An unreleased Pentagon review revealed that outdated intelligence, thinned civilian-protection staffing, and extensive reliance on AI targeting systems all contributed to the US strike that killed more than 150 Iranians, including at least 123 children, at an elementary school in Minab, Iran, through the Palantir AI kill chain, according to Bloomberg.
On February 28, two US Tomahawk missiles struck the Shajarah Tayyebeh Elementary School in Minab, Iran, on the opening day of the US-Iran-Israel war.
On September 18, Bloomberg published an investigation, reporting that some Pentagon personnel knew within mere hours that the US was responsible, exposing stale intelligence, rushed approvals, reduced civilian protection staffing, as well as excessive trust in AI automation.
The Department of War (DoW), formerly Defense Department, opened a formal command investigation shortly after the strike, led by a general officer from outside US Central Command that carried out the US attack on the Iranian children’s school.
AI can accelerate decisions, but speed does not guarantee that the intelligence behind them is accurate.
Palantir AI Kill Chain Deadly Contribution
Leader of the Palantir AI kill chain is its AI-powered targeting platform, Maven Smart System, built by the big data analytics company, under a Pentagon contract worth $1.3 billion.
Palantir’s Maven is the core military backbone of the Pentagon, designed to fuse satellite images, drone footage, and signals intelligence into a wave of quick strike recommendations.
The Palantir AI kill chain operated as more than 1,000 Iranian targets were struck in the first 24 hours of the assault.
According to officials who spoke to Bloomberg, some CENTCOM personnel expected Maven to automatically flag stale, or contradictory intelligence, though it is unclear why they believed the system was built to do that.
The site had once belonged to an Islamic Revolutionary Guard Corps (IRGC) compound. However, commercial imagery showed walls separating the school from the military area by 2017, while 2018 images showed brightly painted walls, a soccer pitch and play markings.
For the Palantir AI database workflow, the problem was that the main military intelligence database still treated the location as a current IRGC facility, not a children’s school. It has been cataloged in US database as the military facility for years, in which the outdated classification was fed into Palantir AI kill chain, Maven, and the school was flagged as a target on day one.
According to a Bloomberg post in June, a US analyst noticed changes at the site as early as 2019 but logged them in a system that was not connected to the main targeting database. The AI kill chain could therefore move quickly without receiving intelligence that might have challenged the old classification.
Maven combines more than 150 data inputs to support targeting and command decisions. According to Bloomberg, Palantir AI model Maven helped reduce work that had previously taken military staff hours to only several minutes during preparations for the Iran operation.
That speed reflects broader Pentagon Palantir AI adoption. In its Q1 2026 business update, Palantir described Maven as moving AI to US warfighters “at speed and scale,” showing how faster data processing has become part of the Pentagon’s military AI strategy.
But investigators found that some Central Command personnel expected Maven to flag outdated intelligence or inconsistencies. The Palantir AI kill chain accelerated information, while operators may have expected it to perform checks beyond what the system was designed or contractually required to do.
Palantir rejected responsibility for intelligence failure. The company said it is not responsible for the underlying data nor identifying intelligence deficiencies, adding that there was no evidence its software caused the Minab strike.
That distinction is important when considering the role of Palantir AI products inside military decision making.
Can Automation Replace Safeguard?
According to Bloomberg, Pentagon civilian harm mitigation teams had been reduced by roughly 90% to fewer than 20 people, while CENTCOM’s team fell from 10 members to one.
No member reviewed Minab before the strike, reducing human checks around Pentagon Palantir operations.
Around three dozen people still participated across the targeting process. Yet Pentagon investigators found high confidence throughout the Palantir AI kill chain that Minab remained a military facility, even though more recent information showed the location had changed.
The Palantir AI database issue therefore went beyond one wrong entry. Automation cannot reconcile intelligence it cannot access; meaning fragmented military systems can allow an outdated assessment to survive even when newer evidence exists elsewhere.
Two people familiar with Palantir AI military contracts told Bloomberg that the US government retains primary responsibility for the quality of information fed into Maven. After Minab, Palantir added new capabilities that can re-review underlying intelligence to identify factors that would disqualify a target and flag inconsistencies and inaccuracies that human review may have missed, according to a person familiar with the changes.
The update shows how Palantir AI models are being adjusted following the strike.
Still, AI kill chain automation depends on current imagery, connected databases and human review.
As Palantir AI military contracts expand, Minab raises a broader strategic question: should military AI progress be measured by how quickly decisions are produced, or by whether systems can recognize when the intelligence behind those decisions is uncertain?
That tension sits at the center of AI war technology. Speed can help commanders manage large operations, but it can also move an outdated classification through several stages before someone has enough time or information to challenge it.
Following Bloomberg’s report, the Pentagon is already working to replace weaknesses in its main targeting database with systems using more automation. Minab shows why Pentagon Palantir integration still depends on updated target libraries, connected intelligence and strong human safeguards.
The UN backed a fact-finding mission and added another level of scrutiny. According to the Associated Press, it found reasonable grounds to believe the US committed war crimes in two strikes that killed civilians.
However, the US government rejected the mission’s findings.
For the US military, the Palantir AI kill chain triggers the deeper tension behind its AI strategy. Automation can deliver speed and scale, but if flawed intelligence moves through the system faster, faster decision making alone cannot be treated as strategic progress.
Inside Telecom provides you with an extensive list of content covering all aspects of the tech industry. Keep an eye on our Intelligent Tech sections to stay informed and up-to-date with our daily articles.
