Pentagon Investigators Say Overreliance on Palantir AI Contributed to US Strike That Killed 123 Iranian Children (gizmodo.com)
- Reference: 0185737220
- News link: https://news.slashdot.org/story/26/09/22/0614238/pentagon-investigators-say-overreliance-on-palantir-ai-contributed-to-us-strike-that-killed-123-iranian-children
- Source link: https://gizmodo.com/pentagon-investigators-say-overreliance-on-palantir-ai-tech-contributed-to-u-s-strike-that-killed-123-iranian-children-2000814477
> According to the officials who spoke to Bloomberg, some personnel inside U.S. Central Command leaned too hard on the AI inside the Maven Smart System, an AI-powered data integration and targeting platform developed by Palantir to accelerate intelligence analysis and decision-making. The Defense Department has reportedly made the software tool a cornerstone of military operations over the past year... Officials said some users expected Maven to flag stale records or contradictions in the intelligence assembled for potential targets, though it is not clear why they thought the system would do that. The Minab [elementary school] site, which was cataloged as an Islamic Revolutionary Guard Corps facility due to outdated data, was fed into Maven with other candidates and came out as a recommended day-one target. Target-list work that once took hours was condensed into minutes.
>
> A Palantir spokesperson told Bloomberg the company "is not responsible for the underlying data nor identifying intelligence deficiencies" and that there is no evidence its software was at fault. Two people familiar with Palantir's Pentagon contracts said the government keeps primary responsibility for the quality of the information that's fed into Maven. After the strike, Palantir added features that "re-review underlying intelligence to identify factors that would disqualify a target and flag inconsistencies and inaccuracies that human review may have missed," a person familiar with the work said. That new functionality is said to have already caught some anomalies...
>
> More than 1,000 Iranian targets were hit in the first 24 hours, and according to Bloomberg's sources, that pace compressed the time available for additional confirmations. Staffing on civilian harm mitigation teams across the U.S. Department of Defense had also fallen by roughly 90% over recent years, shrinking to fewer than 20 people overall. The U.S. Central Command's group shrank from 10 people to one. No member of those teams reviewed the Minab site before the missiles were up in the air.
>
> A [4]separate inquiry by the United Nations' Independent International Fact-Finding Mission on Iran this week reached the conclusion that there were reasonable grounds to believe the United States committed the war crime of launching an indiscriminate attack.
[1] https://gizmodo.com/pentagon-investigators-say-overreliance-on-palantir-ai-tech-contributed-to-u-s-strike-that-killed-123-iranian-children-2000814477
[2] https://www.bloomberg.com/graphics/2026-iran-school-attack/
[3] https://www.latimes.com/world-nation/story/2026-09-21/inside-u-s-kill-chain-that-destroyed-iranian-school
[4] https://www.ohchr.org/sites/default/files/documents/hrbodies/hrcouncil/sessions-regular/session63/a-hrc-63-61-aev.pdf
Something something (Score:2)
Not sending their best.
If only (Score:3)
It wasn't a completely foreseeable consequence of not double checking things. Heads should roll but they won't.
Re:If only (Score:5, Funny)
How could the Haitians eating Hunter Biden's laptop allow this to happen?
Re: (Score:2)
Elections have consequences, make sure to vote in November:
[1] Raskin has already launched investigations into allegations of corrupt government contract deals at Donald Trump Jr.’s venture capital firm and on the Trump administration’s pardon policy. [cnn.com]
Rep. Robert Garcia, the top Democrat on the House Oversight Committee, wants to interview the president and first lady as well as top officials like Vice President JD Vance and White House Chief of Staff Susie Wiles about how the administration has ha
[1] https://www.cnn.com/2026/09/13/politics/democrats-trump-investigations-midterms
Re: (Score:2)
People will really be like "you believe in voting? that pales in effectiveness to my strategy, firebombing a Walmart" and then not firebomb a Walmart
Re:If only (Score:5, Informative)
It was converted ELEVEN FUCKING YEARS AGO.
The changes were visible from space:
"Analysis of satellite imagery shows that the site had been converted into a school by October 2015, when new entrances and perimeter walls were constructed. By 2017, football pitches and assembly areas were visible in the playgrounds, and by 2018, walls around the school compound were painted blue and pink, distinguishing the girlsâ(TM) and boysâ(TM) yards."
