Two Tomahawk missiles hit the Shajarah Tayyebeh Elementary School in Minab, Iran, on February 28, killing over 150 people, including at least 123 children. Within hours, some military personnel knew the United States was responsible. Months later, an unreleased internal Pentagon investigation leaked to Bloomberg pointed fingers at a familiar culprit: an overreliance on artificial intelligence. Specifically, officials blamed Palantir’s Maven Smart System for failing to flag stale intelligence.
It is easy to point the finger at software. Machines don't talk back, and corporate statements offer convenient targets. But blaming a software tool for a catastrophic failure of human oversight ignores how modern military commands actually operate under pressure.
The Anatomy of a Flawed Kill Chain
The tragedy in Minab didn't happen in a vacuum. On the opening day of the conflict, the administration ordered an overwhelming air assault, targeting more than 1,000 locations within the first 24 hours. That compressed timeline forced a massive crunch on intelligence analysts.
Compounding the rush was a structural personnel drain. The military had previously cut back on civilian-protection staffing, leaving fewer trained eyes to manually verify targets. Analysts relied heavily on the Maven Smart System—a platform designed by Palantir to aggregate data and speed up decision-making.
U.S. databases still listed the Minab compound as a military installation based on old records. However, commercial satellite imagery clearly showed that walls and separate entrances had been built by 2017, cutting the school off from the adjacent base and turning it into a functioning civilian institution. The data was outdated, the imagery was ignored or unverified, and the human checkpoint failed.
Shifting Blame to the Algorithm
When things go wrong in high-stakes environments, accountability often trickles downward or sideways into software logs.
Palantir defended its platform quickly. A company spokesperson noted that the software "is not responsible for the underlying data nor identifying intelligence deficiencies," adding that there is no proof the system malfunctioned. True to form, the tech firm responded to the incident by building new features designed to re-review underlying intelligence, flagging inconsistencies that human operators missed.
Yet, software tools are only as good as the directives and habits of the people using them. If commanders prioritize speed over verification and slash the personnel responsible for civilian safety, algorithms will simply accelerate bad decisions. Expecting an AI platform to independently correct systemic institutional failures is a convenient excuse for leadership.
The Broader Fallout and United Nations Findings
The internal Pentagon probe isn't the only body looking at the disaster. A separate investigation by the United Nations' Independent International Fact-Finding Mission on Iran concluded that there were reasonable grounds to believe the United States committed war crimes by launching an indiscriminate attack.
The UN report stated that the U.S. failed in its obligation to verify the target, acting recklessly while knowing there was a substantial risk of striking a civilian object. Despite these findings, Washington has avoided publicly accepting direct responsibility.
What Comes Next for Military AI Integration
As autonomous systems and data integration tools become permanent fixtures in defense operations, the friction between speed and safety will only grow. The temptation to let algorithms handle heavy workloads during massive air campaigns is immense. But when organizations treat software as a substitute for human judgment rather than an assistant, disasters happen.
If military leadership wants to prevent future tragedies, the fix isn't just updating software features or blaming a contractor. It requires restoring the human checkpoints, slowing down the kill chain when data is ambiguous, and holding commanders accountable for the intelligence they choose to act on.
