From Mother Jones · Sophie Hurwitz · · 2m
Killer AI is already here, we're using it in Iran
The US military is deploying AI to select bombing targets in Iran, but most Americans have no idea it's happening.
In brief
The US military is deploying AI to select bombing targets in Iran, but most Americans have no idea it's happening. AI systems are actively selecting targets for lethal strikes in the Iran war with minimal human oversight, yet this reality is absent from public AI safety debates. Originally reported by Mother Jones.
Americans fear AI but don't know the truth
Over 60 percent of Americans worry AI will kill us all, yet a new poll shows almost no one knows the US military is already using AI to select bombing targets in Iran.
The Iran war is an AI warfare experiment
The US-Israel conflict with Iran represents an unprecedented application of AI in warfare, yet most Americans remain unaware their military is deploying these systems.
Most Americans have heard nothing about it
Only 8 percent of Americans have heard a lot about military AI use in Iran. Two-thirds have heard only a little, and 40 percent have heard nothing at all.
AI deployed 2,000 missiles in 96 hours
The Pentagon's AI systems deployed over 2,000 munitions to 2,000 distinct targets in just 96 hours during Operation Epic Fury, according to the Defense Department's chief AI officer.
Claude AI ranks targets for bombing
The Pentagon's Maven Smart System uses Anthropic's Claude AI to semi-autonomously rank bombing targets and draft legal justifications for strikes, the same tool people use for homework and emails.
“People would be pretty shocked that what they're using for their homework or to summarize emails is also being used by intelligence officers.”
AI accuracy drops sharply in bad weather
Claude's target selection is only about 60 percent accurate in good conditions and 30 percent accurate in bad weather, yet the Pentagon claims humans always make final bombing decisions.
School strike killed 150 children
In February, US missiles struck a girls' elementary school in Minab, Iran, killing over 150 people, mostly young children. Seven months later, UN experts called it a crime against humanity.
Pentagon struck 13,000 targets in 38 days
In the first week of the Iran war, the Pentagon struck over 3,000 targets. By the time a ceasefire was announced 38 days later, that number had risen to 13,000 targets.
Tech companies are writing the rules
The Pentagon has placed few checks on what technology companies can do during wartime. Companies are essentially regulating themselves, with their terms of service defining AI use in war more than any laws or government guidelines.
“Their terms of service are defining how AI systems are being used for war to a greater extent than any laws that have been written.”
The bottom line
AI systems are actively selecting targets for lethal strikes in the Iran war with minimal human oversight, yet this reality is absent from public AI safety debates.