[1]https://forensic-architecture.... [forensic-a...ecture.org]
[1] https://forensic-architecture.org/investigation/airstrikes-on-minab
Re: If only (Score:3)
Kind of makes you wonder if he lied about it for a reason
Re: (Score:1)
> It was converted ELEVEN FUCKING YEARS AGO.
I think it is also important to note that this was known very quickly after the attack, but I saw dozens of right wing trolls here, on TV and in the White House saying Iran had used the children as human shields. No one is taking responsibility for any of this.
Re: (Score:3)
Stop spreading bullshit. It's been a school for over a decade now. [1]https://en.wikipedia.org/wiki/... [wikipedia.org]
Investigations confirmed that the school had existed as a civilian institution for more than a decade, located close to, but separate from the IRGCN compound.[21] According to satellite imagery, the building housing the school was walled within the IRGCN compound in 2013 and walled off from the rest of the compound by October 2016 at the latest.[11][20][22] Its new external wall had three entrances separate f
[1] https://en.wikipedia.org/wiki/2026_Minab_school_attack
Re:If only (Score:4, Interesting)
Heads *need* to roll though, but I'm not sure anyone with a say in the matter at the Pentagon realises that, let alone why is really does need to be the case. The US' own judiciary has been pretty clear on this; whenever a lawyer has been caught using AI-generated submissions that contain hallucinations some form of censure has ensured PDQ. It's similar in industry as well; multiple instances of employees using AI without checking the facts that have subsequently been either disciplined, fired, or (if they are senior enough and the screw-up warrants it) issued a golden parachute.
Now, look at the situation on the ground. This school has been there for around a decade, and it is *very* obviously a school - pink and blue painted play areas, with assembly areas and sports facilities suitable for kids; that is totally NOT a military PT facility within the larger complex. Sure, it's surrounded by IRGC facilities, but absolutely no military anywhere else in the world co-locates their military's family accommodation blocks and support facilities close to the operations blocks and facilities to make it easier for their staff & troops, right? Oh, wait, yes, they do, don't they? Not always, sure, but often enough that the US response here matters, because if they try and sweep this under the rug "because AI", then they just made all their similar facilities viable targets "because AI". There's a reason everyone else, including US allies, were quick to publically condemn this, and even say it might be a war crime because of failure to properly assess the strike's potential for collateral damage, and that's to make it absolutely clear they do not consider this have acceptable levels of collateral damage, let alone be a legitimate target.
Don't forget, the IRGC tend to be fanatics even by Iranian standards, and are often either willing to strap on a suicide vest or can motivate others to do it for them. 123 dead kids means a *LOT* of bereaved parents, brothers, sisters, and other family members, many of whom are probably already linked to the IRGC, who are going to need a lot less motivation to do something a little more radical than just march along the street waving flags and chanting "Death to America!" the next time they are asked to do something for their country. What the US does here is totally going to be interpreted as setting precedent by Iran, and quite possibly by other countries and groups within their axis, should it come to a conflict or just an opportunistic attempt at some terrorism.
Re: Gee (Score:1)
And military and intelligence analysts have warned of the terrible consequences of attacking Iran DIRECTLY for years. Thats why no other president dared to do so. Up to these days.
Propaganda, meant to happen, or other? (Score:2)
> "a cascade of preventable failures that included an overreliance on an AI tool built by Palantir."
On the one hand, this administration is desperate because the mid-terms are coming up and they're so far underwater in the polls that they're pushing up seaweed. They'll say anything and everything to gain even a fraction of an approval point, and are grasping at every straw that might allow them to keep breathing long enough to successfully rig the election. So throwing Palantir under the bus would be psychopath logic.
On the other hand, Thiel and Karp are themselves creepy psychopathic fuckers who support
Re: (Score:2)
The simplest explanation is usually the correct one, and in this case, given what I've seen people do with AI, and given my own experience with Claude Code (Opus and Fable models), the normal human inclination to trust it's probably right and click OK is by far the most likely explanation. And remember that frontier models have grown increasingly better over the last year, and one of the newest improvements is that the reasoning models use two models in an adversarial mode where the second model checks and
Re: (Score:2)
What applicable US law would any of those responsible face? Misfeasance is the only one that comes to mind, and I think that's a state level law.
