The strike on the Shajareh Tayyebeh girls’ school in Minab has been framed by many as a grim preview of AI warfare gone wrong. But the harder truth is less futuristic and more familiar: stale intelligence, compressed timelines, thin oversight, and human judgment hollowed out by process. This is the story of what happens when weak tradecraft gets dressed up as precision.
Stale intelligence, shallow review, false confidence, dead children.
Akbare Fori Minab – Follow-on Tomahawk Strike on Shajareh Tayyebeh girls’ school
Minab Wasn’t an AI Accident. It Was a Tradecraft Failure at Maven Speed.
Explore this article’s interactive infographic
The strike on the Shajareh Tayyebeh girls’ school in Minab, Iran should kill off one lazy habit fast: blaming “the AI” as if the machine wandered off on its own. That is not what this looks like.
Based on our assessment, this appears to be something more dangerous and more familiar. A target deck built on stale data. A civilian site left inside a military database long after reality changed. No meaningful pattern-of-life work. A kill chain built for speed. And human beings still present in the loop, but present in the weakest possible way: not as a brake, not as a source of judgment, but as a legal and procedural signature at the end of an already compressed process.
That matters because it is tempting to talk about the Minab strike as a tech story. It is not primarily a tech story.
It is a tradecraft story. A bad one.
The Real Failure Wasn’t the Interface
Available reporting indicates that Operation Epic Fury unfolded as a high-tempo, AI-augmented campaign in which nearly 900 strikes were executed in the first twelve hours, with the Maven Smart System helping compress the kill chain from days to minutes. That kind of speed is exactly what defense bureaucracies and tech firms have been selling for years: decision superiority, tighter fusion, faster targeting, less friction. On paper, that sounds clean. In practice, it gets weird fast when the underlying data is wrong and nobody slows the machine down long enough to notice.
And Minab is what that looks like.
At roughly 08:04 UTC on February 28, 2026, a U.S.-made Tomahawk hit the Shajareh Tayyebeh primary school in Minab while classes were in session, killing at least 175 civilians, most of them girls between seven and twelve. Reporting suggests the site remained classified in a DIA database as a military facility because it had once been associated with an adjacent IRGC naval base. But the broader picture suggests the site had been physically separated from that base by 2016 and had functioned as a school for roughly a decade, with play areas, sports space, and a visible digital footprint that should have made its civilian use hard to miss.
That is not a sensor problem. That is not some exotic edge-case failure of machine learning.
That is database inertia meeting institutional complacency.
Speed Turned Bad Data Into Lethality
In our assessment, the most damning detail is not simply that the data was old. It is that basic targeting discipline appears to have thinned out under pressure.
Reporting suggests there was no meaningful pattern-of-life study before the strike, and that even minimal surveillance likely would have revealed the presence of children and school staff. The same body of reporting also points to the larger compound, including a clinic and pharmacy, being hit in a way that suggests the whole area was treated as one military object despite years of clear civilian use. That is not precision. That is administrative certainty masquerading as operational judgment.
This is where the Maven conversation needs to get more honest.
As best as we can tell from the available reporting, the Maven Smart System pulls together satellite imagery, SIGINT, sensor data, and battlefield reporting into a single targeting interface. It uses automated recognition, sensor fusion, confidence scores, and weapon-pairing recommendations to move targets through a streamlined workflow. Analysts review confidence scores. Humans verify target identity. Commanders sign off on strikes. That all sounds reassuring until you ask the obvious question: what happens when humans are technically involved but functionally outpaced?
That is the real problem.
The “Human in the Loop” Can Become a Legal Fiction
The United States loves the phrase “human in the loop.” It sounds ethical. It sounds controlled. It sounds like a firewall against catastrophe.
But Minab points to a much uglier reality. Human involvement can degrade into rubber-stamping when analysts are pushed through massive target volumes and commanders are asked to bless machine-shaped decisions at operational speed. The available reporting explicitly points toward automation bias here: the tendency to over-trust automated output, especially when it arrives wrapped in clean confidence scores and machine precision. In that environment, the human is not really exercising judgment. He is validating throughput.
That distinction matters more than all the glossy AI branding in the world.
Because the danger in modern warfare is not just autonomy in the sci-fi sense. It is pseudo-deliberation. It is a command climate where the machine recommends, the staff processes, the commander signs, and everyone later points to everyone else. The software did not launch the missile by itself. But neither did a commander meaningfully re-interrogate the target the way the laws of war and basic professional competence should demand.
That is how accountability gets diluted without ever disappearing.
Palantir’s Maven Smart System Didn’t Remove the Human. It Thinned the Human Out.
In our assessment, the Minab strike exposed an accountability vacuum: if an AI-enabled system recommends a target using outdated data and a human approves it under compressed timelines, responsibility becomes diffuse enough to be politically survivable and morally evasive. That may be administratively convenient. It is not strategically healthy, and it is not ethically serious.
There is another lesson here, and it should make U.S. planners uncomfortable.
The future fight will not simply reward whoever has the faster algorithm. It will punish whoever starts believing the algorithm is a substitute for tradecraft. Available reporting suggests these systems remain brittle, vulnerable to data poisoning, false positives, and physical deception. In other words, the same force that skips target validation because the dashboard looks clean is setting itself up to be spoofed, flooded, and manipulated by adversaries who understand that computer vision and machine learning are not magic. They are systems. Systems can be gamed.
What This Means for the U.S. Way of War
That has direct implications for U.S. doctrine and force design.
If the American answer to future war is faster fusion plus thinner human review, then Minab is not an anomaly. It is a warning shot. The operational appeal is obvious: compress the chain, collapse the timeline, hit more targets, stay inside the enemy’s decision cycle.
Fine.
But if your back-end data governance is sloppy, if your analysts are trained to manage queues instead of interrogate reality, and if your commanders get habituated to trusting the machine’s framing of the problem, then your vaunted precision starts to look like recklessness with better software.
That should land hard across the force.
For the intelligence community, this means target databases are no longer back-office housekeeping. They are lethal infrastructure. If a designation persists for years after the ground truth changes, that is not a clerical oversight. That is a future strike package waiting to happen.
For operators and commanders, it means pattern-of-life work cannot become optional just because the interface is cleaner and the target folder populates faster.
For policymakers, it means “human in the loop” is an empty phrase unless the human has both the time and the institutional permission to say no.
The Sobering Lesson
The AI debate has been too fixated on whether machines will one day replace humans in lethal decisions. The more immediate danger is cruder and more plausible.
Humans stay nominally in charge while slowly surrendering the habits that make command judgment real in the first place.
That is the sobering part.
The machine did not erase responsibility. It spread it out, sped it up, and made it easier to hide inside process.
Minab was not just a strike gone wrong. It was a preview of what happens when a military confuses kill-chain compression with good tradecraft. The result was not futuristic. It was grimly ordinary: stale intelligence, shallow review, false confidence, dead children.
That is not the future of war because AI made it so.
That is the future of war if professionals let speed outrun judgment.