low-profile but redoubtable program: Project Maven. Launched in 2017 as an experiment to help military analysts process the massive stream of images sent by drones, Maven has expanded significantly over the past eight years. It has become an AI-assisted targeting and battlefield management system, capable of dramatically increasing the speed of the kill chain.
Specifically, Maven rapidly analyzes satellite imagery to detect movements or identify targets, while providing an instantaneous overview of the operational theater to determine the best strike plan. The system transforms a detected threat into a targeting process, evaluating available solutions and presenting command with a range of options, according to a Pentagon official during a recent online demonstration.
The emergence of generative AI over the past three years has marked a major leap forward, enabling users to interact with the system using natural language and thereby making this technology accessible to a wider audience beyond just military technicians.
When the algorithm misses the mark
The system’s effectiveness raises serious questions. During the first 24 hours of Operation Epic Fury, U.S. forces struck more than 1,000 targets. This unprecedented pace had dramatic consequences on the ground.
On February 28, 2026, the Shajareh Tayyebeh Girls’ Elementary School in the Shahrak-e Al-Mahdi neighborhood of Minab, in Iran’s Hormozgan Province, was destroyed by missile strikes. According to Iranian media, between 175 and 180 people were reportedly killed, the majority of whom were schoolgirls.
Investigations conducted by several major international media outlets quickly established that the United States was reportedly responsible. According to sources familiar with the initial findings, the strike on the Shajareh Tayyebeh school in Minab occurred while the U.S. military was targeting a nearby Islamic Revolutionary Guard Corps naval base. U.S. Central Command had generated the target coordinates based on outdated information provided by the Defense Intelligence Agency, which contributed to the error, according to CNN.
In an official statement, United Nations experts expressed their deep dismay and described the strike on a school as a “grave violation against children, education, and the future of an entire community,” noting that schools are civilian objects expressly protected under international humanitarian law.
Anthropic says no, OpenAI says yes: the U.S. tech industry is divided
The issue of technology providers reveals a deep divide within Silicon Valley. Ethical concerns about AI were already a sensitive issue in the program’s early years. In 2018, more than 3,000 Google employees signed an open letter denouncing their contract with Maven, and several engineers resigned. Google subsequently refused to renew the contract and published an AI charter excluding any participation in weapons systems.
But those days of scruples seem to be over. Google has reportedly lifted some of its restrictions and announced that it will become more involved in military contracts. OpenAI and Elon Musk’s company xAI are also in the running to replace Claude in Maven, the Pentagon said.
After Anthropic was classified as a “supply chain risk” by the Pentagon for refusing to allow its tools to be used for fully automated strikes or mass surveillance, OpenAI landed a defense contract worth hundreds of millions of dollars. A symbolic turnaround for a company that, until recently, presented itself as a champion of the ethical use of AI.
On social media, Iran scores its first victories
On the digital front, Tehran is waging a counteroffensive unlike any seen before. Within 24 hours of the launch of the U.S. and Israeli attacks, dozens of social media accounts affiliated with the Revolutionary Guards began spreading Iranian propaganda messages, some of which reached several million people, according to a study by Clemson University in South Carolina.
AI-generated videos depicting Donald Trump and Israeli Prime Minister Benjamin Netanyahu as Lego figures have flooded social media. Some were accompanied by original rap music, likely also generated by AI. These videos are attributed to the Revayat-e Fath Institute, an Iranian state-affiliated organization, and have been broadcast on Iranian state television channels as well as circulating online.
The goal goes beyond mere entertainment: Iran seeks to undermine Trump by targeting his key political vulnerabilities, blending geopolitical grievances, meme culture, and popular imagery to reach fragmented Western audiences, according to Nancy Snow, a professor specializing in propaganda.
Fake images, real confusion: AI blurs the lines
Beyond trolling, AI is being used to fuel deliberate confusion. This conflict is the first in which AI-generated content is being intentionally used to sow chaos and confusion about what is actually happening on the ground, according to expert Melanie Smith. Censorship rules restricting the flow of information—both from Israel and from Iran, which has imposed a total internet blackout—only exacerbate this confusion.
And the United States is not immune to criticism in this area. Donald Trump has accused Iran of using AI as a weapon of disinformation, even as the White House has faced sharp criticism for releasing a video that mixed real footage of strikes in Iran with clips from action movies and video games.
As the propagandist Jacques Ellul theorized as early as the 1960s, propaganda evolves alongside the communication systems that carry it. In a social media landscape shaped by algorithms and generative AI, it now takes the form of content designed primarily to spread rather than to persuade.
A war consumed like reality TV
Emerson Brooking, of the Atlantic Council’s Digital Forensic Research Lab, points out that war is increasingly being subsumed by the attention economy, becoming a form of commodification of conflict—a strange and unsettling experience for those living through it in real time.
The war in Iran marks a significant turning point. Artificial intelligence is now simultaneously a tool for lethal decision-making on the battlefield, a vehicle for viral propaganda, and a central element in competition between global tech companies.