If you mean the laws SHOULD criminalize and punish this kind of behavior, I agree. They don't for high military or executive branch officials. Not even after they leave office.
Re: (Score:2)
Throw them in the Hague maybe? Just a thought.
Wrong focus (Score:5, Informative)
The real reason is cost/staff cutting and not reviewing the strike plans;
"Hegseth dismantled most of the Pentagon’s civilian harm mitigation, or CHM, units — cutting headcount across a number of teams by roughly 90% to fewer than 20 staff members, people with direct knowledge of the matter said. Centcom’s team was reduced from 10 to 1.
No CHM team member reviewed the Minab site before the strike, according to officials involved in the internal investigation. That was a consequence of the cuts in staffing but also a decision not to involve the group in planning and operations..."
Is it though? (Score:4, Insightful)
The whole selling point of AI is that you'll need less staff which means less human oversight. Well, fucking oops.
Re:Wrong focus (Score:4, Informative)
To top it all off that motherfucker Mike Johnson ended the session early to prevent an impeachment vote for Hegseth. [1]https://www.pbs.org/newshour/p... [pbs.org]
Why do republicans not want to work anymore?
[1] https://www.pbs.org/newshour/politics/speaker-johnson-recesses-house-early-blocking-massies-push-to-impeach-hegseth
Re: Wrong focus (Score:3)
Because Republicans believe the average American to be an uninformed lout. That none of this will ever come back to their party. That they can simply pivot away, stir up a new crisis, and we will all forget.
Re: Wrong focus (Score:3)
Which they know because they kept the receipts from when they fucked up education to make them stupid
Re: (Score:2)
This.
Minimizing civilian casualties is of course always important if you want to even pretend to be a civilized country. But in this case it was, and I'm using the past tense there on purpose, even more important. Before the US started this war, Iran had an aging, sickly, increasingly-unpopular leader who probably would have been dead in a few years from natural causes. But as long as he remained alive and in power, he was a focal point for Iranian dissidents, and if the goal really was to encourage t
Re: (Score:2)
On the upside, through destructive testing we now have a datapoint on the effectiveness of civilian harm mitigation.
Re: (Score:2)
Sounds like plain criminally negligent homicide on mass-scale to me.
The Legitimate Concerns of AI in Warfare Came True (Score:2)
This is an example of one of the legitimate challenges facing the increased integration of AI into warfare. The AI does its work well enough that the humans in the loop who are supposed to be supervising it are lulled into a false confidence of the AI's ability. Thus, when the humans are supervising numerous decisions in a high pressure, time critical schedule, they begin "abdicating" decision authority to the AI by just assuming it's right, resulting in lethal mistakes. You have the veneer of humans in the
Distributed responsibility and transparency (Score:5, Interesting)
> This is an example of one of the legitimate challenges facing the increased integration of AI into warfare.
The more fundamental problem is distributed responsibility and lack of transparency.
Until recently, drone strikes required a long sequence of command decisions, starting with intel gathering, analysis, decision, orders, and remote command of the device. No one is directly responsible when something goes wrong, the failure is distributed among many players with the result that no one has to bear the consequences.
For example, the [1]"collateral murder" [wikipedia.org] incident, where apache helicopters gunned down civilians and journalists.
In normal situations, we have [2]Derek Chauvin [wikipedia.org], who was an individual who caused the death of one man. He went to trial, the evidence was examined, and guilt was assigned. Trials are more-or-less held in the public light, and journalists and interested parties can examine court records and in some cases reanalyze evidence (not all physical evidence is kept or available, but for example, DNA evidence can sometimes be reexamined).
That's how it's supposed to happen.
Contrast that with erroneously ordered drone strikes that hit civilians instead of legitimate targets.
I suppose the biggest, most egregious example would be George Bush taking the country to war because Saddam Hussein had "weapons of mass destruction", and of course one were found, and the security services swore up and down that they told him there were none, but he stated "well, the message didn't get through".
And skated responsibility. Completely.
