A symbol of a strategic turning point: It shows how artificial intelligence is finding its way into modern warfare – and at the same time triggering heated ethical debates.image: Shutterstock/montage watson
Project Maven plays a central role in US military operations, particularly against Iran.
April 5, 2026, 9:48 p.mApril 5, 2026, 9:48 p.m
Alex PIGMAN / afp
A Pentagon AI program, “Project Maven,” is at the center of the military campaign against Iran. A completely new role that shows how modern warfare is fundamentally changing – driven by innovations from Silicon Valley, which is itself wrestling with big ethical questions.
translation
This text was written by our colleagues from French-speaking Switzerland and we translated it for you.
What is that?
“Project Maven” is the US Army’s most important AI program. It was launched in 2017 as an experiment to help military analysts evaluate the vast amounts of drone imagery.
The evaluators were completely overwhelmed – they had to go through image after image to find clues that are often only visible for a fraction of a second. “Maven” was developed to change exactly that: find the needle in the haystack faster.
Eight years later, the program has been massively expanded. It has developed into an AI-supported system for target acquisition and combat management – and has multiplied the speed of the “kill chain”, i.e. the process that extends from the discovery of a target to its destruction (in NATO language: effect chain).
How does this work?
“Maven” combines the functions of military air traffic control with those of a cockpit. Expert Aalok Mehta describes the system as “a kind of additional layer” that brings together sensor data, satellite images and information about own and enemy forces.
Specifically, explains the director of the Wadhwani AI Center in Washington, “Maven” analyzes satellite data at high speed to detect movements or identify targets – and in the process creates “a real-time image of the operational area” to determine the optimal attack plan.
“Maven” turns a detected threat directly into a targeting process: It evaluates available options and presents leadership with a selection of possible courses of action, as a Pentagon representative explained in a recent online demo.
The rise of generative AI in the last three years has made another leap possible: You can now interact with the system in natural language – making the technology suddenly accessible to non-technical people in the military.
This ability is currently provided by the Claude the Anthropic model – although it may not be for long. The US start-up was sanctioned by the Pentagon after publicly refusing to use its AI for fully automated attacks or to monitor American citizens.
Why did Google reject it?
The ethical questions surrounding AI were already controversial in the early years of the program – back when Google was still the original contractor. In 2018, more than 3,000 employees of the Mountain View tech giant signed an open letter against the project because, for them, it crossed a red line. Several engineers took the consequences and resigned.
Google later refused to renew the contract and published an AI charter that barred any involvement in weapons systems. The episode exposed a deep rift in Silicon Valley: on the one hand, engineers who see autonomous targeting as an ethical red line – on the other hand, military officials who consider it indispensable.
Google has since lifted these restrictions and announced that it will become more involved in military contracts. In addition to Google, OpenAI – Anthropic’s main competitor – and Elon Musk’s xAI company are also in the running to replace “Claude” in “Maven”, as the Pentagon announced.
What role does Palantir play?
In 2024, Palantir Technologies took over the space vacated by Google. The company, which was set up in part with money from the CIA, was strongly focused on intelligence work from the start.
The company has now become the main contractor for “Maven” – and its AI technology forms the operational backbone of the program. For CEO Alex Karp, the world is divided into two camps: those who have this technology – and those who don’t.
For him, it is crucial that the West master skills that the rest of the world does not have. A system that can compress several hours of the “kill chain” into a few seconds simply makes every opponent obsolete, he believes.
What results have been achieved so far?
The Pentagon and Palantir Technologies did not want to comment on the specific achievements of “Maven” in the war against Iran. However, the high pace of US attacks suggests that the system has significantly accelerated the process of target acquisition and strikes.
In the first 24 hours of Operation Epic Fury, US forces attacked more than 1,000 targets. According to consistent media reports, this also includes the fatal attack on a school that was housed in a former military building. A Pentagon investigation is ongoing. (dal)