Syria was the first social media war, where the Syrian government harnessed the power of social media to spread misinformation. Ukraine was the first drone war, taking combat beyond the trenches. Now, the Iran conflict is the first artificial intelligence war, as the world’s strongest military embraces the technology.
Admiral Brad Cooper, the head of US Central Command, said earlier this month the Pentagon was “leveraging a variety of advanced AI tools” to sift through “vast amounts of data in seconds” and make “smarter decisions faster than the enemy can react.” Here’s how AI is being used in the conflict so far.
Targeting. The US military has been preparing to use AI for years. In 2017, the Pentagon launched Project Maven, an initiative to use computers to sort through swaths of video footage and other data. Palantir took the lead on the project. Creating Maven Smart System, a $1.3 billion piece of software that combines battlefield data, intelligence reports, and surveillance footage on a map, and uses AI to detect targets. The system is integrated with Anthropic’s Claude large language model (LLM), which processes the data and turns it into actions, recommending what should be hit and with what weapon, in real time.
A human still makes the final decision, but the model drastically speeds up the “kill chain,” or the process of identifying, prioritizing, and eviscerating the target. Reporting suggests Maven helped the US strike more than 1,000 targets in the first 24 hours of the campaign.
Bureaucracy. “When people think about the military using AI they picture Iron Man and autonomous weapons,” says Eurasia Group Geotechnology expert Scott Bade,, “But that's not really the way AI is being used." Instead, much of AI’s impact is much more mundane. According to Bade most of the military’s AI integration is bureaucratic and operational. “The Pentagon is a giant office building,” meaning that AI is largely being used for predictive maintenance, logistics, and moving supplies more efficiently.
Intelligence. But AI is also transforming military intelligence. Some military officials claim they have produced intelligence reports that “no human hands” have touched, as AI sifts through vast transcripts far faster than any human can and quickly draws insights. “There used to be data that the military had but couldn’t use because it didn’t have the capacity to analyze it,” says Bade.
But an AI’s outputs are only as good as its inputs. LLMs, in particular, can produce recommendations without clear explanations, which can make it difficult for military personnel to assess their accuracy.
The concerns. The use of AI is raising concerns about the future of combat. This was shown in Israel’s use of AI-assisted targeting in Gaza after Oct. 7, where the accusation was not that a machine was independently choosing who lived or died, but that operators had become too comfortable with its recommendations. “They just accepted what the machine recommended pretty uncritically,” Bade said, “The IDF did some tests and decided they were confident with the margin of error.” The fear is not that killer robots could take over the military, but of automation bias: the tendency to trust a fast, confident system more than a slow, skeptical human.
AI is becoming a focal point when things go wrong, even when there isn’t evidence that it was used. On the first day of the US-Iran war, a Tomahawk cruise missile hit an elementary school in Minab, southern Iran. At least 168 people were killed, many of them children. Reports showed the US military had thought it was hitting a nearby Islamic Revolutionary Guard Corps naval base.
Further investigations found that outdated intelligence was likely to blame for the mistaken targeting. Nevertheless, in the days that followed, a group of Democrats in Congress called for an investigation into whether Maven was used and for oversight over AI’s use on the battlefield. “I think a lot of people have been quick to jump to blaming AI,” said Bade, when human error is to blame.
“Supply chain risk.” While the US uses AI in the Iran conflict, it is also in an active feud with one company behind the technology. Earlier this month, the Pentagon took the rare step of designating Anthropic a “supply chain risk” – effectively blacklisting it – for guardrails on its technology. The designation means that federal agencies – and anyone who wants to do business with the US military – must stop using software from the company. Anthropic is suing in response, alleging that the designation is punitive and constitutes a breach of contract.
Regardless of how the lawsuit pans out, the bigger question may be whether the Pentagon can disentangle the technology from its systems in the middle of an active conflict. In July 2025, the government signed a $200 million contract to embed Anthropic into its workflows. Military contractors have said it would take months to extract Claude from its processes, and that it would not be “quick nor painless.” It powers software like Project Maven, which has been incorporated across all branches of the military, making it inextricable – especially in the middle of the war.
It is unprecedented for the government to pick a fight with such an embedded part of its defense ecosystem in the middle of a war. And that may be the clearest sign yet of how important AI has become to modern warfare. It’s no longer the stuff of distant science fiction, but part of the machinery of war itself.