To bring this closer to home, software engineers implement all sorts of insecure, inconvenient, and privacy violating features with no moral pushback. Even if it's completely obvious that something is an anti-feature, people do it because a) their job depends on it, and b) it's someone else's decision. And if there's a breech and ten million social security numbers get made public, the company simply fixes the problem and continues. No responsibility is assigned.
I was satisfied to see videos of the "narco terrorist" boats get destroyed, and then the survivors killed. Not the actual killing, but just knowing that the event was recorded and analyzed in the public sphere, and there was a conversation about it, and it was investigated. That seems like it was handled the right way.
Death is permanent, it can't be undone. If we're going to kill people, regardless of whether it's AI or anything else, it should be transparent and put into the public discourse.
No one is assigned blame for screwing up, that's the more fundamental problem.
[1] https://en.wikipedia.org/wiki/July_12,_2007,_Baghdad_airstrike
[2] https://en.wikipedia.org/wiki/Derek_Chauvin
AI & PMCs s make handy scapegoats (Score:2)
As far as regimes opposed to the US are concerned, maybe they'll start poisoning AI and open source data with deliberately false and misleading information because the intelligence are too stupid to do some due diligence. Maybe that's what happened here, or the AI just hallucinated something. I doubt it would be hard to drop some poisoned information in the wild and have missiles hit targets with nothing of strategic value in them.
needful AI (Score:1)
A many previous posters have described, forced dependence on
i see 2 possibilities ... (Score:2)
- unchecked ai mistake
- deliberate targeting now brushed under the rug of unchecked mistake
those girls killed were mostly daughters of irgc officials. it seems reckless and counterproductive, which it is, but if you want to deal a devastating blow to crush any opposition and believe that it will work, it becomes a plausible decision. it's right there in line with killing a head of state who is at the same time the revered leader of an entire major religion, along with a good part of his family and entourage
Re: (Score:2)
So those in charge are murderous and treasonous war crime monsters who undertook unconstitutional actions as Congress never even green lit this? Sounds like the Epstein files are likely accurate and the real death toll isn’t those foreign children, but your own and your neighbors. Time for prison or better, the permanent solution.
Re: i see 2 possibilities ... (Score:2)
How much evidence of this do we need? Blowing up those boats and then killing some of their crews should have proved it. We are the international criminals. We are the bad guys.
Re: (Score:2)
Well, possibly. If so, this should get everybody involved life sentences. But we do not even seem to be able to prosecute child rape against the felon-in-chief.
Also need to add that this failed spectacularly. The Irak war cannot be won by the US. All the US can try is to bribe them now.
It was not me, it was the AI! (Score:2)
Here we are folks, at that crossroad where people are now going to blame every mistake on AI like it's a get out of jail free card. We either hold the AI companies responsible, the person using it without checks and balances or both. There are no other options and it needs to start now, lots of folks are going to get hurt by not starting this now.
Re: (Score:2)
And it didn't work at Nuremberg either.
This is how (Score:3)
When experts say AI will destroy humanity, people think of robot “kill all humans” armies and super intelligence taking over. How it already almost happened is slop brained “secretary of war” kegsbreath [1]demanded [apnews.com] we use AI in the military despite warnings. Then everything from AI slop reports leading to bombing over a hundred children in a school to slop reports of nuclear weapons materials being shipped by China nearly started WWIII. It’s not going to end the world by solving the physics of the universe causing a singularity that ends existence, it’s forcing brain rot onto those in charge until they blindly do what the stupid word spewing box commands that will end it.
[1] https://apnews.com/article/anthropic-hegseth-ai-pentagon-military-3d86c9296fe953ec0591fcde6a613aba
Re: (Score:2)
It seems like the issue of AI hallucination is widely promulgated and widely known. I mean, this site has an overrepresentation of technicians who tend to think of highly esoteric tech information as common knowledge just because they work with it every day.....but I don't think this quirk of AI behavior is in that category.
And yet, people keep making this same mistake precisely where it matters. We get stories like this every other day. I have seen it at my own job, where my own boss who still has some
Re: (Score:2)
> People who should know better let their wish for a reliable AI override their knowledge that AI is not reliable, and they just move forward with it.
Non technical people think even less, and are obviously less knowledgeable yet they too believe the squawk box without question. It’s so much easier to not have to think, as if it’s trying to flex an atrophied limb that now just causes pain on any attempt.
Incompentence (Score:2)
Taking AI without some sort of verification and guardrails is pure unadulterated incompetence.
Re: (Score:2)
Guardrails would invalidate the [1]Nuremberg defense [wikipedia.org]
.
[1] https://en.wikipedia.org/wiki/Superior_orders
Re: (Score:2)
Everyone who can remove "guardrails" does eventually remove them.
Local police stations removed them to search flock cameras for tits.
Most certainly the military removed them to search for targets.
I also have an "uncensored" model or two in case I need information on topics that someone finds "sensitive".
You surely don't believe that the dario andropic, scam slopman, elona or cuckerberg develop their plans for world domination with censored models.
So Palantir *and* the DoD (Score:2)
are both guilty of war crimes. How 'bout a trillion dollars in blood money to the parents?
Re: (Score:2)
> How 'bout a trillion dollars in blood money
Why not? After all, it won't be coming out of their pockets.
You don't need AI to screw up targeting (Score:2)
We've been doing [1]that for years [wikipedia.org]. AI allow you to fuck up much faster and more efficiently however.
[1] https://en.wikipedia.org/wiki/United_States_bombing_of_the_Chinese_embassy_in_Belgrade
Murder by LLM ... (Score:2)
Should still get them life-sentences.
Who is accountable? (Score:2)
Nobody? Honest mistake? I guess nobody has to take responsibility for anything anymore. THANKS AI!
Old data or incorrect data? (Score:2)
Everyone knows that evil regimes don't follow any sort of rules of engagement or military codes of conduct like the West does. They will deliberately put their military forces in or near schools and hospitals. Human shields tend to work in only one direction. If those sites don't get taken out, they have a base of operations. If they do get taken out, they can go to the court of public opinion and try to deflect blame.
So we all know they are lying (Score:5, Insightful)
We know it was just basic incompetence and a complete disregard for human life. This will of course provide an excuse and change the conversation though.
It also fits in with the current narrative the big AI companies are pushing because they would really like a very specific set of government regulations that coincidentally are good for the existing players but would prevent any new players from entering the market now that they have become established and that they are close to doing their IPOs...
Whatever the case at least we got them to admit that they murdered 123 children.
The war can't end until after the midterms because Trump lost because of course he lost everything Trump touches dies. This means we're going to have to pay Iran reparations to the tune of $300 billion US taxpayer dollars. That's $300 billion of your dollars going to Iran.
Trump can't do something that weak and pathetic right before the midterm elections. He'll do it right after.
That'll give the 77 million people who thought it was a good idea to put the king of debt through bankrupts every business he runs in charge of the country he promised to run like a business.... I honestly do not know how our species is going to survive the next 25 years.
Re: So we all know they are lying (Score:1)
Mother earth is a formidable killer. If humanity dont behave, she erases humans and come up with some new species, in a few thousands/millions years.
AI never lies, it just has bad human data? (Score:3)
On: "A Palantir spokesperson told Bloomberg the company "is not responsible for the underlying data nor identifying intelligence deficiencies" and that there is no evidence its software was at fault."
Is AI now an aspect of Mother Nature? Reminds me of:
[1]https://2001.fandom.com/wiki/F... [fandom.com]
""Let me put it this way, Mr. Amer. The [Palantir] 9000 series is the most reliable computer ever made. No 9000 computer has ever made a mistake, or distorted information. We are all, by any practical definition of the
[1] https://2001.fandom.com/wiki/Featured_Quotes
Re: So we all know they are lying (Score:2)
I personally bet on Horizon Zero Dawn scenario and this story only convinces me I'm making a good bet.
Re: (Score:2)
Not sure why this was modded flamebait. I don't see anything grossly inaccurate and any editorializing isn't exactly couched in fantasy.
Re: (Score:2)
> Not sure why this was modded flamebait.
Err, habit. Its that guy.
But ok, first he conflates one of his pet hates, "big AI" , with Palantir.
The substance is not horribly wrong, but the way he says it might be accurately described as "flamebait". Certainly inflammatory. And there is no "he's right, but does not have to be such a douche about it" moderation. The juvenile name-calling is as bad as Trump.